Datasets:

WilliamWen

/

summarization_cata

like 0

Data will be made available on request.Fossil fuels are essential building blocks in the petrochemical industries for producing materials such as plastics, synthetic fibres, rubbers, lubricants, detergents, and solvents (Keim, 2010; Speight, 2011). However, their non-renewable nature poses a sustainability risk, prompting the search for sustainable alternatives based on renewable biomass sources (Ozturk et al., 2017). Oil produced from pyrolysis of lignocellulosic biomass, such as oil palm biomass, can be a sustainable alternative to fossil fuels, given its carbon-neutral properties with low sulfur and nitrogen content (Martínez et al., 2014; Palamanit et al., 2019). During pyrolysis, the thermal decomposition of oil palm biomass may produce more than 300 chemical compounds in the oil, which can be used as precursors for synthesizing petrochemical products (Keim, 2010; Machado et al., 2022). Oil palm biomass-derived pyrolysis oil mainly consists of oxygenated compounds due to its high oxygen content in raw biomass. Such oil requires modification to improve the hydrocarbon content (Palamanit et al., 2019). Co-pyrolysis of oil palm biomass with plastics like polypropylene (PP) is a promising method for increasing hydrocarbon content (Al-Maari et al., 2021). PP, rich in carbon and hydrogen, provides the hydrocarbon pool required for the deoxygenation reaction of oxygenated compounds from biomass to form hydrocarbons such as aliphatic and aromatic hydrocarbons in oil.Solid acidic catalysts can further promote the deoxygenation reactions (i.e., dehydration, decarbonylation, and decarboxylation) of pyrolytic volatiles to improve the hydrocarbon content in pyrolysis oil (Hassan et al., 2019; Shafaghat et al., 2019; Zhang et al., 2016). Due to its excellent catalytic performance for deoxygenation, which generates hydrocarbons such as olefins, aliphatic, and aromatic hydrocarbons, high-acidity zeolites have been widely used in several studies (Balasundram et al., 2018; Wang et al., 2020). In such a reaction, oxygen is typically removed by releasing by-products such as water, carbon dioxide, and carbon monoxide (Hassan et al., 2019). Zeolites, nonetheless, are microporous catalysts. Thus, micropore-related flow restriction can affect their deoxygenation catalytic performance, especially if relatively large molecules such as lignin-derived compounds are involved (Shafaghat et al., 2019). Such flow restriction can also cause coke formation, create pore blockage, catalyst deactivation, and catalyst poisoning, thereby reducing the performance of catalysts (Hassan et al., 2019; Shafaghat et al., 2019). To address this, mesoporous acidic catalysts such as titania (TiO2) and alumina (Al2O3) based catalysts with larger pore sizes were introduced, allowing large molecules to diffuse and reducing pore blockage and catalyst deactivation (Lu et al., 2010; Zhou et al., 2019). The high chemical and thermal stabilities of TiO2 and/or Al2O3-based catalysts have also sparked interest (Bagheri et al., 2014; Paranjpe, 2017). It has been proposed that doping of metals such as nickel, copper, molybdenum, cobalt, palladium, and cerium into TiO2 and/or Al2O3-based catalysts can improve deoxygenation (Bagheri et al., 2014; Lu et al., 2010; Zhou et al., 2019).Several works have evaluated TiO2 and/or Al2O3-based catalysts in oil upgrading through catalytic pyrolysis and co-pyrolysis in a nitrogen atmosphere. Dong et al. (2019) compared the catalytic performances of titania-based catalysts doped with different metals, including copper (10% Cu/TiO2), iron (10% Fe/TiO2), and molybdenum (10% Mo/TiO2) on the phenol conversion during the catalytic pyrolysis of corn straw lignin at 450 °C. They reported that the highest total phenol conversion was attained using 10% Mo/TiO2, followed by 10% Cu/TiO2, TiO2, and then 10% Fe/TiO2. Lu et al. (2010) studied the catalytic upgrading of oil from pyrolysis of poplar wood using the titania, zirconia, and titania-zirconia-based catalysts doped with cerium, ruthenium, and palladium at 600 °C. In general, all the catalysts reduced the sugars (i.e., levoglucosan) in the oil, while titania-zirconia-based catalysts yielded a high amount of hydrocarbons and ketones. TiO2-based catalysts promoted the formation of phenols. Mysore Prabhakara et al. (2021) investigated the catalytic performance of γ-Al2O3, dolomite, and hydrotalcite (HTC) MG70 with the addition of 20 wt.% Na2CO3 into the catalysts during the catalytic pyrolysis of beechwood at 500 °C. All these catalysts significantly reduced the oxygenated compounds and enhanced the formation of aliphatic, monoaromatic, and polyaromatic hydrocarbons. Zhou et al. (2019) investigated the utilization of NiO/γ-Al2O3 catalyst on the dehydration reaction mechanism during the pyrolysis of rice husks. Weak acid sites on Al2O3 were discovered to facilitate the dehydration reaction the most throughout the process. In addition, Imran et al. (2014) reported that the alumina-supported sodium carbonate (Na2CO3/γ-Al2O3) catalyst improved the quality of oil from the pyrolysis of wood fibers.No studies have used titania and alumina-based catalysts in the co-pyrolysis of OPT and PP to improve the targeted oil composition. The mesopores in these catalysts may facilitate the diffusion rate of large molecules (i.e., compounds derived from the thermal degradation of OPT and PP) through the pores of the catalysts and promote the conversion into hydrocarbons during catalytic co-pyrolysis. This study investigated the catalytic performance of a titania-based catalyst doped with nickel-molybdenum (Ni–Mo/TiO2) and an alumina-based catalyst with nickel (Ni/Al2O3) for the upgrade of oil generated from co-pyrolysis of OPT and PP. The effect of the catalysts on the oil composition was evaluated.OPT was collected from an oil palm plantation in Saratok, Sarawak. OPT was pre-dried in the oven at 105 °C for 24 h, ground (Fritsch rotary mill, PULVERISETTE 14), and sieved (Fritsch sieve shaker, ANALYSETTE 3 PRO) to obtain the samples with a particle size of 500 μm and below. Locally sourced PP food containers were cut into smaller sizes and sieved using a sieve shaker (Fritsch, ANALYSETTE 3 PRO) to obtain samples with a particle size of 500 μm and below. The sieved PP was stored under ambient conditions before use. Two catalysts used in this study, Ni–Mo/TiO2 and Ni/Al2O3, were synthesized based on the impregnation method reported by Aqsha et al. (2015).The catalysts’ specific surface area, average pore diameter, and pore volume were determined via nitrogen adsorption-desorption isotherm analysis (Brunauer–Emmett–Teller (BET) surface area and pore size analyzer, Quantachrome Nova 4200e). Before the analysis, the samples were degassed at 200 °C for 12 h to remove any surface-adsorbed residual moisture.The crystallinity of the catalysts was investigated using powder X-ray diffraction (XRD) (X-ray Diffractometer, Rigaku SmartLab). Cu-Kα radiation (λ = 0.154 nm) was used to measure the diffraction patterns in the range of 2θ from 5 to 100°.XRF was used to analyze the composition of the catalysts with an accelerating voltage of 15 kV and a current of 30 μA (Bruker S2 PUMA).The acidity of the catalysts was determined through NH3-temperature programmed desorption (TPD) analysis (Micromeritics Chemisorb 2750). The sample was pre-treated by heating it from room temperature to 200 °C in helium gas flow for 120 min. Adsorption of NH3 was carried out at 100 °C for 60 min (5% in He, v/v), followed by helium purging at the same temperature for another 60 min. Following that, NH3 desorption was carried out by heating from 50 to 800 °C at a ramping rate of 10 °C min-1 and holding at the final temperature of 800 °C for 15 min.The catalytic co-pyrolysis was carried out in a horizontal tube furnace (MTI, GSL-1100X) with a 400 mL min-1 nitrogen flow rate to form an inert condition in the tube furnace. 3 g of OPT and PP mixture sample (weight ratio of OPT: PP of 1:1) with 0.3 g of catalyst were loaded into the reactor and nitrogen purged for 5 min. The reactor was heated to the desired operating temperature (i.e., 500, 600, and 700 °C) at a heating rate of 10 °C min-1, with a holding time of 40 min. Afterwards, the reactor was cooled down to 200 °C while continuously purged with nitrogen gas. A cold trap in an ice bath (2–3 °C) was connected to the tube reactor outlet to collect the liquid product (oil) from the experiment. The collected oil was stored at 2–7 °C until further analysis. The non-condensable gases were released into the environment. The product yield obtained from the experiments was calculated using Equations (1)–(3) . (1) P y r o l y s i s o i l y i e l d ( w t . % ) = M a s s o f p y r o l y s i s o i l o b t a i n e d ( g ) M a s s o f s a m p l e ( g ) x 100 % (2) S o l i d y i e l d * ( w t . % ) = M a s s o f s o l i d o b t a i n e d ( g ) M a s s o f s a m p l e ( g ) x 100 % *Solid yield refers to all solid residues collected from the experiments, including feedstock residue, catalysts, and coke. (3) G a s y i e l d ( w t . % ) = 100 w t . % – p y r o l y s i s o i l y i e l d ( w t . % ) – s o l i d y i e l d ( w t . % ) The composition of pyrolysis oil was determined using a gas chromatography-mass spectrometer (GC-MS) with an HP-5MS column (Agilent, 30 m length x 0.25 mm inner diameter x 0.25 m film thickness) (Agilent, 6890 N). The column oven was programmed to operate at 40 °C for 3 min. Afterwards, it was heated from 40 to 200 °C at the rate of 8 °C min-1 with a holding time of 10 min. The temperature was then ramped from 200 to 220 °C at a rate of 10 °C min-1 and held for 10 min. The column was kept at a pressure of 7.04 psi and a flow rate of 1 mL min-1 of helium. The split ratio of 50:1 was used in the analysis. Before the analysis, 0.2 g of pyrolysis oil was diluted in 10 mL of acetone. A syringe filter was used to filter the diluted oil sample before it was transferred to the GC sample vial and injected into the equipment via auto-injection mode for analysis. Compounds were identified by comparing the NIST08 mass spectral data library entries. Table 1 presents the textural properties (i.e., specific surface area, pore volume, and average pore diameter) of Ni–Mo/TiO2 and Ni/Al2O3. The lower specific surface area of Ni–Mo/TiO2 compared to Ni/Al2O3 could be attributed to the accumulation of two types of metal particles on the catalyst's surface or within its pores (Kumar et al., 2019). Both catalysts are categorized as mesoporous since their average pore diameter sizes are between 2 and 50 nm (Thommes et al., 2015). The large pores allow large molecules, such as lignin-derived compounds, to flow in and out of the catalysts' pores for higher conversion of the compounds during the catalytic co-pyrolysis (Lu et al., 2010). Fig. 1 depicts the acidities of the catalysts analyzed with NH3-TPD, which reveals the acid site distribution. The temperature region where the ammonia desorption peak has located indicates the types of acid sites (i.e., weak, medium, and strong acid sites) on the surface of both catalysts. The weak acid sites correspond to the ammonia desorption peak at temperatures less than 250 °C. In comparison, the medium acid sites appear in the temperature region between 250 and 500 °C. The ammonia desorption peak, which appears at temperatures above 500 °C, represents strong acid sites (Phan et al., 2020). Strong acid sites with higher acid strength likely provide higher catalytic cracking activity for converting the compounds into desirable products through the catalyst (Li et al., 2020). Fig. 1(a) shows that most ammonia desorption peaks are between 250 and 500 °C, indicating the presence of medium acid sites for Ni–Mo/TiO2. On the other hand, weak, medium, and strong acid sites are present on Ni/Al2O3 catalyst surface as the ammonia desorption peaks are detected in all three temperature regions (Fig. 1(b)). A higher peak intensity value in Ni–Mo/TiO2 relative to that in Ni/Al2O3 contributes to the higher acidity in the former catalyst (Table 1). Fig. 2 shows the powder XRD patterns of Ni–Mo/TiO2 (upper) and Ni/Al2O3 (bottom) catalysts, respectively. Numerous peaks appear on the pattern of Ni–Mo/TiO2, indicating the presence of a mix of phases. Indexing reveals three major oxide phases, i.e., anatase (TiO2), molybdenum oxide (Mo9O26), and nickel oxide (NiO2). An intense peak at 2θ of 25.3° is detected for TiO2 phase, along with weak peaks at 2θ of 37.8°, 48.0°, 53.9°, 55.1°, and 62.7° (COD#96-720-6076). For Mo9O26 phase, intense peaks are observed at 2θ of 24.9° and 25.3°, while weak peaks are present at 2θ of 27.3°, 32.2°, and 33.0° (ICSD#98-002-7510). NiO2 has a weak characteristic peak at 2θ of 37.1° (ICSD#98-007-8698). Ni/Al2O3 has two-phase components, i.e., nickel oxide (NiO) and alumina (Al2O3). The intense peaks of NiO are observed at 2θ of 37.2° and 43.3° while the weak peak is detected at 2θ of 62.93° (ICDD#03-065-6920). On the other hand, the characteristic peaks of Al2O3 are observed at 2θ of 46.0° and 66.8° (ICDD#00-004-0858). The catalyst's composition from XRF analyses is presented in Table S1 in Supplementary Information. Fig. 3 depicts the product yield obtained from non-catalytic and catalytic co-pyrolysis of OPT and PP with Ni–Mo/TiO2 and Ni/Al2O3 at temperatures ranging from 500 to 700 °C. The solid yield in Fig. 3 refers to all solid residues collected from the experiments, including feedstock residue, catalysts, and coke. When the temperature rises from 500 to 700 °C, the solid yield decreases for non-catalytic and catalytic conditions due to the decomposition of char present in the solid fraction into the oil and gas with the rising temperature. According to Zhou et al. (2013), char formation is more favorable at a lower temperature (450 °C) due to the lower decomposition rate of the feedstocks. At temperatures above 450 °C, the decomposition of feedstocks into condensable volatiles and non-condensable gases improves while char formation decreases. The pyrolysis oil yield in non-catalytic co-pyrolysis is maintained at 16 wt.% from 500 to 600 °C and drops to 11.50 wt.% when the temperature rises to 700 °C. The pyrolysis oil yield in catalytic co-pyrolysis with Ni–Mo/TiO2 increases from 12.67 to 19.50 wt.% with the rise in temperature from 500 to 600 °C. Further increase of temperature to 700 °C reduces the oil yield to 17 wt.% due to the enhancement of the secondary reactions of the primary volatiles into the gaseous products at higher temperatures (>600 °C) (Fan et al., 2017; Zhou et al., 2013). The highest pyrolysis oil yield obtained from catalytic co-pyrolysis of Ni/Al2O3 is 17.17 wt.% at 500 °C, followed by a reduction to 12.33 wt.% at 600 °C. Such oil yield reduction is likely due to the increase of gas yield by 10 wt.% at this temperature. Ni/Al2O3 has been shown to improve the formation of gaseous hydrocarbons rather than liquid hydrocarbons during catalytic cracking of OPT and PP (Lin et al., 2020; Singh et al., 2019; Xue et al., 2017). This finding is consistent with the lower amount of liquid hydrocarbons obtained at 600 °C, as shown in Fig. 4 . On the other hand, the gas yield increases with rising temperatures from 500 to 700 °C for three cases (Fig. 3). The secondary reaction of primary volatiles into lighter compounds at higher temperatures results in the formation of non-condensable gases, increasing gas yield with temperature (Hassan et al., 2019). Fig. 4 shows the oil composition obtained from the non-catalytic and catalytic co-pyrolysis of OPT and PP with Ni–Mo/TiO2 and Ni/Al2O3 in the temperature range of 500–700 °C. The oil from non-catalytic co-pyrolysis consists mainly of oxygenated (39.74–52.10%) and phenolic compounds (34.01–41.85%). The oil contains a small amount of hydrocarbons (5.19–10.22%), as evidenced by the relatively low GC-MS relative area for these components. During non-catalytic co-pyrolysis, the oxygenated and phenolic compounds are generated from the thermal decomposition of OPT (i.e., hemicellulose, cellulose, and lignin) (Palamanit et al., 2019; Stefanidis et al., 2014). The thermal degradation of PP produces hydrocarbons via a series of reactions that include random chain scission, mid-chain β-scission, end chain β-scission, radical recombination, and hydrogen transfer reactions (Singh et al., 2019; Xue et al., 2017).When Ni–Mo/TiO2 and Ni/Al2O3 are used as the catalysts in the co-pyrolysis of OPT and PP, the hydrocarbons contained in the oil are significantly increased, as shown by an increase in the GC-MS relative area of up to 54.07–58.18% and 37.28–68.77%, respectively (Fig. 4). The amount of phenolic compounds is reduced, with the reduction in the GC-MS relative area for Ni–Mo/TiO2 (down to 8.46–20.16%) and Ni/Al2O3 (down to 2.93–14.56%). The presence of catalyst generally reduces the amount of oxygenated compounds, although no clear trend can be drawn concerning the parametric effect of temperature and catalyst type. Fig. 5 illustrates the proposed reaction mechanism for the hydrocarbon formation from the analyses based on relevant previous works (Dai et al., 2020; Lin et al., 2020; Singh et al., 2019; Xue et al., 2017). The increase of the hydrocarbon content in the catalytic co-pyrolysis is due to the catalytic cracking of PP and deoxygenation of oxygenated and phenolic compounds promoted by Ni–Mo/TiO2 and Ni/Al2O3 catalyst in addition to the thermal decomposition of PP (Fig. 5).The two catalysts used here rely on the presence of both metal (Ni and Ni–Mo) and acidic (TiO2 and Al2O3) sites to provide high deoxygenation ability and thus improve hydrocarbon production. The oxygenated and phenolic compounds undergo deoxygenation reactions via dehydration, decarbonylation, and decarboxylation to form hydrocarbons (Fig. 5) (Dai et al., 2020). The oxygen in the oil is removed during deoxygenation reactions with water, carbon dioxide, and carbon monoxide released as by-products. The acidic sites in the two catalysts, TiO2 and Al2O3, tend to the occurrence of dehydration reaction over decarbonylation and decarboxylation reactions, resulting in the removal of oxygen from the oil and its subsequent combination with hydrogen to form water as a by-product (Ding et al., 2020). This reaction pathway nonetheless consumes the hydrogen in the oil, which is required to produce hydrocarbon. The presence of metal sites, namely Ni and Ni–Mo, in the two catalysts is expected to partially counteract this pathway, resulting in a more dominant occurrence of decarbonylation and decarboxylation reactions in Ni–Mo/TiO2 and Ni/Al2O3-catalyzed co-pyrolysis of OPT and PP (Balasundram et al., 2018; Dai et al., 2020). Higher acidity of Ni–Mo/TiO2 relative to Ni/Al2O3 (Table 1) due to more abundant acidic sites and synergy between Ni and Mo leads to the formation of a higher amount of hydrocarbons from the catalytic co-pyrolysis of OPT and PP (Fig. 4).During the catalytic co-pyrolysis, the hemicellulose, cellulose, and lignin present in OPT undergo thermal decomposition to produce primary products or intermediates. Afterwards, these products and intermediates diffuse through the pores of Ni–Mo/TiO2 and Ni/Al2O3 and undergo catalytic cracking and deoxygenation reactions to produce secondary products (Balasundram et al., 2018; Lin et al., 2020). The thermal decomposition of hemicellulose primarily yields ketones, furans, and acids, which are then catalytically cracked into smaller oxygenates (i.e., acetic acid, acetone, and simple furans) and olefins on the acidic sites of the catalysts (Dai et al., 2020). Conversely, cellulose is degraded to form anhydrosugars as primary products (Lin et al., 2009). The acidic sites in the two catalysts aid in the dehydration of anhydrosugars to produce more furans. Likewise, the catalytic cracking and deoxygenation of furans form smaller oxygenates and olefins (Dai et al., 2020; Praveen Kumar and Srinivas, 2020). Table 2 shows the decrease of acids and furans in pyrolysis oil after adding Ni–Mo/TiO2 and Ni/Al2O3 catalysts. The result suggests their conversion into olefins, which are the important precursors for the formation of hydrocarbons (Peng et al., 2022). The presence of Ni and Mo in Ni–Mo/TiO2 promotes the decarbonylation and decarboxylation of oxygenated compounds (i.e., ketones, acids, and furans), producing olefins for the subsequent production of hydrocarbons (Balasundram et al., 2018; Xue et al., 2021). Despite this, the amount of ketones in the oil increases after adding these two catalysts (Table 2). This is likely due to catalyst-promoted radical interactions between OPT and PP (Lin et al., 2020).Compared to hemicellulose and cellulose, lignin has a more complex structure, thus producing larger molecules of oligomers during thermal decomposition (Jiang et al., 2010; Lu et al., 2010; Stefanidis et al., 2014). The mesoporous structure of Ni–Mo/TiO2 and Ni/Al2O3 catalysts with wide channels allow for higher diffusion of these lignin-derived oligomers, resulting in high conversion into simple phenols (Lu et al., 2010), which are then converted into olefins via deoxygenation (Hassan et al., 2019; Xue et al., 2017). During the thermal decomposition of PP, olefins can be produced via radical recombination and hydrogen transfer reactions of PP-derived radicals (Singh et al., 2019; Xue et al., 2017). These olefins would produce cyclic hydrocarbons via isomerization and oligomerization. The acidic sites in the catalysts have previously been reported to aid in the isomerization and oligomerization reactions resulting in the formation of cyclic hydrocarbons. Fig. 6 shows a higher amount of cyclic hydrocarbons in the oil derived from the catalytic co-pyrolysis than that from the non-catalytic co-pyrolysis (Peng et al., 2022).Aliphatic hydrocarbons, on the other hand, are produced during PP decomposition through random chain scission, β-scission, radical recombination, and hydrogen transfer reactions (Singh et al., 2019; Xue et al., 2017). Fig. 6 depicts an increase in aliphatic hydrocarbons in the oil produced by catalytic co-pyrolysis compared to non-catalytic co-pyrolysis. The metal sites (i.e., Ni and Ni–Mo) in the two catalysts promote the hydrogen transfer reactions (Peng et al., 2022). The presence of Mo in Ni–Mo/TiO2 promotes the transfer of electrons from Mo to Ni, which enhances the catalyst's electron density and thus improves the hydrogen transfer reaction (Maluf and Assaf, 2009). The lower relative amount of aliphatic hydrocarbons observed in Fig. 6 compared to cyclic hydrocarbons is consistent with the nature of aliphatic hydrocarbons as intermediates. Furthermore, some aliphatic hydrocarbons may go through additional isomerization and oligomerization reactions to become cyclic hydrocarbons, facilitated by the acidic sites of the catalysts (Xue et al., 2017). Table 3 compares the catalytic performances of the catalysts used in this work with other works (Imran et al., 2014; Lu et al., 2010; Mysore Prabhakara et al., 2021). Significantly higher content of hydrocarbons is obtained with the use of Ni–Mo/TiO2 and Ni/Al2O3 as compared to the other TiO2 and Al2O3-based catalysts. However, this is also contributed by the addition of PP as the co-feeding material that provides a sufficient hydrogen source. High oxygenated compounds in the oil reported in the other works are expected, mainly from the decomposition of the wood biomass in the presence of TiO2 and Al2O3-based catalysts.Ni–Mo/TiO2 and Ni/Al2O3 are mesoporous acidic catalysts based on nitrogen adsorption-desorption isotherm and NH3-TPD analyses. Between 500 and 700 °C, the pyrolysis oil yields from the catalytic co-pyrolysis of OPT and PP using Ni–Mo/TiO2 and Ni/Al2O3 were 12.67–19.50 wt.% and 12.33–17.17 wt.%, respectively. The acidic properties of both catalysts enhanced the production of hydrocarbon in oil by facilitating the deoxygenation of oxygenated and phenolic compounds and the catalytic cracking of PP. By adding transition metals (Ni and Mo) into the acidic TiO2 and Al2O3-based catalysts, the deoxygenation mechanism was shifted towards decarbonylation and decarboxylation, removing oxygen from oil as carbon dioxide and carbon monoxide gases, which can conserve hydrogen for hydrocarbon formation. Compared to the non-catalytic co-pyrolysis case, the high amount of cyclic hydrocarbons in oil from catalytic co-pyrolysis with Ni–Mo/TiO2 and Ni/Al2O3 catalysts indicates their high catalytic ability in promoting the isomerization and oligomerization reactions of olefins and aliphatic hydrocarbons.Liza Melia Terry: Methodology, Validation, Formal analysis, Investigation, Visualization, Writing – original draft. Melvin Xin Jie Wee: Methodology, Resources. Jiuan Jing Chew: Supervision, Resources, Writing – review & editing. Deni Shidqi Khaerudini: Resources, Writing – review & editing. Nono Darsono: Resources, Writing – review & editing. Aqsha Aqsha: Conceptualization, Resources, Funding acquisition, Writing – review & editing. Agus Saptoro: Resources, Writing – review & editing. Jaka Sunarso: Supervision, Resources, Writing – review & editing, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Liza Melia Terry gratefully acknowledges the Tun Taib Scholarship from Sarawak Foundation. The authors acknowledge the facilities, scientific, and technical support from Advanced Characterization Laboratories Serpong, National Research and Innovation Agency through E-Layanan Sains, Badan Riset dan Inovasi Nasional. The authors also acknowledge the facilities for GC-MS analysis and funding support from Curtin University Malaysia through Strategic Research Incentives (SRI).The following is the Supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.envres.2023.115550.

Pyrolysis oil from oil palm biomass can be a sustainable alternative to fossil fuels and the precursor for synthesizing petrochemical products due to its carbon-neutral properties and low sulfur and nitrogen content. This work investigated the effect of applying mesoporous acidic catalysts, Ni–Mo/TiO2 and Ni/Al2O3, in a catalytic co-pyrolysis of oil palm trunk (OPT) and polypropylene (PP) from 500 to 700 °C. The obtained oil yields varied between 12.67 and 19.50 wt.% and 12.33–17.17 wt.% for Ni–Mo/TiO2 and Ni/Al2O3, respectively. The hydrocarbon content in oil significantly increased up to 54.07–58.18% and 37.28–68.77% after adding Ni–Mo/TiO2 and Ni/Al2O3, respectively. The phenolic compounds content was substantially reduced to 8.46–20.16% for Ni–Mo/TiO2 and 2.93–14.56% for Ni/Al2O3. Minor reduction in oxygenated compounds was noticed from catalytic co-pyrolysis, though the parametric effects of temperature and catalyst type remain unclear. The enhanced deoxygenation and cracking of phenolic and oxygenated compounds and the PP decomposition resulted in increased hydrocarbon production in oil during catalytic co-pyrolysis. Catalyst addition also promoted the isomerization and oligomerization reactions, enhancing the formation of cyclic relative to aliphatic hydrocarbon.

Biomass represents a sustainable alternative carbon source compared to fossil resources like oil, gas and coal [1–3]. Considering the limited reserves of these fossil resources, growing research efforts are being devoted to the development of efficient catalytic systems for biomass valorisation into biofuels and biobased chemicals [1,3,4]. For the upgrading of biobased compounds into valuable chemicals, metallic catalysts are often required for one or more step(s) in a multi-step reaction that may involve hydrogenation, oxidation and/or hydrogenolysis [3,4]. On one hand, noble metal catalysts, such as Au, Pt, Pd and Ru nanoparticles, often exhibit excellent catalytic performance in specific reactions [3]; on the other hand, their high cost limits the extension of their application from lab-scale to the industry. Moreover, these catalysts often suffer from stability issues since the nanoparticles tend to aggregate and thus decrease their activity under hydrothermal reaction conditions [4,5]. As such, there is a strong need for developing noble-metal-free catalysts, which ideally should have comparable performance and better stability compared to those noble metal catalysts [3]. Among the biobased compounds that typically require the use of noble metal catalysts for their oxidation, glycerol is an attractive platform molecule [6,7]. It is produced in large amounts (above 1 million tons crude glycerol in 2016) as the major side product from the biodiesel industry by transesterification of vegetable oils with methanol [4,8]. This led to an oversupply of glycerol and, therefore, has prompted both academia and industry to develop efficient catalytic routes to convert it into several valuable chemical products [9–11]. Lactic acid and alkyl lactates can be produced from glycerol through a dehydrogenation-rearrangement pathway (Scheme 1 ) [8,12–14]. Lactic acid has a wide range of applications, including that as monomer of poly-lactic acid, a biodegradable bio-polymer with various applications in the food, pharmaceutical and packaging industry [12]. Currently, lactic acid is produced by fermentation of carbohydrates, which generates large amounts of salts in the product work-up section and has a relatively low volumetric production rate [15,16]. The chemocatalytic route involving the dehydrogenation of glycerol and consecutive rearrangement of the triose intermediates (Scheme 1) is considered a viable, sustainable alternative to the fermentation process [12]. This chemocatalytic route implies a nominal formation of H2 and in this sense can be correlated to the use of glycerol as feedstock for the sustainable production of H2 through aqueous-phase reforming (APR) [7,11,17]. Hydrogen is widely used in current chemical industry (e.g. ammonia synthesis, Fischer-Tropsch process, steel industry and various hydrogenation reactions) and in the power fuel cell systems as a clean power source [2,11,18]. Clearly, routes that allow producing H2 from a renewable source such as biomass represent a sustainable alternative to the current production through methane steam reforming, which is based on a fossil resource and requires extremely harsh conditions [2,19].The conversion of glycerol into lactic acid requires metallic sites for the first step, i.e. the dehydrogenative oxidation, and a base or a combination of Brønsted and Lewis acid sites for the second step (Scheme 1). Most studies used noble metal catalysts for the first step, such as Pt, Pd, Au and their alloys [12,20–22]. Pt/C was used for the hydrogenolysis of glycerol under He atmosphere and gave 55% selectivity to lactic acid at 95% conversion of glycerol [23,24]. Supported Au and its alloy catalysts (AuPt/TiO2) were firstly used with O2 as the oxidant, reaching 30% glycerol conversion and 86% selectivity to lactic acid at 90 °C [21]. The first report of a bifunctional catalyst for the conversion of glycerol into lactic acid without adding a base employed Pt supported on a zeolite (Sn-MFI) and achieved an excellent 81% selectivity towards lactic acid at 90% conversion of glycerol under O2 (6 bar) at a relatively mild temperature (90 °C) [14]. Catalysts based on non-noble transition metals, such as Ni, Co and Cu, were also found to be active in converting glycerol to lactic acid under inert atmosphere in the presence of a base [20,25–29]. A Ni/graphite catalyst tested at 250 °C for 2 h yielded 89% lactic acid at full glycerol conversion [20]. A series of 30%CuO/ZrO2 catalysts were also developed and reached 95% yield of lactic acid at 200 °C [29]. A recent study reported a 20%Co3O4/CeO2 catalyst that achieved 80% selectivity to lactic acid with 85% glycerol conversion at 250 °C for 8 h [27]. All these non-noble metal catalysts were employed in the presence of a homogeneous base (NaOH) and at relatively high reaction temperatures (200–250 °C), under which conditions the base alone would display a significant activity in the conversion of glycerol to lactic acid [30,31]. An additional drawback of the Ni, Cu and Co-based systems is the high metal-to-glycerol ratio that was needed to achieve acceptable reaction rates. Moreover, the Cu and Co-based catalysts suffered remarkable loss of activity upon reuse, probably due to leaching of metal species under the hydrothermal conditions [27,29]. If the conversion of glycerol to lactic acid (salt) is carried out under inert atmosphere, the initial dehydrogenative oxidation step (Scheme 1) nominally liberates one molecule of H2 per molecule of glycerol [14,25]. However, the hydrogen generated in such system is highly diluted by N2 in most cases and is thus difficult to collect. In this context, it is more attractive to utilise in-situ the hydrogen removed from glycerol in the reduction of relevant target compounds. Here, we report a bimetallic Ni-Co catalyst supported on CeO2 with remarkably high activity in the transfer hydrogenation between glycerol and several H2 acceptors, under relatively mild hydrothermal conditions (160 °C) and in the presence of NaOH as promotor. The choice of investigating a Ni-based catalyst was inspired by the above-mentioned activity of this metal in converting glycerol to lactic acid, combined with its well-known activity in catalysing hydrogenation reactions as significantly cheaper alternative to noble metals (e.g. Pt and Pd) [3,32]. The idea of using Ni in a bimetallic system was justified by previous reports that showed that the catalytic performance of Ni could be enhanced by incorporating another component, such as Co or Cu, which led to stronger metal-support interaction with consequent smaller metal particle size [3,4,33]. Namely, bimetallic Ni-based catalysts supported on ZrO2 showed much better performance in the dry reforming of methane (Ni-Co) or in the oxidative steam reforming of methanol (Ni-Cu) compared to their monometallic counterparts [34–36]. In this work, different oxides were tested as support for the Ni-based catalysts, with CeO2 leading to the highest activity in glycerol conversion. Our bimetallic Ni-Co catalytic system was also compared to its monometallic counterparts, showing higher activity and allowing to reach very high conversion of glycerol with excellent selectivity towards lactic acid, and to combine this reaction with the efficient hydrogenation of several unsaturated compounds in a one-pot process.Glycerol (99%), 1,3-dihydroxyacetone dimer (97%), glyceraldehyde (90%), glycolic acid (99%), lactic acid (98%), pyruvic aldehyde (40 wt% in H2O), cyclohexene (99%), cyclohexane (99.5%), sodium hydroxide (98%), benzene (99.9%), levulinic acid (99%), 4-hydroxypentanoic acid, γ-valerolactone (99%), nickel(II) nitrate hexahydrate (98.5%), cobalt(II) nitrate hexahydrate (98%), copper(II) nitrate hemi(pentahydrate) (98%), titanium oxide (P25), magnesium oxide (99%) cerium oxide (nanopowder, nominally < 25 nm, though some large particles were observed by TEM; this compound is denoted as CeO2 for the sake of simplicity, though it contains both CeIV and CeIII and it is thus actually CeO2-x), zirconium oxide (nanopowder, < 100 nm) were purchased from Sigma Aldrich. Glyceric acid (20 wt% in H2O), nitrobenzene (99.5%), aniline (98%), azobenzene (98%), azoxybenzene (98%) were purchased from TCI Chemicals. Active carbon Norit SX1G was purchased from Cabot. The H2O used in this work was always of MilliQ grade. All chemicals were used without further purification.A wet impregnation method was used for the preparation of catalysts based on Ni, Co, Cu, NiCo, NiCu supported on CeO2 and ZrO2. Typically, CeO2 (2 g) was mixed with an aqueous solution of Ni(NO3)2 or Co(NO3)2 or Cu(NO3)2 or the combination of two of them (2 M, with the volume of the solution being defined by the target loading of Ni, Co and Cu). The slurry was stirred at room temperature until the water evaporated. The solid mixture was then dried at 100 °C overnight. The resulting solids were milled to fine powder and then calcined at 550 °C in the oven under static air (heating rate 3 °C/min). The calcined catalysts were further reduced in a tube oven under H2 flow (99.9% and 200 mL/min) at 400 °C (heating rate 3 °C/min) for 2 h. The gas flow was switched to N2 for 1 h to wipe away the adsorbed H2 on the catalyst surface before taking the catalyst out from the tube oven. A typical reduced catalyst prepared by this method was named as 10NiCo/CeO2, in which 10 stands for the total loading of Ni and Co (wt%), in which the mass ratio between Ni and Co is always kept as 1:1. In addition, as a reference, the catalyst was also used directly after calcination at 550 °C without further reduction in H2, which was named as 10NiCo/CeO2-C.The catalytic experiments were carried out in a 100 mL Parr stainless steel autoclave reactor equipped with a Teflon liner and an overhead stirrer. In a typical test, a predetermined amount of the catalyst together with a mixture of aqueous solution of glycerol (0.5 M in 20 mL), NaOH (0.15 mol) and the selected hydrogen acceptor (0.2 mol, as organic phase) were loaded into the reactor. The reaction was performed under N2 (20 bar) for 4.5 h at 160 ᵒC (extra heating time 0.5 h) at a stirring speed of 800 rpm. Next, the reactor was depressurised and the reaction content (in two phases) was taken separately and filtered to remove the catalyst. The organic phase was analysed by gas chromatography using a Thermo Trace GC equipped with a Restek Stabilwax-DA column (30 m × 0.32 mm ×1 μm) and a FID detector. The aqueous phase was first neutralised and diluted by H2SO4 (1 M), then analysed by high performance liquid chromatography (HPLC, Agilent Technologies 1200 series, Bio-Rad AminexHPX-87H 300 × 7.8 mm column) at T = 60 °C, with 0.5 mM H2SO4 as eluent (flow rate: 0.55 mL/min) using a combination of refractive index detector and ultra-violet detector. For the analysis of nitrobenzene and its products, conversion and selectivity were determined by GC analysis using an Agilent Technologies 7980B GC equipped with an Agilent DB-5#6 (5%-Phenyl)-methylpolysiloxane column (15 m, 320 μm ID). The identification of the products was performed by GC-mass spectrometry (GC–MS) on an HP 6890 Series GC equipped with a Restek Rxi-5Si MS fused silica column (30 m, 250 μm ID) coupled to an HP 5973 Mass Selective Detector. Each component was calibrated using solutions of the individual components at 4 different concentrations.For the catalyst recycle tests, a small amount of the reaction mixture was collected for analysis, the remaining mixture was filtered and the catalyst was recovered. The catalyst was washed first with H2O (20 mL), then with ethanol (20 mL), and this procedure was repeated 3 times, after which the solid was dried overnight at 100 ᵒC. This solid was used for another batch experiment.Transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM) and energy-dispersive X-ray spectroscopy (EDX) mapping measurements were performed on a FEI Tecnai T20 electron microscope operating at 200 keV with an Oxford Xmax 80 T detector. The samples were prepared by ultra-sonication in ethanol followed by drop-casting of the material on a copper grid.Nitrogen physisorption isotherms were measured at −196 °C using a Micromeritics ASAP 2420 apparatus. The Brunauer-Emmet-Teller (BET) method was used to calculate the specific surface area. The Barrett-Joyner-Halenda (BJH) method was used to calculate the pore volume.Inductively-coupled plasma optical emission spectrometry (ICP-OES) was performed using a Perkin Elmer Optima 7000 DV instrument in order to obtain the actual metal loadings on the supports.X-ray photoelectron spectroscopy (XPS) was measured by mounting the catalysts on a conductive tape adhered to the XPS sample holder. No further treatment was carried out prior to the XPS measurement. Then, the sample was loaded into the load lock and the pressure was reduced below 1·10−7 mbar. The XPS measurements were performed using a Surface Science SSX-100 ESCA instrument equipped with a monochromatic Al Kα X-ray source (hν =1486.6 eV). During the measurement, the pressure was kept below 2·10-9 mbar in the analysis chamber. For acquiring the data, a spot size with a 600 μm diameter was used. The neutraliser was on to avoid charging effects. All XPS spectra were analysed using the Winspec software package developed by LISE laboratory, University of Namur, Belgium, including Shirley background subtraction and peak deconvolution.Hydrogen-temperature programmed reduction (H2-TPR) measurements were performed on an Autochem II 2920 from Micromeritics. In a typical experiment, 80 mg of sample was pre-treated at 500 °C (heating rate 10 °C/min) for 1 h in a flow of He (30 mL/min). Subsequently, the sample was cooled down to 50 °C under the same flow of He. The reduction analysis was performed from 50 to 900 °C (10 °C/min) in a 30 mL/min flow of 5 vol.% H2 in He.X-ray diffraction (XRD) measurements were performed on a D8 Advance Bruker diffractometer with a CuKα 1 radiation (λ = 1.5418 Å). The XRD patterns were collected under 40 kV and 40 mA in the range of 10-80°.Definitions:The glycerol conversion (Conv./%) is defined by Eq. (1): (1) Conv . = C g , 0 - C ( g ) C ( g , 0 ) × 100 % in which C(g) is the molar concentration of glycerol after a certain reaction time and C(g,0) is the initial concentration of glycerol.Product selectivity for a compound P is defined by Eq. (2): (2) S p   = C ( p ) C ( g , 0 ) - C ( g ) × 100 % in which C(p) is the molar concentration of a product after a certain reaction time.The yield of transfer hydrogenation is defined by Eq. (3): (3) Y trans - H  = ∑ ( x * n p 1 ) n ( g , 0 ) - n ( g ) × 100 % in which x is the number of hydrogen atoms needed for the reduction of product 1, n(p1) is the molar amount of product 1, n(g) is the molar amount of glycerol after a certain reaction time and n(g,0) is the initial molar amount of glycerol.The term “lactic acid” is used in this article to describe the product obtained from the reaction mixture, which is actually sodium lactate (mixed with a small portion of lactic acid from hydrolysis).Our study of the conversion of glycerol into lactic acid coupled with the transfer hydrogenation to an unsaturated compound started with the investigation of the catalytic behaviour of Ni catalysts (10 wt%) as a function of the type of the material (activated carbon (AC) and various metal oxides) on which the metal particles were supported by wet impregnation. The five catalysts were tested at 160 °C in the presence of NaOH as promotor and using a model compound as cyclohexene as the H2 acceptor (Table 1 ). Ni supported on AC, MgO and TiO2 showed relatively low activity (entries 1–3, Table 1), whereas the activity was significantly higher when nanosized CeO2 and nanosized ZrO2 were used as support for Ni (glycerol conversion 53% and 63%, respectively; entry 4–5, Table 1), in line with previous reports on other (de)hydrogenation reactions [4,33]. In all cases, high selectivity towards lactic acid (> 91%) was observed. This is attributed to the presence of NaOH, which effectively promotes the deprotonation of glycerol and catalyses the successive isomerisation of the intermediates (glyceraldehyde and dihydroxyacetone) into the lactic acid salt (Scheme 1), thus granting very high selectivity towards the desired product [21,22,28,37]. Small amounts of glyceric acid, glycolic acid and propanediol were detected as side products (Table 1). Glyceric acid is formed through the further dehydrogenation of glyceraldehyde and glycolic acid probably originates from oxidative CC bond cleavage of glyceric acid [13]. Propanediol (as a mixture of 1,2- and 1,3-isomers) probably forms via the hydrogenolysis of glycerol [38–40]. In addition, for all reactions, very minor amounts of glyceraldehyde and propanoic acid were observed as side products, with selectivity below 0.2% for each of them.Based on this preliminary study, CeO2 and ZrO2 were selected as supports for further study of Ni-based catalysts. Then, we aimed at improving the catalytic performance by incorporating another metallic component, i.e. Co or Cu [3,4,33]. The activity of the bimetallic catalysts was compared to the monometallic counterparts (Table 2 ), while keeping the total loading of metal at 10 wt% (and with 1:1 mass ratio for the bimetallic systems). The incorporation of Co into the catalyst formulation was highly beneficial when CeO2 was used as support (10NiCo/CeO2), leading to 91% glycerol conversion (entry 1, Table 2) compared to 53% conversion obtained over 10Ni/CeO2 and 46% conversion over 10Co/CeO2 (entry 5, Table 2). Also the incorporation of Cu enhanced the activity compared to the monometallic counterparts, though the effect was less marked (compare entry 2 in Table 2 to entry 5 in Table 1 and entry 6 in Table 2). On the other hand, the 10NiCo/ZrO2 catalyst showed almost the same activity as the monometallic 10Ni/ZrO2, (compare entry 3 in Table 2 with entry 4 in Table 1), whereas the incorporation of Cu proved more beneficial when ZrO2 was the support, reaching 80% glycerol conversion (entry 4, Table 2). These results indicate a complex interplay between the type of metals and the supports. The benefit brought about by the bimetallic formulation will be elucidated further in the case of the optimum catalyst, i.e. 10NiCo/CeO2 (vide infra). In all these tests, the selectivity towards lactic acid remained very high (94–96%). Glyceric acid, glycolic acid and propanediol were detected as the main side products, with selectivity < 6% in total. Though the incorporation of Cu enhanced the activity of the Ni-based catalysts, leaching of metal species was observed in the basic medium under hydrothermal conditions, with significant amount of brown Cu-containing precipitate deposition on the stirring bar and reactor walls [28,33,41–43]. Therefore, 10NiCo/CeO2 was selected for further investigation aimed both at a deeper evaluation of the catalytic performance and at understanding the relationship between structure and catalytic behaviour.The catalysts presented in this work were prepared by wet impregnation, followed by calcination and finally reduction by H2. The actual loading of Ni and/or Co determined by ICP-OES (Table 3 ) was found to be very close to the nominal 10 wt% loading. In the bimetallic Ni-Co catalyst, the actual loading of Ni and Co is 5.6 wt% for both metals, which is slightly higher than the theoretical 5 wt%. The BET surface area was measured before and after loading Ni and Co, showing only a slight decrease (from 32 to 28 m2/g) compared to the fresh CeO2 support.To investigate the possible organisation of Ni, Co and Ni-Co species in crystalline phases on the CeO2 support, the catalysts were further characterised by XRD before and after reduction (Fig. 1 ). The materials before reduction (Fig. 1A) display the characteristic peaks of the CeO2 support together with the typical peaks of NiO (in 10Ni/CeO2-C) or Co3O4 (in 10Co/CeO2-C) [34,35,44]. The bimetallic 10NiCo/CeO2-C shows a broad peak at 36.7°, which is slightly shifted compared to the Co3O4 peak (37°) and has been attributed to the mixed oxide NiCo2O4 [34,45–47]. After reduction at 400 °C in H2 flow, besides the peaks of the CeO2 support, only one peak at 44.7° belonging to metallic Ni can be seen in the pattern of 10Ni/CeO2 (Fig. 1B). On the other hand, no signals stemming from Co and/or Ni phases were observed in 10Co/CeO2 and 10NiCo/CeO2. These results suggest that relatively large crystalline Ni particles formed upon reduction in 10Ni/CeO2, while the Co or Ni-Co species obtained after reduction were highly dispersed in the other two catalysts [45–48].To achieve deeper insight on the dispersion of Ni, Co and bimetallic Ni-Co catalysts supported on nanosized CeO2, TEM and STEM-EDX-mapping were used to investigate the average size of these metallic domains (Figure S1, 2 and 3). Since the atomic mass of cerium is much higher than that of nickel or cobalt, it is hard to determine the particle size of Ni, Co or Ni-Co alloy on the CeO2 support based on TEM pictures (Figure S1), as the darker zones are not necessarily corresponding to Ni or Co domains.Analysis by STEM coupled with EDX mapping was more informative as it allows identifying the elemental composition within the image (Fig. 2 ). The large green domains in Fig. 2A and 3B indicate the presence of Ni-containing nanoparticles on CeO2. Based on the XRD data (Fig. 1A), these domains are identified as large NiO nanoparticles (mainly around 100 nm, with some smaller particles, see Fig. 2A) in the sample before reduction (10Ni/CeO2-C), and to large domains of metallic Ni (around 75 nm, Fig. 2B) after the sample was reduced (10Ni/CeO2). For the monometallic material prepared by supporting Co on CeO2 and prior to reduction (10Co/CeO2-C), the Co3O4 identified by XRD (Fig. 1A) was found to be better dispersed on the CeO2 support (Fig. 2C) compared to NiO on CeO2. The 10Co/CeO2 material obtained upon reduction showed nearly homogeneously dispersed Co species (Fig. 2D), which indicates that the particle size of Co is lower than the detection limit of EDX-mapping (around 30 nm). The relatively small size of the Co nanoparticles is also in agreement with the absence of any signal due to metallic Co in the XRD pattern of 10Co/CeO2 (Fig. 1B), which suggests a strong metal-support interaction between Co and CeO2 [4,33,35,45,46,48].STEM and EDX-mapping of the Ni-Co bimetallic material prior to reduction (10NiCo/CeO2-C), showed that both Ni and Co are nearly homogeneously dispersed on the CeO2 surface (Fig. 4 A–D). This demonstrates that the presence of Co prevents the aggregation of Ni species, in contrast to the large domains observed in 10Ni/CeO2-C. After reduction at 400 °C under H2, Ni and Co still preserve very good dispersion, with no large metal particles (i.e. > 30 nm) being visible (Fig. 3 H). The strong interaction between Co and the CeO2 support, which promotes the observed high dispersion of both Co and Ni on the surface, has been shown to be related to the formation of a thin layer of reduced CeOx at the interface with the metallic Co [35]. Based on our results, we infer that this feature prevents Ni from forming large particles in the process of calcination and reduction [33,35,46].The reducibility of Ni, Co and Ni-Co supported on CeO2 was further investigated by H2-TPR (Fig. 4). The support, CeO2, exhibited two dominant peaks centred at 490 °C (from 300 to 550 °C) and 880 °C (from 700 to above 900 °C), which are attributed to the reduction of surface ceria and bulk ceria, respectively [35,49]. Besides the reduction peaks of CeO2 at 420 and 815 °C, which are slightly shifted to lower temperature, the monometallic 10Ni/CeO2-C displays two peaks at 213 °C (minor) and 320 °C (dominant), which are attributed to the reduction of adsorbed oxygen and NiO, respectively [35,50]. The monometallic 10Co/CeO2-C showed two main peaks at 260 and 315 °C, which are attributed to the two-step reduction Co3O4→CoO→Co [51,52]. The large and broad shoulder extending from 350 to 500 °C is probably due the reduction of surface CeO2. Compared to the monometallic Ni catalyst, the significant increase of the intensity of the reduction peak of surface CeO2 in the monometallic Co catalyst supports the existence of a strong metal-support interaction between Co and CeO2, which is in agreement with the formation of a thin layer of reduced support on the metallic Co surface reported in the literature [35,48]. The 10NiCo/CeO2-C material showed almost identical profile as the one of 10Co/CeO2-C, with all the peaks shifted by ca. 5 °C to lower temperature. This suggests that, in the bimetallic Ni-Co catalyst, the reduction behaviour is mainly dictated by the presence of Co, including the strong metal-support interaction indicated by the broad shoulder between 350 and 500 °C. This result explains the observed much better dispersion of the metal species in the bimetallic Ni-Co catalyst compared to the monometallic Ni catalyst (Figs. 2 and 4) [35].The characterisation by EDX-mapping and H2-TPR indicates a geometrical effect of the presence of Co on the dispersion of Ni on the CeO2 support. To investigate further the interaction between Co, Ni and the support, selected catalysts were analysed by XPS (Figure S2-4). The XPS signal of the Ni 2p3/2 core level region of the unreduced 5Ni/CeO2-C catalyst was deconvoluted into 3 main peaks: at 853.6 eV, assigned to NiO; at 855.6 eV, attributed to Ni(OH)2 and/or NiO(OH); and a satellite peak at 860.6 eV [53–56]. Similar peaks were identified by deconvoluting the Ni 2p3/2 signal of the unreduced 10NiCo/CeO2-C catalyst (Figure S2.A and B). After reduction (Figure S2.C and D), in addition to the 3 peaks mentioned above, the deconvolution allowed identifying a peak ascribed to Ni0 (at 852.3 eV) in catalysts 5Ni/CeO2 and 10 NiCo/CeO2 [54,55,57]. These data confirm the successful reduction to metallic Ni. The fact that the majority of the XPS signal stems from oxidised Ni species can be explained considering that XPS is a surface technique (information from the top 1–10 nm of the material) and that the surface of the particles is expected to tend to oxidise in contact with air and moisture [58,59]. The XPS signal of the Co 2p3/2 core level region of the unreduced 5Co/CeO2-C catalysts was deconvoluted into 3 main peaks: at 779.5 eV, assigned to cobalt oxides (CoO and/or Co3O4); at 781.5 eV, ascribed to Co(OH)2; and a satellite peak at 785.5 eV [55,60,61]. Analogous peaks were identified by deconvoluting the Co 2p3/2 signal of the unreduced 10NiCo/CeO2-C catalyst (Figure S3.A and B). After reduction (Figure S3.C and D), in addition to the 3 peaks mentioned above, the deconvolution showed a peak assigned to Co0 (at 778.0 eV) in the catalysts 5Co/CeO2 and 10 NiCo/CeO2 [55,62]. Similarly to what discussed in the case of the supported Ni particles, the presence of oxidised Co species in the reduced samples is attributed to the formation of a layer of oxides and hydroxides at the surface of the particles, generated by contact with air and moisture. The features of the XPS signal of the Ce 3d core level region support the anticipated strong interaction between Co and CeO2 (Figure S4). This is indicated by the surface reduction of Ce4+ and the increase in Ce3+ observed in the XPS spectra of the Co-containing catalysts (whereas this effect is absent in the spectra of the catalysts containing Ni but no Co). This matches well with the literature and with our H2-TPR results [35,52,54]. The XPS data are not conclusive on possible synergistic electronic effects between Ni and Co. Therefore, we infer that the main reason for the improved catalytic performance of bimetallic 10NiCo/CeO2 is the smaller size and better dispersion of the Ni-containing particles.Based on this characterisation study, the optimum activity observed with the bimetallic 10NiCo/CeO2 catalyst is attributed to presence of the more active Ni compared to the monometallic 10Co/CeO2, and to the better dispersion of the active metallic species compared to the monometallic 10Ni/CeO2 catalyst. To further confirm the nature of the active sites, unreduced Ni, Co and bimetallic Ni-Co catalysts were tested under the same conditions employed for the reduced catalysts (Table S1). In the unreduced materials, the metal oxides (NiO, Co3O4 and NiCo2O4) would be the catalytic sites rather than the metallic sites. All the unreduced catalysts had significantly lower activity compared to the reduced ones (Table 1 and 2), with the conversion of glycerol being < 16% in all cases. These results confirm that the metallic sites are the active site in this transfer hydrogenation reaction between glycerol and cyclohexene, in agreement with what shown in the literature [27–29].The Ni, Co and Ni-Co catalysts with different loading (2, 5 and 10 wt%) supported on CeO2 were tested to gain better understanding on the effect of the Ni and Co composition (Fig. 5 ). With the Ni/CeO2 catalysts, the conversion of glycerol increased with the metal loading up to 5 wt% Ni, at which it reached 55%, whereas it remained nearly constant upon further increase to 10 wt % of Ni. This trend is completely different from the one observed with the Co/CeO2 and NiCo/CeO2 catalysts, for which the glycerol conversion and the lactic acid yield exhibited an increasing trend with the increase in metal loading (Fig. 5A). The performance of these catalysts can be analysed also in terms of turnover number (TON) (Fig. 5B). These data show that the TON is nearly constant as a function of metal loading for the monometallic Co-catalysts, whereas an increasing loading of Ni causes a gradual decrease in TON, which is more marked for the monometallic Ni-catalysts compared to the bimetallic Ni-Co materials. These trends are in agreement with the tendency of Ni to form large particles at high loading (see Fig. 3.A–B), which implies that a smaller fraction of the metal is available to act as active site, thus leading to the observed lower TON. On the other hand, Co maintains small metallic domains on the CeO2 surface also at 10 wt% metal loading (Fig. 2.C–D), thus enabling to have a nearly constant TON as a function of metal loading. The highest TON was observed for 2Ni/CeO2 and 2NiCo/CeO2, whereas among the catalysts with 10 wt% metal loading, the highest TON was found for 10NiCo/CeO2, despite the decrease compared to the 2 wt% material. This confirms the higher intrinsic activity of Ni compared to Co in catalysing the dehydrogenative oxidation of glycerol. Non-noble metal catalysts are generally used with high loading to give high productivity. Indeed, when the catalytic performance is compared in terms of productivity (Fig. 5C) the highest value among the tested catalysts is obtained with the material with the highest TON among those with 10 wt% metal loading, i.e. 10NiCo/CeO2. This underlines the benefit of the presence of Co in combination with Ni on the catalytic performance [33–35,44,48].The 5NiCo/CeO2 catalyst, which achieved intermediate glycerol conversion at 160 °C, was selected for investigating the effect of the reaction temperature (in the range 140 to 200 °C, Figure S2). The conversion of glycerol increased with higher reaction temperature, from 11% (at 140 °C) to 99% (at 200 °C), while the selectivity to lactic acid remained > 98%. The selectivity towards the transfer hydrogenation was steady at around 25% in all range of temperatures. It should be noted that, when only NaOH was used in the reaction system, the conversion of glycerol was rather low, though it increased from 1.6 to 16% (from 140 to 200 °C, Figure S5). This demonstrates the need for a heterogeneous catalyst to carry out the dehydrogenation reaction in this range of relatively mild temperatures [30,31].To further investigate the effects of the catalyst amount on this reaction, different weights of catalyst (from 0.025 to 0.15 g) were used, while all other parameters were kept constant. The results show a gradual increase in the conversion of glycerol from 29% to > 99.9% upon increase of the loading of the 10NiCo/CeO2 catalyst (Figure S6).The role of NaOH was studied in more detail by varying the molar ratio between NaOH and glycerol (from 0 to 2, Figure S7). Without the addition of NaOH, both the conversion of glycerol and the selectivity to lactic acid were very low (conversion of glycerol = 3.5%). If the molar ratio between NaOH and glycerol was increased, the conversion of glycerol gradually increased reaching 91% with 85% yield of lactic acid salt at NaOH/glycerol = 1.5. However, a further increase in the NaOH/glycerol molar ratio to 2 caused a decrease in the conversion of glycerol to 81%, thus indicating that the employed ratio (1.5) is the optimum value. These results confirm that the presence of a base like NaOH in the reaction mixture is critical to induce the deprotonation of one of the hydroxyl groups of glycerol, thus promoting the dehydrogenation of glycerol [21,28]. Moreover, NaOH can catalyse the isomerisation of glyceraldehyde and dihydroxyacetone and lead to the formation of sodium lactate with very high selectivity.The reaction profile as a function of the reaction time was studied with the 10NiCo/CeO2 catalyst (Figure S8). The conversion of glycerol increased almost linearly within the first 4.5 h, corresponding to a productivity of lactic acid of 17.4 g(LA)/(g(metal)h). After 6.5 h of reaction, almost complete glycerol conversion (97%) was achieved, with 93% lactic acid (salt) yield. The selectivity towards lactic acid stayed always above 90% and the total selectivity towards by-products (glyceric acid, glycolic acid and propanediol) was around 4%. The selectivity towards the transfer hydrogenation slightly decreased with the reaction time, from 31% to 26%. These results suggest that under the employed reaction conditions the dehydrogenation of glycerol is the rate-determining step, and that once the dihydroxyacetone and/or glyceraldehyde formed, they would be transformed into lactic acid (salt) in a very fast and selective way.Catalyst 10NiCo/CeO2 was also selected for a reusability test (Fig. 6 ). The fresh catalyst showed 91% conversion of glycerol and 85% yield to lactic acid, while recycling after straightforward washing and drying led to a slight, gradual decrease in activity. After 5 runs, the conversion of glycerol decreased to 73%, while the selectivity towards lactic acid remained unaltered (> 94%). Meanwhile, the selectivity in the transfer hydrogenation gradually increased from 24 to 28% between the first and the fifth run. The gradual loss of activity is probably caused by the leaching of a small fraction of the active components in the alkaline hydrothermal reaction system, since the loading of Ni and Co decreased from 5.6 wt% (each) in the fresh catalyst to 4.4 wt% (each) after 5 runs (entry 5, Table 3).During the optimisation of the Ni-based catalyst presented above, cyclohexene was employed as hydrogen acceptor in the transfer hydrogenation reaction from glycerol. To expand the scope of applicability of the transfer hydrogenation, we tested a set of H2 acceptors with different features (a biobased compound as levulinic acid, an aromatic compound as benzene, a compound containing both an aromatic ring and another reducible group as nitrobenzene and a linear, terminal alkene as 1-decene). While cyclohexene and 1-decene were selected as model compounds, the hydrogenation of benzene, nitrobenzene and levulinic acid is of potential industrial relevance [63–69]. The tests were carried out with a 1:1 molar ratio between glycerol and the hydrogen acceptor, at 160 °C under N2 atmosphere (Scheme 2 and Table 4 ).When levulinic acid was employed as the H2 acceptor, two main products were observed: 4-hydroxypentanoic acid (27% yield), obtained by hydrogenation of the carbonyl group of levulinic acid, and γ-valerolactone (48% yield), obtained by subsequent dehydration (Scheme 2 and entry 1 in Table 4). γ-Valerolactone can be used as food additive, solvent and precursor for polymers [6,68,70,71]. This reaction also gave an 86% yield of lactic acid at 87% glycerol conversion with a very good 88% selectivity in the transfer hydrogenation.When benzene was tested as H2 acceptor, a very high selectivity (97%) in the transfer hydrogenation from glycerol was observed, with cyclohexane being the only product (corresponding to complete reduction of benzene). The reduction of benzene is the industrial route for the production of cyclohexane, which is employed as precursor in the synthesis of adipic acid used in the manufacturing of nylon [72,73]. The yield achieved here (25%) is promising considering that under the employed reaction conditions (1:1 molar ratio between glycerol and benzene), the maximum theoretical yield of cyclohexane is 33%. These results were coupled with 79% conversion of glycerol and 77% yield of lactic acid (entry 2, Table 4).When nitrobenzene was employed as hydrogen acceptor, the reduction of the nitro group is expected to be favoured over the reaction of the aromatic ring. Indeed, the observed products (azoxybenzene with 59% yield, azobenzene with 18% yield and aniline with 7.5% yield) all originate from the reduction of the nitro group (Scheme 2) [63,74–76]. These are all industrially valuable products, with azoxybenzene being utilised in dyes, reducing agents and polymerisation inhibitors; azobenzene being used in dyes, indicators and as additive in polymers; and aniline finding application in producing pesticides, dyes and as the precursor to polyurethane [77–79]. For this reaction, the selectivity in the transfer hydrogenation from glycerol was > 100%. This can be explained considering the strong oxidative ability of nitrobenzene, which led to the further oxidation of the triose intermediates to glyceric acid and glycolic acid (entry 3, Table 4), similarly to what is generally observed in the oxidation of glycerol in the presence of O2 [25,80–82]. Therefore, glyceric acid (52% yield) becomes the major product under these conditions, with lactic acid being obtained in much lower yield (23%).When 1-decene was selected as a linear H2 acceptor with a primary CC bond, 92% conversion of glycerol and 91% yield of lactic acid was achieved after reaction, while 85% of decene was hydrogenated to decane, corresponding to a remarkably high 94% selectivity in the transfer hydrogenation (entry 4, Table 4). This is much higher than what was found when using cyclohexene as the H2 acceptor (entry 5, Table 4). This result is probably due to the higher accessibility of the CC bond in a linear alkene with a terminal double bond as 1-decene compared to the more sterically-hindered cyclohexene.The study of substrate scope for the transfer hydrogenation reaction from glycerol demonstrated that our catalytic system based on 10NiCo/CeO2 is able to efficiently promote the conversion glycerol to lactic acid while exploiting the liberated hydrogen in the reduction of different unsaturated compounds to achieve the synthesis of useful target products without requiring an external H2 source.Bimetallic Ni-Co catalysts supported on CeO2 were prepared and tested for the transfer hydrogenation from glycerol to various unsaturated compounds, in which lactic acid and the corresponding hydrogenated products were obtained in a one-pot batch reaction. Introducing Co into the formulation of the Ni-based catalysts was crucial to prevent the aggregation of Ni into large particles. This was proven by the higher activity of the bimetallic 10NiCo/CeO2 catalyst compared the Ni- or Co-based counterparts, and by characterisation of the catalytic materials by EDX-mapping and H2-TPR, which demonstrated the high dispersion of Ni-Co sites on the CeO2 support. The bimetallic 10NiCo/CeO2 catalyst exhibited very high activity (91% glycerol conversion) and selectivity to lactic acid (94%) at 160 °C, 4.5 h under N2 atmosphere in the presence of NaOH as promoter. This result demonstrates that excellent conversion and selectivity can be achieved using a catalyst with a relatively low loading of Ni and Co and that operates at milder reaction conditions compared to other non-noble metal catalysts for glycerol dehydrogenation reactions [20,25–29]. Moreover, various H2 acceptors (levulinic acid, benzene, nitrobenzene, 1-decene, cyclohexene) were tested in the transfer hydrogenation from glycerol, exploiting in-situ the hydrogen liberated in the dehydrogenative oxidation of glycerol to generate several useful products.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.We would like to thank the financial support from the China Scholarship Council for the Ph.D. grant of Zhenchen Tang, the technical support from Leon Rohrbach, Jan Henk Marsman, Erwin Wilbers, Anne Appeldoorn and Marcel de Vries, the TEM-EDX support from Dr. Marc Stuart and the ICP-OES support from Johannes van der Velde. We also acknowledge Dr. Matteo Miola for useful scientific discussion of the XPS data.Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.apcatb.2019.118273.The following is Supplementary data to this article:

Bimetallic Ni-Co catalysts supported on nanosized CeO2 were prepared and investigated as heterogeneous catalysts for the transfer hydrogenation between glycerol and various H2 acceptors (levulinic acid, benzene, nitrobenzene, 1-decene, cyclohexene) to selectively produce lactic acid (salt) and the target hydrogenated compound. The bimetallic NiCo/CeO2 catalyst showed much higher activity than the monometallic Ni or Co counterparts (with equal total metal mass), thus indicating strong synergetic effects. The interaction between the metallic sites and the CeO2 support was thoroughly characterised by means of transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM), energy-dispersive X-ray spectroscopy (EDX) mapping, X-ray photoelectron spectroscopy (XPS), hydrogen-temperature programmed reduction (H2-TPR) and X-ray diffraction (XRD). Combining characterisation and catalytic results proved that the Ni species are intrinsically more active than Co species, but that incorporating Co into the catalyst formulation prevented the formation of large Ni particles and led to highly dispersed metal nanoparticles on CeO2, thus leading to the observed enhanced activity for the bimetallic system. The highest yield of lactic acid (salt) achieved in this work was 93% at 97% glycerol conversion (160 °C, 6.5 h at 20 bar N2, NaOH: glycerol = 1.5). The NiCo/CeO2 catalyst also exhibited high activity and selectivity towards the target hydrogenated products in the transfer hydrogenation reactions between glycerol and various H2 acceptors. Batch recycle experiments showed good reusability, with retention of 80% of the original activity after 5 runs.

No data was used for the research described in the article.The efficient emission control of unburned methane in power plants and vehicles that use natural gas as a potential bridge fuel in the transition toward renewable energy is of vital importance [1,2], given that this pollutant is strongly involved in the greenhouse effect. Its potent greenhouse effect is around 25 times higher than that of CO2. Presently, the most adequate approach to minimize the negative impact of the release of residual methane to the atmosphere is catalytic oxidation, which allows the direct conversion of the hydrocarbon to carbon dioxide and water. Probably there is no doubt that noble metal-based catalysts, particularly Pd catalysts, are the systems with the highest intrinsic oxidation power for this abatement strategy [3–7]. However, although large efforts are continuously being made to increase its thermal and chemical stability under operating conditions, its wide use is fundamentally penalized by economic reasons [8]. Thus, the proposal of cheaper, highly efficient, alternative catalysts is a challenge of relevant interest. Most studies have been focused on the use of transition metal oxides, namely nickel [9,10], manganese [11,12], copper [13] or iron [14]. However, it is widely accepted that spinel cobalt oxide (Co3O4) is the most attractive oxide phase for the lean methane catalytic combustion owing to the presence of variable valance states (Co3+/Co2+), its lower bonding energy of Co-O bonds and the high mobility of active oxygen species capable of activating the C-H bond [15–17]. Nevertheless, bulk cobalt oxide, as well as other bulk transition oxides, usually exhibit very poor textural and structural properties, especially when synthesized by simple methodology routes [18,19]. Thus, their good behavior is mainly assigned to their high metallic content (>70 %wt.), thereby resulting in a markedly low intrinsic activity. For this reason, several strategies have been proposed in order to enhance the performance of Co3O4-based catalysts with the ultimate goal of maximizing the population of active sites. The selection of the support for depositing the active phase is the first obvious approach to take into consideration. Furthermore, advances in the optimized design of supported catalysts are highly relevant since the final configuration of a commercial catalytic unit will be a structured catalyst operating with large gas flows [20]. These catalysts will be surely prepared by washcoating a thin catalytic layer (metal oxide/support) onto the surface a monolithic/foam substrate.In this sense, it must be stated that, owing to the high affinity of cobalt for most typical inorganic supports (γ-Al2O3, SiO2 or MgO), a certain fixation of cobalt as less active CoAl2O4, CoSiO3 or Co-Mg mixed oxides must be assumed [21,22]. This unavoidably involves the use of relatively high Co loadings (20–40 %wt.) to compensate partially the useless presence of a fraction of deposited Co. The use of alternative supports such ceria or alpha-alumina prevented this strong undesired interaction but their relatively low intrinsic surface area do not usually lead to a substantially improvement in behavior of the resultant composite catalysts [23,24]. A complementary option to adjust the amount and/or reactivity of oxygen species is the addition of a promoter that could improve the reducibility of the resultant catalyst at low temperatures. Based on its comparable ionic radius and coordination and oxidation states to cobalt, nickel is the most preferred promoter. The incorporation of nickel is mainly justified by the notable activity shown by the NiCo2O4 spinel that can be formed from the interaction between cobalt and nickel. This approach is quite interesting for methane oxidative abatement [25–28], but requires a large amount of nickel (around 50 % of the Co content for a Ni/Co molar ratio of 0.5). In addition, the synthesis of stoichiometric nickel cobaltite is largely dependent on very well controlled synthesis conditions in terms of calcination temperature and selected preparation route, usually oriented to the synthesis of the mixed oxide in its bulk form. In other words, it would be of interest to explore alternatives for taking advantage of the known beneficial effects of nickel promoter, without the need of large amounts of this additive and using a relatively simple route for obtaining an active Ni-promoted cobalt catalyst.Therefore, the objective of this work is the study of Ni/Co-Al2O3 catalysts for the oxidation of methane under conditions similar to those found in the exhaust of vehicular natural gas engines (relatively low residence times, and presence of water and carbon dioxide). Thus, for a total metal loading of 30 % by weight, the effect of the addition of 5 % and 10 %wt.Ni on cobalt catalysts with a content of 25 % and 20 %wt.Co, respectively, was investigated. These samples were prepared by sequential precipitation of cobalt and nickel, with an intermediate calcination step. Along with these catalysts, monometallic cobalt (20 %, 25 % and 30 %wt.) and nickel (30 %wt.) catalysts with a content of 30 %wt. (30Co and 30Ni, respectively) were synthesized as well for comparative purposes.All oxide catalysts were synthesized following a precipitation route over a thermally-stabilized (calcined at 850 C for 8 h) γ-Al2O3 (Saint Gobain), which selected as the support. Three cobalt oxide catalysts, namely 20Co, 25Co and 30Co samples, were prepared by precipitation of aqueous solution of cobalt nitrate hexahydrate with an adjusted concentration to obtain the desired nominal Co loading (20, 25 and 30 wt. %, respectively), at 80 °C using an aqueous solution of sodium carbonate (1.2 M) until reaching a pH of 8.5. After precipitation, the precursors were dried at 110 °C overnight.Then, the catalyst precursors were calcined at 600 C for 8 h in static air. In the case of the reference nickel catalyst (30Ni, with a nominal Ni content of 30 %wt.), the starting salt was nickel nitrate hexahydrate. This sample was also submitted to the same aforementioned thermal treatment.Two bimetallic Ni-Co catalysts were obtained by sequential precipitation using the same metallic salts and precipitating conditions (pH = 8.5, 80 ºC). Thus, nickel was added to the previously prepared 20Co and 25Co samples with a nominal content of 10 % and 5 %wt., respectively. Thus, the total metallic loading of these samples was fixed at 30 %wt. Finally, the samples were again calcined at 600 C for additional 4 h. In this way, both monometallic and bimetallic catalysts were activated under identical thermal conditions. The resulting Ni-Co catalysts were designated as 10Ni/20Co and 5Ni/25Co.The supported catalysts were characterized by a wide number of analytical techniques, including scanning electron microscopy (SEM) coupled to energy dispersive X-ray spectroscopy (EDX), scanning transmission electron microscopy - high angle annular dark field (STEM-HAADF) coupled to EDX mapping, wavelength dispersive X-ray fluorescence (WDXRF), N2 physisorption, X-Ray diffraction (XRD), Raman spectroscopy, X-Ray photoelectron spectroscopy (XPS), temperature-programmed reduction with hydrogen (H2-TPR) and temperature programmed reaction with methane (CH4-TPRe). Experimental details on each of these techniques are included in the Supplementary Material.The activity of the synthesized catalysts for the oxidation of residual methane was determined in a fixed bed reactor (Microactivity by PID Eng&Tech S.L.) between 200 and 600 °C. The reaction products were quantified with an on-line gas chromatograph (Agilent Technologies 7890 N) equipped with a thermal conductivity detector. In each reaction test one gram of catalyst (particle size 0.25–0.30 mm) diluted with 1 g of inert quartz (particle size 0.5–0.8 mm) was used. A reaction mixture of composition 1 %CH4/10 %O2/89 %N2 was used with a total flow rate of 500 mL min−1, which represents an approximate space velocity of 60,000 h−1. To ensure that the mass and heat transfer effects were not affecting the kinetic results, the inter- and intraphase concentration and temperature gradients (Table S1, Supplementary Material) were verify to be negligible according to the criteria proposed by Eurokin [29]. The absence of mass and heat transfer limitations within the reactor was evaluated not only under differential conditions (X < 20 %) but also under the least favorable conditions (450–600 ºC). Additionally, the stability of the most promising catalyst with time on stream was evaluated at constant temperature (575 °C) for a total reaction interval of 150 h under alternate dry, humid (10 %) or CO2-rich (10 %) conditions while maintaining the O2/CH4 molar ratio at 10.Prior to the discussion of the characterization results of the prepared catalysts, it is highly relevant to remark that the variety of oxide phases that can be present in Co- and/or Ni-containing gamma-alumina supported catalysts thermally activated at moderate temperatures (600 °C) is wide. In addition to the expected Co3O4 and NiO oxides, and obviously the γ-Al2O3 support, the presence of mixed spinels such as CoAl2O4 and NiAl2O4 is normally unavoidable. These new metallic oxides are formed due to the strong interaction between the Co and Ni species and the support that results in the partial insertion of Co or Ni atoms into the lattice of the gamma alumina. Moreover, it is commonly accepted that the morphology of these spinels will be essentially amorphous since its transformation into a crystalline structure requires calcination temperatures as high as 800–850 °C [30,31]. Besides, the formation of Ni/Co mixed oxides can occur. Thus, based on these considerations both monometallic (20Co, 25Co, 30Co and 30Ni) and bimetallic (5Ni/25Co and 10Ni/20Co) catalysts were thoroughly investigated by a wide number of analytical techniques including N2-physisorption, SEM coupled to EDX, XRF, STEM-HAADF coupled to EELS or EDX, Raman spectroscopy, XPS, H2-TPR and CH4-TPRe. Table 1 include the textural properties of the metal oxide catalysts. The corresponding pore size distribution are included in Fig. S1, Supplementary Material. The thermally-stabilized (calcined at 850 °C for 8 h) blank alumina support showed a surface area of around 140 m2 g-1 and a pore volume of 0.56 cm3 g−1. Its pore size distribution was bimodal with maxima located at 110 and 150 Å. After the addition of increasing amounts of cobalt (20Co, 25Co and 30Co samples), the surface area appreciably decreased to 120–108 m2 g−1 due to pore blocking. Accordingly, their pore volume was notably affected since it decreased to 0.35–0.29 cm3 g−1. The resultant narrower average pore size was in the 94–98 Å range. It was then evident that cobalt species preferentially deposited on the larger pores of the support (150 Å). In the case of the nickel catalyst (30Ni sample), the addition of the metal affected the textural properties to lesser extent when compared with its cobalt counterpart with the same loading (30Co sample). Thus, a surface area close to 130 m2 g−1 was observed. This was probably connected to a trade-off effect between the pore blocking of the support by nickel and the newly formed NiAl2O4 phase with a high intrinsic surface area. This rationale was supported by the notable surface area (170 m2 g−1) of an as-prepared bulk NiAl2O4, which was prepared by precipitation and calcined at 600 °C.Regarding the bimetallic catalysts, the addition of nickel (5–10 %wt.) to the Co/Al2O3 samples produced a loss of specific surface area0 around 7–8 % with respect to the corresponding monometallic sample with the same Co content (25Co and 20Co samples). Furthermore, while the pore volume remained almost constant, a slight increase in the mean pore size (from 94 to 98–107 Å) was found. In view of these results, it could be concluded that the deposition of the promoter had no marked effect on the textural properties, as relatively similar surface areas, pore volumes, and pore diameters were obtained independently from the Ni/Co ratio of the bimetallic samples.The microstructural morphology of the four monometallic samples (20Co, 25Co, 30Co and 30Ni) and the two bimetallic samples (5Ni/25Co and 10Ni/20Co) was examined by SEM. Irrespective of the composition of the catalysts, the micrographs (Fig. S2, Supplementary Material) revealed a heterogeneous surface on which irregular particles with sizes ranging from 5 to 20 µm are arranged with an aggregated morphology. Elemental identification and quantitative compositional information could be obtained by an energy dispersive X-Ray analyzer. Thus, the average surface composition of various defined regions (40 ×40 µm with a sampling depth of about 1 µm) for each catalyst was determined. Table 2 compares the bulk and surface composition as analyzed by XRF and EDX. As for the monometallic samples, an expected surface enrichment was found as revealed by their comparatively higher metal (Co or Ni)/Al molar ratios in relation to the respective bulk molar ratios. Particularly, this ratio at the surface as determined by EDX increased by a factor of 1.6–2.1 in the case of the Co-containing catalysts, and a factor of 1.1 in the case of the 30Ni sample.The monometallic 30Co and 30 Ni samples were also examined by scanning transmission electron microscopy–high-angle annular dark field (STEM–HAADF). Additionally, EELS elemental maps (Fig. S3, Supplementary Material) were obtained for certain regions in each sample to examine the spatial distribution of these metals in the catalysts. It was revealed that both Co and Ni were homogeneously distributed over the surface and no large uncoated support regions were apparently observed. This suggested a relative good metallic coverage of the alumina surface. The samples were characterized by the presence of polycrystallites (in some cases formed by the apparent attachment of smaller crystallites) with sizes ranging from 10 to 40 nm. It is worth pointing out the detection of crystalline phases on the surface of the 30Ni catalyst was comparatively less frequent, thereby suggesting the deposited metallic species on this sample exhibited a more amorphous nature.As for the bimetallic Ni-Co catalysts, it must be pointed that, although the Ni/Co molar ratio at the surface was higher than the corresponding bulk ratio, this increase was not very marked, from 0.23 to 0.25 over the 5Ni/25Co sample and from 0.58 to 0.60 over the 10Ni/20Co sample. This suggested a partial Ni diffusion into the cobalt catalytic layer. Likewise, surface chemical mapping, in this case carried out by STEM-HAADF coupled to EDX, was carried out to study the distribution of both metals on the surface of the bimetallic catalysts. As seen in the compositional maps included in Figs. 1 and 2, both cobalt and nickel were relatively well dispersed over the surface, with no visible clustering or agglomeration of either metal. Seemingly, the mixing between cobalt and nickel seemed to be equally intimate for both Ni-Co catalysts.X-ray diffraction analysis was used to identify the crystalline phases present in each oxide catalyst. The corresponding patterns are included in Fig. 3. The monometallic cobalt catalysts (20Co, 25Co and 30Co) showed the characteristic signals of a cubic spinelic phase (2θ = 19.2, 31.4, 37.1, 45.1, 59.6 and 65.5°) that would be in agreement with the formation of Co3O4 (ICDD 00–042–1467) and/or CoAl2O4 (ICDD 00–044–0160). Certainly, as will be evidenced later by both Raman spectroscopy and H2-TPR analysis, these samples consisted of a mixture of these cobalt oxides. However, while assuming the present cobalt aluminate will be preferentially amorphous under mild calcination at 600 °C, the visible diffraction signals in these patterns could be exclusively assigned to highly crystalline Co3O4. On the other hand, the reference 30Ni catalyst evidenced the typical signals of a cubic phase at 2θ = 37.2; 43.1; 62.9 and 75.4° corresponding to the presence of nickel oxide NiO (ICDD 00–089–7131). It must pointed out that although the formation of nickel aluminate is highly likely, this could not be detected probably due to its poor crystallinity. Recall that no clear signals attributable to crystalline NiAl2O4 (ICDD 00–078–1601) were observed. However, and similar to the results found for the Co catalysts, the existence of this spinel will be verified by redox and structural studies. Finally, a weak signal attributable to γ-alumina (ICDD 01–074–2206) support was also observed at 2θ = 67.2° over these four monometallic samples. The diffractograms of the bimetallic Ni-Co catalysts did not reveal the presence of segregated NiO, which suggested that this oxide was finely dispersed on the Co/Al2O3 matrix. Thus, only diffraction signals related to Co3O4 were noted. The crystallite size of this oxide (Table 1) was determined from the full width half maximum of the characteristic signal at 37.1°, using the Bragg equation. It thus ranged from 19 to 21 nm for the 20Co and 25Co samples to 35 nm for the 30Co catalyst. Interestingly, the addition of nickel to the Co/Al2O3 samples did not significantly alter the crystallite size (17–19 nm). On the other hand, the crystallite size of the NiO phase in the 30Ni catalyst was 14 nm.The examination of the structure of the oxide catalysts was carried out by Raman spectroscopy ( Fig. 4). The Raman spectra of the 20Co, 25Co and 30Co samples displayed the five typical vibration modes of Co3O4 at 196, 480, 520, 619 and 687 cm−1 [32]. The presence of CoAl2O4 was also evidenced by the two shoulders located at 706 and 725 cm−1 [33]. Apparently the contribution of these two additional signals was more marked in the case of the 20Co and 25Co, thus suggesting that the formation of cobalt aluminate would be favored with lower loadings of cobalt. Therefore, the cobalt phases present in the studied Co/Al2O3 catalysts would be a mixture of Co3O4 and CoAl2O4, with a higher relative abundance of the aluminate phase when the total Co content of the sample was lower. Lastly, the Raman spectra of the 30Ni catalyst was dominated by a wide signal located at 545 cm−1, which would be coherent with the presence of a mixture of NiO (its main Raman mode is located at 510 cm−1) and NiAl2O4 (its main Raman mode is located at 574 cm−1). The existence of the nickel spinel was further evidenced by the weaker signals at 746 cm−1 and 835 cm−1 [34]. On the other hand, the addition of nickel to the Co-Al2O3 catalysts did not substantially modify the spectra of the resulting samples (5Ni/25Co and 10Ni/20Co). Thus, the only observable Raman modes coincided with those corresponding to the parent cobalt catalyst, namely a mixed contribution of Co3O4 and CoAl2O4 phases. The marked presence of NiO and/or NiAl2O4 could be ruled out in these bimetallic catalysts.The surface composition of the samples and, more importantly, the distribution of the various metallic (cobalt and nickel) and oxygen species was investigated by analyzing the Co2p3/2 (777–792 eV), Ni2p3/2 (850–870 eV) and O1s (526–538 eV) XPS spectra of the samples, as shown in Fig. 5. Prior to the analysis in the XPS chamber, the as-calcined oxide catalysts were stored in airtight polyethylene containers in order to limit their exposure to ambient air. The Co2p3/2 spectra were deconvoluted into three main and two satellite contributions. The main contributions were located at 779.5, 780.7 and 782.4 eV, and were tentatively attributed to the presence of Co3+(Co3O4), Co2+(Co3O4 and/or CoAl2O4) and Co2+(CoO) species, respectively [35]. For all oxide catalysts, the relative abundance of the signal related to CoO was lower than 10 % of the total Co2p3/2 signal. This species was assumed to be formed by reduction under the vacuum conditions in the XPS chamber. The two signals located at 785.5 and 789.5 eV were assigned to the shake-up satellite peaks from Co2+ and Co3+ ions.Following a similar procedure, the Ni2p3/2 spectra were deconvoluted into five signals. The three main signals were centered at around 853.9, 855.4 and 856.9 eV and were associated with the presence of Ni2+(NiO), Ni2+(nickel belonging to a spinelic phase) and Ni3+(Ni2O3) species, respectively [36]. The satellite contribution of the spectra was dominated by an intense signal located at 861.0 eV, characteristic of the presence of Ni2+, and a small shoulder at 865.3 eV, which was a consequence of the relatively reduced presence of Ni3+ ions in these samples. Finally, the O1s spectra of the samples was characterized by three signals located at 529.3, 531.3 and 532.6 eV, which were attributed to oxygen species from the crystalline lattice (Olatt), superficially adsorbed oxygen species (Oads), and carbonate and hydroxyl species, respectively [37]. From the quantification of the aforementioned spectra, the elemental surface composition and the distribution of ionic species could be determined, as summarized in Table 2.As for the monometallic cobalt catalysts, it was observed that the Co3+/Co2+ molar ratio was in the 0.60–0.69 range, which was markedly lower than that expected for the exclusive presence of Co3O4. These moderate ratios suggested the presence of Co2+-rich oxides such as CoAl2O4, as previously pointed out by Raman spectroscopy. In fact, it could be inferred that cobalt aluminate was preferentially formed for low Co loadings, since the 20Co sample showed the lowest Co3+/Co2+ ratio (0.60). On the other hand, the Ni2p3/2 spectrum of the pure nickel catalyst (30Ni) clearly evidenced the presence of comparable amounts of nickel oxide and nickel aluminate. Hence, in addition to traces of Ni3+ species, the observed nickel was in the form of NiO (36 %) and Ni2AlO4 (49 %). The incorporation of nickel markedly affected the distribution of cobalt species on the 5Ni/25Co and 10Ni/20Co catalysts. Interestingly, the addition of this promoter favored the presence of Co3+ cations. This enrichment was a priori related to the partial insertion of Ni2+ ions into the structure of the Co3O4, which would imply the generation of the mixed NiCo2O4 spinel to some extent. In this sense, since the increased population of Co3+ ions was more noticeable for the 5Ni/25Co catalyst (with a Co3+/Co2+ molar ratio of 0.80) in comparison with the 10Ni/20Co sample (with a Co3+/Co2+ molar ratio of 0.64), a more extensive formation of nickel cobaltite was likely for low concentrations of the promoter. Accordingly, these samples showed a high population of Ni2+ species related a spinel-like phase (NiCo2O4) at the cost of Ni2+ as NiO. On the other hand, the presence at the surface of Ni3+ species (as Ni2O3) was also observed over the bimetallic Ni-Co and 30Ni samples, which was favored for low Ni loadings. Finally, it must be remarked that all these structural changes induced by nickel addition on the surface of the alumina supported cobalt catalysts led to an increase of lattice oxygen species. These are widely accepted to play a key role in methane oxidation [38]. Hence, the Olatt/Oads molar ratios were between 0.50 (10Ni/20Co) and 0.63 (5Ni/25Co), apparently higher than those of the respective Ni-free counterparts (0.25 for 20Co and 0.43 for 25Co).The analysis of the metallic catalysts by temperature-programmed reduction with hydrogen (H2-TPR) could be also helpful in identifying the nature of the oxide species present in each sample. The corresponding profiles are compared in Fig. 6. The redox behavior of the monometallic catalysts (20Co, 25Co, 30Co and 30Ni) was initially discussed in order to facilitate the subsequent interpretation of the results corresponding to the bimetallic Ni-Co samples. As for the Co-containing catalysts, two reduction events were clearly observable. Above 800 °C no measurable H2 consumption was noticed. Thus, the low-temperature uptake at 250–500 °C was assigned to the reduction of free Co3O4, according to the two-stage Co3+ → Co2+ → Co0 process [39]. The stoichiometric H2:Co molar ratio of this step is 1.33. The high-temperature consumption in the 550–750 °C corresponded to the reduction of the present cobalt aluminate [40]. The H2:Co stoichiometry for full reduction of this oxide is 1. Note that the presence of this highly stable oxide was in agreement with the results derived from both Raman and XPS spectroscopies. Table 3 includes the total H2 uptake of each monometallic sample, which increased from 3.5 to 5.2 mmol H2 g−1. The comparison of these values with the theoretical consumption expected when assuming that all cobalt was exclusively present as Co3O4 (which would vary from 4.2 to 6.1 mmol H2 g−1) resulted in reducibility degrees around 84–85 %. From these values, the relative distribution of Co atoms as Co3O4 or CoAl2O4 could be estimated. Hence, the abundance of cobalt as cobalt oxide gradually increased with the Co loading from 35 % to 37 % and 39 % over the 20Co, 25Co and 30Co samples, respectively.On the other hand, the fixation of the deposited metal as aluminate due to the strong metal-support interaction was observed for the 30Ni nickel catalyst as well. Therefore, its reduction trace also revealed two distinct H2 uptakes at moderate (400 °C) and high (700 °C) temperatures, which were associated with the presence of NiO and NiAl2O4, respectively [41], in consonance with the Raman and XPS results. It is worth pointing out that the stability of nickel aluminate was significantly higher than that of cobalt aluminate since its full reduction needed temperatures higher than 800 °C. Both oxides present a H2:Ni stoichiometry of 1. A relative good agreement was found between the experimental (4.5 mmol H2 g−1) and theoretical (4.6 mmol H2 g−1) consumptions. Consequently, a reducibility close to 100 % was evidenced. An estimation of the relative contribution of each nickel species suggested a roughly similar population of both oxide phases (43 %Ni as NiO and 57 %Ni as NiAl2O4).The incorporation of nickel to the 20Co and 25Co samples did not significantly altered the shape of the corresponding redox patterns since these also showed two reduction uptakes at low (250–475 °C) and high temperatures (550–750 °C). In view of the reduction pattern of the 30Ni sample, it was reasonable to expect that the reduction of the Ni2+ species present in the bimetallic samples (preferentially as free NiO) would occur mainly at the low temperature window, thus simultaneously coinciding with the reduction of Co3O4 species. Likewise, a small uptake at around 170 °C, which was not observed in the monometallic Co-Al2O3 counterparts, was visible. This consumption was assigned to the reduction of finely dispersed NiO nickel species [42,43], and was comparatively more noticeable for the 5Ni/25Co catalyst. In addition, the Ni-Co samples exhibited an appreciable shoulder at around 800 °C that was related to the reduction of nickel aluminate, probably due the strong interaction of added nickel with trace amounts of uncovered alumina.As shown in Table 3, owing to their higher total metallic loading the quantitative analysis of the reduction profiles expectedly evidenced a higher H2 uptake for the bimetallic samples (5Ni/25Co and 10Ni/20Co) in comparison with the respective Ni-free cobalt counterparts (25Co and 20Co, respectively). Thus, the overall reducibility increased from 4.4 to 5.2 mmol H2 g−1 in the case of 25Co and 5Ni/25Co catalysts, and from 3.5 to 5.2 mmol H2 g−1 in the case of the 20Co and 10Ni/20Co catalysts. Note that the total uptake of the Ni/Co samples (5.2 mmol H2 g−1) was virtually identical to that of the 30Co catalyst (5.2 mmol H2 g−1). Also relevant was the fact the reducibility, within the experimental error, of the bimetallic samples was promoted after the addition of nickel. Thus, it increased from 84 % over the 25Co sample to 90 % over the 5Ni/25Co sample, and from 84 % over the 20Co sample to 92 % over the 10Ni/20Co sample. This suggested that the incorporation of nickel promoted the presence of Co3+ cations with a higher H2 consumption per Co (1.5). As revealed by XPS, the simultaneous presence of nickel and cobalt could result in the formation of NiCo2O4-like spinel that ultimately increased the catalyst overall reducibility. Moreover, keeping in mind that the catalytic activity in the methane oxidation is expected to be mainly controlled by oxygen species consumed in the low-temperature range, it was found that the introduction of nickel was efficient for achieving this purpose. Hence, this uptake increased from 1.2 (20Co) to 1.6 mmol H2 g−1 (10Ni/20Co), and from 1.7 (25Co) to 2.0 mmol H2 g−1 (5Ni/25Co). In this latter case, a comparable uptake was found with respect to the 30Co catalyst.The reactivity of the available oxygen species present in the synthesized catalysts was complementary investigated by monitoring the conversion of methane in the absence of oxygen at increasing temperature (CH4-TPRe). The explored temperature range was 50–600 °C with a heating ramp of 10 °C min−1. The samples were then kept at 600 °C for 15 min. The composition of the product stream was followed by mass spectrometry (m/z = 44 (CO2), 28 (CO) and 2 (H2) signals). The resulting profiles of the bimetallic Ni-Co and monometallic (30Co and 30Ni) catalysts are shown in Fig. 7. Theoretically, methane is expected to be oxidized to carbon oxides at relatively low temperatures by active oxygen species at the catalyst surface. This will result in a progressive reduction of the metallic oxides, and a concomitant high-temperature conversion of methane into reforming products including CO, H2 and CO2, and/or cracking products (H2 and carbonaceous deposits) that will be catalyzed by partially reduced or metallic cobalt and/or nickel. Following this rationale, which is schematically depicted in Fig. S4 (Supplementary Material), the most relevant findings derived by this characterization technique were essentially those corresponding to the low temperature range, at which the complete oxidation of methane would be favorably occurring.The CH4-TPRe profiles revealed the formation of substantial amounts of CO2 at two relatively well-discernible temperature windows. On the one hand, the signal detected at lower temperatures (400–450 °C) was attributed to the gradual complete oxidation of methane by oxygen species. Note that no CO or H2 were detected in this temperature range. On the other hand, when the total oxidation process was no longer possible, the progressive reduction of the catalyst by methane then activated the transformation of the feed into CO, CO2 and H2, as can be evidenced by the co-existence of these three products at higher temperatures (500–550 °C). Moreover, the XRD analysis of the spent samples evidenced the presence of metallic cobalt (ICDD 00–015–0806) and nickel (ICDD 00–001–1258), and crystalline coke (ICDD 01–075–1621) (Fig. S5, Supplementary Material).As aforementioned, only the oxygen species involved in the low-temperature CO2 formation signal will be assumed to be highly active in the catalytic combustion reaction. After a proper quantification of the amount of formed CO2, the corresponding amount of consumed oxygen species could be estimated. In this sense, the 5Ni/25Co bimetallic catalyst showed the largest consumption (0.16 mmol O2 g−1) followed by the 10Ni/20Co sample (0.09 mmol O2 g−1) and the Co and Ni monometallic catalysts (0.08 and 0.04 mmol O2 g−1, respectively). In addition, it is worth pointing out that the 5Ni/25Co the oxidation reaction also started at significantly lower temperatures (200 °C) in comparison with the other samples (300–500 °C).The efficiency in the oxidation of methane into carbon dioxide of the four samples having the same nominal metallic content (30 %wt. %), namely 5Ni/25Co, 10Ni/20Co, 30Co and 30Ni catalysts, was analyzed operating at 300 mL CH4 g−1 h−1 between 200 and 600 °C. Three consecutive light-off tests were conducted over each catalyst. After the first test, which could be understood as an equilibration step of the catalyst under reaction conditions, a certain decrease in conversion was observed. Interestingly, no significant differences in conversion were found between the second and third tests resulting in a virtually identical light-off curve. Thus, the conversion profiles shown in Fig. 8 correspond to the third catalytic reaction run. All Co-based samples exhibited 100 % CO2 selectivity in the whole temperature range. Nevertheless, substantial amounts of carbon monoxide were formed over the 30Ni sample, leading to CO2 selectivity of only 90 % even at the highest reaction temperatures (600 ºC). It was observed that bimetallic catalysts exhibited a considerably better performance compared with the monometallic samples. The T50 values, listed in Table 4, were similar for the two monometallic catalysts (550 °C) and higher than those shown by the bimetallic counterparts (535 °C for 10Ni/20Co and 525 °C for 5Ni/25Co). Table 5.The specific reaction rates, calculated using the differential method (for conversions less than 20 %) at 450 °C, revealed a higher intrinsic activity of the 5Ni/25Co catalyst (0.80 mmol CH4 h−1 g−1), compared with the monometallic 30Co and 30Ni samples (Table 4). The other investigated bimetallic sample (10Ni/20Co) showed an intermediate behavior (0.63 mmol CH4 h−1 g−1). When referred to the total metallic loading, the best intrinsic activity of the 5Ni/25Co sample was also evidenced. From the correlations depicted in Fig. 9, the observed catalytic activity trend was coherent with the abundance of Co3+ species in the samples. Thus, the 5Ni/25Co catalyst presented the highest Co3+/Co2+ molar ratio due to the more efficient insertion of Ni2+ ions in the structure of the Co3O4 spinel leading to the generation of the nickel cobaltite-like species. As shown in Fig. S6 (Supplementary Material), this dependence was also valid when referred to the reaction rate normalized per gram of metal. The excellent behavior of this mixed oxide as oxidation catalyst for a variety of hydrocarbons [44,45], carbon monoxide [46], carbonaceous particulate matter [47] and methane [48] as well has been previously reported. On the other hand, it must be pointed out that both NiAl2O4 and CoAl2O4 spinel are not particularly active for the complete oxidation of methane [49,50], owing to their relatively low reducibility and highly stable oxygen species that penalized methane oxidative conversion by the Mars–van Krevelen mechanism. Besides, their formation could be detrimental for the generation of NiCo2O4 due to the decrease in the amount of available Co3O4 and Ni for their mutual interaction. In our study, the formation of this highly active mixed spinel was apparently enhanced with adding small amounts of nickel, since a Ni content as high as 10 %wt. did not lead to a better efficiency than the 30Co catalyst. This was probably owing to the fact that the incorporated Ni was more efficiently dispersed over the 5Ni/25Co catalyst in comparison with the 10Ni/20Co counterpart, as evidenced by its lower NiO/Ni molar ratio. This favored the interaction between Co3O4 and the deposited Ni to form NiCo2O4 to a larger extent as suggested by its large amount of Co3+. This increased presence of easily reducible Co3+ was accompanied by a concomitant higher presence of active lattice oxygen species that were able to activate the oxidation of methane at relatively low temperatures. This was also evidenced by the strong dependence of the intrinsic activity with the Olatt/Oads molar ratio and the amount of consumed oxygen at low temperatures in the CH4-TPRe runs (Fig. 9). On the other hand, it was found that the intrinsic activity of Olatt species present in the 30Ni catalyst was significantly lower than that exhibited by the Olatt species in the Co-containing catalysts.The Mars-van Krevelen mechanism, also known as the redox mechanism, has been widely used for kinetics modeling of methane oxidation over metal oxides. This is based on the assumption of a constant oxygen surface concentration on the catalyst, with reaction occurring by interaction between a molecule of reactant and an oxidized portion of the catalyst. Thus, the model assumes that the oxidation of the hydrocarbon occurs in two steps. In the first step, the compound react with the lattice oxygen resulting in its reduction and the corresponding formation of oxygen vacant site. In the second step, the reduced metal oxide is reoxidized by the gas phase oxygen present in the feed. In the steady state, the rates of the reduction and oxidation steps must be equal. Then, the kinetic equation (Eq. 1) can be expressed as: (1) ( − r ) = k red k ox P CH 4 P O 2 k ox P O 2 + γ k red P CH 4 where kred is the rate constant of the oxidation of the hydrocarbon by the lattice oxygen, kox the rate constant of the lattice re-oxidation and γ is the overall stoichiometry of the reaction. For conditions with oxygen excess (in our case, a PO2/PCH4 ratio of 10 at the inlet of the reactor), the term koxPO2 is considerably larger than γkredPCH4. Consequently, the rate equation simplifies to a power rate law equation (Eq. 2). (2) ( − r ) ≅ k red P CH 4 Accordingly, the integral method was applied to estimate the apparent activation energy when assuming a first pseudo-order for methane and a zeroth pseudo-order for oxygen [38,51]. Conversions between 10 % and 90 % were fit to the following linearized equation for the integral reactor (Eq. 3) where X is the fractional conversion of methane, k0 is the pre-exponential factor of the Arrhenius equation and FCH40/W is the weight hourly space velocity. The goodness of the numerical fit is depicted in Fig. S7 (Supplementary Material). It was observed that the apparent activation energy of the 30Ni catalyst (128 kJ mol−1) was markedly higher than that of the cobalt catalysts, in line with the lower activity of this catalyst for complete oxidation. The bimetallic catalysts and the 30Co catalyst showed a relatively similar value between 90 and 103 kJ mol−1. It is worth pointing out that this range of values was appreciably higher than that found for this reaction catalyzed by bulk Co3O4 (70–75 kJ mol−1) [52–54], thereby suggesting that the intrinsic activity of the examined cobalt catalysts was negatively affected by the presence of cobalt aluminate. (3) ln − ln 1 − X = ln k 0 C CH 4 0 W F CH 4 0 − E a RT Finally, given the presence of notable amounts of water vapor and carbon dioxide in the real exhaust gases of a natural gas engine, an attempt to evaluate the stability of the most efficient catalyst, namely the 5Ni/25Co sample, with time on stream was made. Thus, the evolution of conversion at 575 °C was examined when the composition was alternated following this sequence: 1 %CH4/10 %O2/N2, 1 %CH4/10 %CO2 /10 %O2/N2, 1 %CH4/10 %O2/N2, 1%CH4/10 % H2O 10 %O2/N2, 1 %CH4/10 %O2/N2, and 1 %CH4/10 % H2O/10 %CO2/10 %O2/N2. For each composition, a reaction time interval of 25 h was analyzed, with an accumulated time on stream of 150 h ( Fig. 10). During the first 15–20 h under base conditions (absence of water and CO2) a slight decrease in conversion from 80 % to 70 % was noticed. Then this conversion was stable, and was not affected by the addition of carbon dioxide for additional 25 h. Therefore, after an initial equilibration of the catalyst under reaction conditions, a relatively good thermal stability and resistance to the presence of CO2 was evidenced (75 h time on stream). However, after the admission of water into the reactor during additional 25 h, conversion dropped to a stable value of 40 % due to water adsorption on the surface [55]. Interestingly, when water was subsequently cut off, the methane conversion was almost fully recovered (65 %) upon returning to dry conditions. Thus, it was evidenced that this temporary inhibiting effect of water did not result in a remarkable irreversible deactivation of the sample. The catalyst was submitted to a further analysis under humid conditions but combined with the addition of carbon dioxide as well (25 h). Again, a decrease in conversion to 35 % was appreciated due to competitive effects caused by water.The state of the used catalyst in this long-term run was carried out by N2 physisorption, XRD and CH4-TPRe. The textural analysis revealed a slight decrease in surface area to 101 m2 g−1 (107 m2 g−1 for the fresh counterpart), thus suggesting the sintering of the active Co3O4 phase as in parallel confirmed by XRD. It is worth highlighting that irreversible poisoning was ruled out in view of the composition of the gas flow at the reactor inlet (CH4/O2/H2O/CO2). Besides, the formation of carbonaceous deposits (coke) was not observed given the net oxidizing character of the feedstream (PO2/PCH4=10 at the inlet of the reactor) that inhibited the eventual decomposition/cracking of methane. Hence, an enlargement of the crystallite size (25 nm, 19 nm for the fresh sample) was verified. These structural changes led in turn to a poorer oxidation ability at low temperatures judging from the results by CH4-TPRe analysis (Fig. S8, Supplementary Material). A shift of around 10 °C was noted for the peak oxidation temperature, from 410 °C (fresh sample) to 420 °C (used catalyst). However, it must be pointed out that the total amount of active oxygen species was not substantially modified (0.16 mmol O2 g−1).From a structural point of view, the monometallic samples consisted of a mixture of crystalline Co3O4 and amorphous cobalt aluminate in the case of the Co-containing catalysts (20Co, 25Co and 30Co), and a mixture of crystalline NiO and nickel aluminate in the case of the 30Ni sample. The formation of these undesired Al-based spinels due to the unavoidable strong interaction between the transition metal and gamma alumina was appreciable since around 40–65 % of the deposited metal was fixed as a metal-Al mixed oxide. It is worth pointing out that the generation of these aluminates was unfavored with the metallic loading.As revealed by STEM-HAADF coupled to chemical mapping the added nickel was homogeneously deposited on the surface of the corresponding cobalt catalyst as no clusters or visible agglomerates were distinguished. Thus, a relative good dispersion of the promoter could be inferred. As a result, the overall redox properties of the bimetallic catalysts were enhanced, which was essentially attributed to the formation of a new NiCo2O4-like spinel that increased the relative population of Co3+ species in the resulting Ni-Co samples. Hence, these structural changes induced by nickel led to an increase in the amount and mobility/reactivity of lattice oxygen species at lower temperatures with respect with the reference pure Co counterparts, which eventually resulted in a higher intrinsic activity and lower ignition temperatures for methane abatement. The optimal catalyst composition, which globally enhanced the abundance of Co3+ by a proper combination of highly active Co3O4 and NiCo2O4 phases, was that of the 5Ni/25Co sample. The 10Ni/20Co and the 30Co catalysts exhibited a similar efficiency. Therefore, this study demonstrated that the synergistic effect between the two metal sites is an efficient strategy to activate lattice oxygen species, which can affect the catalytic oxidation activity significantly. Andoni Choya: Investigation, Writing - original draft. Beatriz de Rivas: Methodology, Formal analysis, Validation. Jose Ignacio Gutiérrez-Ortiz: Methodology, Formal analysis, Funding acquisition. Rubén López-Fonseca: Conceptualization, Writing - review & editing, Supervision, Funding acquisition, Project administration.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This research was funded by the Spanish Ministry of Science and Innovation (PID2019-107105RB-I00 AEI/FEDER, UE), Basque Government (IT1509-22) and the University of The Basque Country UPV/EHU (DOCREC21/23). The authors wish to thank the technical and human support provided by SGIker (UPV/EHU). In addition, authors acknowledge the use of instrumentation as well as the technical advice provided by the National Facility ELECMI ICTS, node ‘Advanced Microscopy Laboratory’ at University of Zaragoza.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jece.2022.108816. Supplementary material. .

In this work bimetallic Ni catalysts supported over Co-Al2O3 and monometallic Co-Al2O3 and Ni-Al2O3 catalysts were examined for the complete oxidation of methane. With a 30 % total metallic loading, the samples were synthesized by a sequential precipitation route. All samples were characterized by nitrogen physisorption, X-ray fluorescence, X-ray diffraction, Raman spectroscopy, scanning electron microscopy, scanning-transmission electron microscopy, X-Ray photoelectron spectroscopy, and temperature-programmed reduction with hydrogen and methane. Their catalytic performance was investigated in the temperature range of 200–600 °C with a space velocity of 60.000 h−1. The bimetallic catalysts showed a better behavior in the oxidation reaction than the monometallic counterparts, mainly due to the good dispersion of Ni on the surface of the Co-Al2O3 samples. This has enabled the insertion of Ni2+ ions into the cobalt spinel lattice, which in turn provoked an increase in the amount of Co3+ species, and a subsequent enhanced mobility of oxygen species in the spinel. In this sense, the 5Ni/25Co catalyst showed the best performance, thus reducing the value of the T50 by 25 °C with respect to the monometallic catalysts.

Brunauer–Emmett–TellerBarret–Joyner–HalendaCold gas efficiencyCassava rhizomeEnergy dispersive spectroscopyEquivalence ratioHigher heating valueInternational Union of Pure and Applied ChemistryLower heating valueMobil Composition of Matter No. 41Relative pressureScanning electron microscopeSimulated flue gasX-ray powder diffractionX-ray FluorescenceBiomass is one of the potential renewable energy resources [1]. Gasification is a promising thermochemical conversion, the main driver for converting biomass composition into useful gases and chemicals. Gaseous fuels from biomass gasification can be sources of producer gas (CO, H 2 , CH4, CO2), and syngas (CO, H 2 ). However, the inherent drawbacks of biomass are low energy density, hydrophilic materials, bulky volume and short time storage. Torrefaction is a pre-treatment process at a temperature of 200–300 °C in an inert atmosphere to increase the volumetric energy density, which can enhance the biomass conversion efficiency [2]. During torrefaction, the original component in biomass such as volatile compounds, lignocellulosic materials, inter and intra-molecular hydrogen and CO, CH bonds are destructed at different temperatures [3–5]. Generally, conventional torrefaction is carried out in a nitrogen atmosphere, which leads to a higher operating cost, stemming from the requirement of separation of N 2 from air. Oxidative torrefaction is another special torrefaction process, in which biomass is torrefied in an oxidative environment (containing 3–10 vol.% O 2 ). The study of Wang et al. [6] indicated that torrefied sawdust’s properties and its pellets in oxidative exposure, such as density and higher heating value, were close to those in inert atmospheres. From the work of Chen et al. [7], it was reported that higher heating value of liquid product derived from the torrefied palm oil fiber pellets in inert and oxidative exposure was in range of 10.10–13.20MJ/kg, which could increase to 23.20–28.70MJ/kg after dehydration. From research by Li et al. [8] which torrefied pine and poplar under CO2 carrier gas at temperature ranging from 220–340 °C. They concluded the higher temperature plays the reaction of deacetylation and dehydration while the mainly reaction decarboxylation occurs at the low temperature torrefaction. It can thus be concluded that using combustion flue gases as carrier gases for the torrefaction of biomass is feasible.Cassava rhizome (CR) biomass, agricultural residues in Thailand, can be converted into a gaseous product by gasification or pyrolysis. Many studies have evaluated CR behavior in thermo-chemical processes such as combustion, gasification, and pyrolysis. Previous studies have investigated cassava residues during later process such as fast pyrolysis of stalk and rhizome of cassava plants by a pyrolysis GC/MS [9]. There were some slow pyrolysis researches of palm kernel cake and cassava pulp residue in a fixed-bed reactor [10]. Homchat et al. [11] conducted slow pyrolysis of fresh and dried CR in a large scale metal kiln which resulted in less charcoal than fresh CR, due to the effect of the moisture content. Most of previous researches investigated the type of reactor; however, high oxygen component in CR (38-57 wt%) caused low heating value and oxygenated compound emission in the bio-oil product. In a few recent studies, it has been reported that torrefied biomass can significantly affect the efficiency of biomass gasification. Phanphanich and Mani [12] investigated the fuel characteristics and grindability of pine chips and logging residues torrefied at temperatures ranging from 225 °C to 300 °C and 30 min residence time. They found that high hemicellulose and lignin in the biomass produce more tar during the gasification. Tremel et al. [13] found that the overall gasification efficiency and carbon conversion efficiency of the entrained flow gasifier was observed to be superior for the smaller ( 160 μ m ) particles torrefied biomass compared to that of the larger ( 250 μ m ) particles.Various zeolite catalysts such as dolomite, olivine, and metal oxide have been introduced in biomass gasification or pyrolysis in order to improve the quality of the product. Several catalysts have been tested, either for coal or biomass gasification i.e., dolomite, fluid catalytic cracking catalysts (FCC) [14], and metal based catalysts [15]. Particularly, zeolites are widely applied in more than 90% of petrochemical and refining industries. During the thermo-chemical process, zeolite catalyzes to upgrade biomass (i.e., cellulose, cellobiose, D-glucose and xylitol) at moderate temperatures of 400–600 °C and enhances the yields of aromatic and aliphatic hydrocarbons. MCM-41 zeolite properties are a regular array of uniform and one-dimension mesopores. The extremely high surface area of ca. 900–1000 m2/g makes these materials promising candidates as catalysts or as catalysts support. Generally, metal such as nickel (Ni), shows excellent catalytic activity. Previous works have recorded improvement of catalytic activity and stability in steam gasification of biomass through Ni/MCM-41 [16], partial oxidation of CH4 [17], and CO 2 reforming of CH4 [18,19]. Many supporters of Ni catalyst, such as MgO, Al2O3, ZrO2, and CeO2 were tested in the activities. Moreover, porous structure of MCM-41 interaction with Ni metal are important for catalytic process during steam reforming of hydrocarbons into light products ( C 1 C 5 ) or the gasoline ( C 5 C 12 ) [16,20].MCM-41 can be synthesized from waste, such as cold fly ash or rice husk. Because abundance of silicon source composed in the solid waste. Research by Li et al. [8] found that the BET surface area and average pore diameter of MCM-41 synthesized from coal fly ash, were 1347 m2/g and 3.80 nm, respectively. Illite is a raw material in many industrial applications particularly in ceramics and refractories. Almost all the illite clay waste in Thailand was disposed of in the mining area after mining and dressing illite clay, which caused an environmental problem. To date there have been no systematic studies of the recovery of illite waste for MCM-41 synthesization. Illite waste can be major a silica (Si) source for MCM-41 zeolite synthesis. Illite waste was treated with an alkaline solution or silica, which make alumina to be the first extracted from clay with hot alkaline solution and consequently this process resulted in the supernatants. Then, the supernatants were applied as a starting material for MCM-41 zeolite synthesis by hydrothermal processing.According to, new trend of the renewable energy and zero waste and circular economy, the utilization of illite waste as the raw material for zeolite synthesis was focused on this work. The objective of this synergy study is to propose the value-added pathway on solving of the illite mining waste, flue gas emission and drawback of CR fuel. The transition metal Ni can be loaded on MCM-41 by impregnation or post-synthesis which is generally a desirable method. The focus of this work is to study MCM-41 synthesized by illite waste (Ni/MCM-41) for catalytic gasification of torrefied CR at 700 °C for 30 min for generation of high-quality gas products.Torrefaction of CR particle size of 0.425–0.850 mm and 0.850–2 mm was conducted at temperature of 260 °C for 60 min in nitrogen gas atmosphere and simulated flue gas (SFG) and used as a raw material for gasification. In case of SFG mixed, CO2 (15 vol.%) and O 2 (5 vol.%) in N 2 balance was applied in this work. The picture of the CR samples is displayed in Fig. 1. The element of initial CR before torrefaction such as carbon, hydrogen, nitrogen, and oxygen were 37.60, 5.41, 0.37, and 55.93, respectively. The properties of CR sample are listed in Table 1. The chemical compositions and phase analysis of illite waste were characterized by X-ray Fluorescence (XRF) and X-ray powder diffraction (XRD) techniques. The elements that are found in the highest quantities are O, Si, Al, Fe, K, and Na. These are also the major elements found in illite waste. The chemical composition of illite sample mainly consisted of 73.01wt.% SiO2, 16.52 wt% Al2O3, 5.28 wt% K 2 O and 2.38wt.% Fe2O3 and low contents of MgO, TiO2, Na2O, and SO3. Phase analysis of illite powders was determined by XRD (PANalytical, model X’ Pert Pro) with 40 kV, Cu K α radiation. The scanning ranges from 10-60° with a step size of 0.02 are shown in Fig. 2(a). Microstructure of illite sample was measured by scanning electron microscope (SEM) (Hitachi, model SU-5000) as shown in Fig. 2(b). Illite was fused with NaOH with the weight ratio of NaOH-Illite at 1.2:1. Calcined temperature of sample was 550 °C for 1 h. The illite fusion was slowly dissolved in 22.50 ml of deionized water for 24 h. The supernatant was obtained after filtration of suspension. 3.45 g of cetyltrimethylammonium bromide (99%, Aldrich Chem Co) was dissolved in 45 ml of deionized water (D.I.), mixed with 5.4 ml of ammonium hydroxide and continuously stirred at 25 °C for 30 min. After addition of 8.33–13 ml of tetraethyl orthosilicate (98%, Aldrich Chem Co) stirring continued until a homogeneous mixture emerged, with pH = 10.5–11.5 adjustment by acetic acid. The mixed liquid was transferred to a Teflon-lined stainless-steel autoclave and heated at 110 °C for 72 h. The precipitated powder of MCM-41 was filtered and washed with deionized water. MCM-41 was dried at 105 °C in oven and then calcined in air at 540 °C for 4 h. 5Ni/MCM-41 catalyst was prepared by impregnation and evaporation. A certain amount nickel (II) nitrate hexahydrate (Ni (NO3)2 ⋅ 6H2O, 98.5%, Aldrich Chem Co) loading (5 wt%) was dissolved in ethanol. MCM-41 powder was added to the mixture and stirred for 3 h followed by evaporation of the mixture at 50 °C. The solids obtained were calcined in a muffle furnace at 550 °C for 4 h with a heating rate of 1 °C/min in the presence of air.A downflow gasifier system consists of five main parts: (1) biomass feeder (2) carrier gas unit, (3) stainless steel reactor and catalyst holder, (4) condenser, and (5) gas filters and collection unit. Gasification zone and catalytic section temperatures were set at 700 °C, and 500 °C, respectively. The 5Ni/MCM-41 catalyst was mixed with silicon carbide (SiC) in the ratio of 1:55 and placed in a top holder section. Three thermocouples were fitted at the top and bottom part of the reactor, and at catalyst holder. The reactor was purged with N 2 to avoid combustion before operating. The carrier gas ( N 2 ) entered the reactor along with the gasifying agent ( O 2 ) which was fed into the bottom of the reactor. The ratio of N 2 and O 2 was adjusted to a target equivalence ratio (ER) of 0.4. The CR was continuously fed into the system of 1.0 g/min for a 30 min. The condensate products, such as tar were retained in condensers and gas washers. Gas produced was measured by means of volumetric gas meter after separation of condensate before conveyed into the main gas line by a vacuum pump at a flow rate of 0.5 L/h. The gaseous products such as CO2, CO, H2, CH4 and other hydrocarbons as C x H y were measured by Gasboard-3100p instrument. The solid portion was later collected for further analysis.X-ray diffraction pattern of synthesized catalyst was derived by XRD (Rigaku TTRAX III) with low angle range of 0.5–5° and XRD (PANalytical, X’ Pert Pro) for wide range of 20–70°. Ni metal dispersion of catalysts was analyzed by Energy dispersive X-ray spectroscope (EDS). Surface area, average pore size, and total pore volume of the fresh catalysts were determined by N 2 adsorption and desorption Brunauer–Emmett–Teller (BET) isotherms. Pore size, pore size distribution and pore volume were obtained by the Barret–Joyner–Halenda (BJH) pore analysis.As expected, at any torrefaction conditions, the oxygen component was decreased in the sample while more carbon was retained in the torrefied CR. For torrefaction in N 2 atmosphere, the carbon content for CR size 0.425–0.850 mm and 0.850–2 mm were 53.40 wt% and 51.49 wt%, respectively. Carbon contents in torrefied CR under SFG atmosphere drastically improved when compared to the original CR as noticed in Table 1. This effect of SFG is prominent for carbon and oxygen components in all torrefied CR particle sizes. The decrease of hydrogen and oxygen in torrefaction of CR process because of the breakage of OC and CC bonds [20]. After undergoing SFG torrefaction, OC atomic ratio (O/C) and HC atomic ratio (H/C) were reduced to become less than those under N 2 carrier gas. Proximate analysis of these samples was shown in Table 1, the loss of some of the organics affected the loss of the volatiles while ash content increased. Additionally, the heating value of CR was originally 15.6–15.9 MJ/kg and after torrefaction with N 2 and SFG, heating value was increased to 20.20–22.07 MJ/kg and 22.07–24.37 MJ/kg, respectively. These results can be expressed by the low-energy bond of HC and OC reduction and high energy bond of CC. Nitrogen and ash content increased in torrefied CR. This is simply attributed to the fact that all the components containing nitrogen and other minerals (in ash) retain in the biomass solid phase, whereas C, H, and O leave the solid.The XRD pattern of MCM-41 after calcination is illustrated in Fig. 3(a). The MCM-41 catalyst samples display the high intensity peak in 2 θ of 2.16°, 3.74°, 4.30°, and 5.72° and sharp diffraction peak ( d 100 ), ( d 110 ), ( d 200 ) and ( d 210 ), respectively. These diffraction peaks indicate a long-range ordered hexagonal mesoporous structure of MCM-41 synthesis from illite waste. The hexagonal diameter and pore wall thickness can be calculated by equation of a 0 = 2d 100 ∕ 3  [21]. The a 0 of MCM-41 was 45.71 Å.The XRD pattern of fresh 5Ni/MCM-41 is exhibited in Fig. 3(b). The diffraction peaks of Ni phase at 42° and 50° occurred at the NiO crystalline phase at 37°, 43° and 51°. The other crystalline such as Ni2SiO4 (nickel silicate) and Al2SiO5 (aluminum silicate) were observed at 26° and 61°, respectively. It can be concluded that Ni atoms displayed good dispersion in the support porous. The nitrogen adsorption–desorption isotherm and pore size distributions of catalyst sample are illustrated in Fig. 4(a) and 4(b). All the samples are isotherm type IV according to IUPAC classification. In Fig. 4, the immediate increase in the region of 0.25 < P/P0 < 0.40 is related to the capillary condensation inside the mesoporous wall [17,19]. In general, the long range at higher relative pressure suggested that the adsorption continued on the surfaces of MCM-41 sample at P/P0 > 0.45 due to an increase in pore size. The isotherm of MCM-41 sample presented mesoporous filling steps with pore size larger than 40 Å. In Fig. 4b, the isotherm shows an identical shape, although the adsorption capacity decreased with Ni loading on MCM-41 support because the particles of nickel oxide finely dispersed inside the MCM-41 supported porous by impregnation with ethanol. The textural properties of synthesized catalysts are presented in Table 2. The surface area of 5Ni/MCM-41 catalysts was slightly decreased from 804.03 m2/g to 737.88 m2/g. The 5Ni/MCM-41 pore diameter was close to MCM-41 supported sample, because of the dispersion of Ni metal particles in MCM-41 and less blockage of MCM-41 supported pore by the impregnation method [22]. It can be noted that the porosity of the catalyst is not significantly changed in this work. SEM technique was applied for analysis of the surface topology and to assess the dispersion of Ni components covered on supporter. SEM image of illite waste, MCM-41 support and 5Ni/MCM-41 can be seen in Fig. 5. All the samples contained irregular shapes and grain sizes [23]. The surface morphology of MCM-41 synthesized at pH 10.5–11.5 gave well-order with diameter size around 100 nm. The small amount of Ni particles can be observed on MCM-41 surface, whereas some Ni particles were found inside the MCM-41 pores [24,25]. Ni metal dispersion was obtained for 3.6 and 5.6 ± 0.2 wt% by EDS for analysis. In this work, the torrefied CR was used in catalytic gasification. The presence of nickel enhanced the gas fraction of product yield. Liquid yield from CR size 0.425–0.850 mm was lower than the larger size in gasification with a catalyst. Carbon and hydrogen conversion were calculated as the molar of CO and H 2 produced. The syngas composition from no catalyst gasification experiment consisted of CO (3.32–7.90 vol.%), H 2 (2.48–3.02 vol.%), CH4 (2.23–3.29 vol.%), C x H y (0.07–0.16 vol.%) and CO2 (9.30–16.52 vol.%). Torrefied CR had lower O/C ratio, and when it was gasified, the torrefied CR produced lower CO2. Gasification with 5Ni/MCM-41 showed higher H 2 and CO for torrefied CR with lower CO 2 concentration, in tune with the findings of Tapasavi et al. [26]. Tar was effectively removed by 5Ni/MCM-41 catalyst. Typically, Ni based catalysts exhibit high tar cracking. This metal property along with the ongoing Boudouard and water gas shift reaction activities allow favorable composition adjustment of H 2 and CO more than that of CO2 in the product gas. The CO and H 2 concentration obviously increased with the use of a catalyst, as can be seen in Table 3. In the downflow reactor, the gasification of torrefied CR reactions can be explained the following main equations (Eqs. (1)–(6)). Devolatilization in gasification of torrefied CR occurs more than others since volatiles react with themselves in gas–gas phase. In addition, the water-gas shift reaction can increase the CO2 amount in syngas. These are explained in gas-phase phenomena of volatile gases released during gasification. In gas phase reactions, the water gas shift reaction is important for increasing the H 2 in syngas, while the methanation influences the CH4 product. The Boudouard reaction converts CO2 into CO. Because temperatures are below 1000 °C, this reaction is in equilibrium and the CO remains in the synthesized gas. (1) Boudouard reaction  C + CO 2 → 2 CO (2) Water-gas shift reaction  CO 2 + H 2 ↔ CO + H 2 O (3) Methanation reaction  C + 2 H 2 → CH 4 (4) C + 3 H 2 → CH 4 + H 2 O The nitrogen compounds contained in materials enhanced the direct release of isocyanic acid as part of volatile matter during degradation of CR at low temperature in downflow reactor. HNCO can react with steam from water gas shift reaction, yielding ammonia (NH3) and CO2. The moisture in CR could participate in the reaction involving hydrogen cyanide (HCN) and give rise to NH3 and CO as per equation below [27]. This results in a comparatively steady amount of CO. (5) HNCO + H 2 O → NH 3 + CO 2 (6) HCN + H 2 O → NH 3 + CO The effect of 5Ni/MCM-41 catalyst on torrefied CR gasification was studied. The catalyst holder was placed on the top of reactor. The operating temperature of catalyst was set at 500 °C, which is generally the range for tar cracking in the gasification, similar to the use of Ni catalyst in gasification temperature of 500 °C and 600 °C of cedar wood and sunflower stalk, respectively [16,25]. Surface acidity active sites of 5Ni/MCM-41 can improved the cracking or reforming reaction. In this work, the gas production and carbon conversion slightly increased from 66 to 73 wt% and 75 to 80%, respectively. H conversion obviously increased from 18.47 to 27.39% as illustrated in Fig. 6. Having surface acidity active sites on 5Ni/MCM-41 can improve the cracking or reforming reactions. All researches reported over 60% conversion of reforming biomass tar over Ni-based catalysts [28–30]. Moreover, synthesized 5Ni/MCM-41 can crack the vapor fraction in the system which increase the product gas volume due to the decomposition of gaseous components of the synthesized gas obtained during cracking of liquid products [31]. 5Ni/MCM-41 will drive Sabatier reaction forward and yield more CH4 as described in equation below [32]. (7) CO 2 + 4 H 2 → CH 4 + 2 H 2 O Combined with the characterization results, the trend of the catalytic performance of the active Ni surface areas and dispersion can be clearly observed. Ni/MCM-41 enhanced the higher H2 production from CR gasification, suggesting that some small Ni particle sized of 3 nm maybe inside the MCM-41 porous which promoted water gas shift and reforming reactions of C x H y and CH4. This is because of the reactants’ longer residence time inside of the MCM-41 pores [20].Cold gas efficiency (CGE) is the fraction of energy output over the energy input. The heating value of the gas products obtained from the gasification of torrefied CR is presented in Fig. 7. A lower O/C ratio in torrefied CR indicate the increasing amount of C and the calorific value which implies higher gasification efficiency. The ranges of CGE varies with heating value of gas products. The minimum of gas heating value (lower heating value) was 7.78–8.37 MJ/kg without catalyst gasification, while the maximum range was 9.38–10.03 MJ/kg. These results indicated a high CO, H 2 and hydrocarbon gaseous mixture in the gas product. Gasification efficiency in gasification without catalyst has some issues worth mentioning as per following. Utilizing CR size 0.425–0.850 mm will yield slightly more CGE than larger CR while smaller CR obtaining from torrefied under SFG will yield considerably 10% less than larger size. CGE is fundamentally calculated from heating value of produced gas and CR biomass sample, yet there is another parameter namely gas-volumetric which induced CGE to become disproportional and not corresponding with other results. For example, gasification of CR size 0.850–2 mm dropped for 3%, which is statistically insignificant. It can be concluded that addition of Ni/MCM-41 enhanced overall efficiency. Catalytic gasification of CR size 0.85–2 mm obtaining from torrefied under SFG showed outstanding result in term of CH conversion and H 2 /CO data.Gasification of torrefied CR with 5Ni/MCM-41 catalyst was investigated in this work under partial oxidative atmosphere such as simulated flue gas from typical power plant. Results from the experiments confirmed the benefits of torrefaction of CR prior to gasification under described atmosphere. Illite waste can be utilized as a precursor to MCM-41 synthesis and added as a catalyst in gasification of torrefied CR. This work has revealed the synergy of utilizing torrefaction and catalyst from waste for energy generation considering from its efficiency in various factors such as gasification of torrefied CR under SFG atmosphere and N 2 with catalyst would yield 10% less unfavorable liquid. MCM-41 supported Ni metal was prepared by ethanol assisted impregnation method. NiO dispersed throughout on MCM-41 supported and partially formed into Ni2SiO4 which showed high surface area and pore volume. 5Ni/MCM-41 plays an important role on total gas yields and CO and H 2 conversions with its excellent properties in gasification and tar cracking. In summary, applying SFG for CR torrefaction is a promising technique to produce quality green fuel at economical cost for further utilization as renewable energy via catalytic gasification process employing illite waste as precursor for synthesizing effective catalyst.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by Royal Golden Jubilee Ph.D. Programme, Thailand [grant no. PHD/0212/2557]. Additional research scholarships were provided by Overseas Academic Presentation Scholarship for Graduate Students, Thailand, and the 90th Anniversary of Chulalongkorn University Fund (Ratchadaphiseksomphot Endowment Fund), Thailand . The authors would like to thank the National Metal and Materials Technology Center (MTEC) and Interdisciplinary Program in Environmental Science, Graduate School, Chulalongkorn University.

In this work, torrefaction of cassava rhizome (CR) under nitrogen gas ( N 2 ) and a simulated flue gas (CO2 (15 vol.%) and O 2 (5 vol.%) in N 2 balance) atmosphere was examined in a downflow reactor at 260ºC for a residence time of 60 min to produce a superior solid fuel for subsequent 5Ni/MCM-41 catalytic gasification of CR utilization. Mesoporous molecular sieves (MCM-41 zeolite) was synthesized from illite waste as a silica source. The MCM-41 synthesis was carried out by hydrothermal and post-synthesis for Ni loading. Various characterization techniques, such as XRD, SEM, and BET were employed to thoroughly characterize catalyst. High surface area (737.88 m2/g) and a typical type IV pattern of hysteresis loop (0.25 –0.40) obtained 5Ni/MCM-41 catalyst is calculated by N 2 adsorption–desorption technique. Catalyst characterization and discussion of results are presented in this work. 5Ni/MCM-41 catalyst strongly enhances the H 2 and CO production from gasification of torrefied CR at a temperature of 700ºC. Carbon and hydrogen conversions were 80.17% and 27.39%, respectively while liquid yield was lower than 10 wt%. The syngas from the conversion maintained H 2 /CO ratio of 0.55 with the highest gaseous efficiency of 49.35%. Obviously, synergy of synthesized 5Ni/MCM-41 catalyst and torrefied CR with gasification is valuable useful as potential renewable energy generation process.

Utilization of CO2 is currently a hot topic in catalysis due to the chance to decrease anthropogenic CO2 emissions on the one hand and to recycle it as a C1 source in exchange to fossil fuels on the other hand. So called power-to-gas (PtG) and power-to-liquid (PtL) technologies enable chemical storage of surplus energy from regenerative sources by reaction of renewable H2 with CO2 to energy carrier such as methane (PtG) or liquid fuels (PtL) [1–3]. Especially the PtG technology has high potential as a chemical energy storage technology since infrastructure for fast energy generation as well as a natural gas grid based on fossil natural gas is already well established and a state-of-the-art-technology. Hence, renewably produced CH4 via PtG can be easily feed into the existing gas grid and in a future perspective completely replace fossil natural gas.Ni is the state of the art catalyst for CO2 methanation (Eq. (1)) already since its discovery by Paul Sabatier in 1902 [4] and has been center of several studies on various supports, whereat reviews can be found elsewhere [1–3,5–14]. (1) CO2 + 4 H2 → CH4 + 2 H2O Besides Ni, also other metals are active in CO2 methanation [15,16]. Mills and Steffgen classified the important metals for methanation catalysts by its activity (Ru > Fe > Ni > Co > Mo) and selectivity to methane (Ni > Co > Fe > Ru) [17].Ni shows high activity with a very good selectivity to CH4. Nevertheless, traditional Ni-catalysts suffer from deactivation by sintering of the Ni particles upon heat evolution from the highly exothermic methanation reaction [18]. Deposition of coke and formation of volatile nickel carbonyls contribute to additional catalyst deactivation [19,20]. Besides, Ni is of toxicological concern. The sequences of Mills and Steffgen point out, that Fe has a very high activity for CO2 activation but suffers from low selectivity. In contrast to Ni, iron is not toxic, is much more abundant and hence around 180 times cheaper than nickel.Surprisingly, only a few studies focus on optimization of Fe based catalysts for CO2 methanation. Kirchner et al. investigated bare iron oxide samples in the CO2 methanation and obtained best activity for nano-sized γ-Fe2O3 with maximum CH4 yield of 60 % at 400 °C and ambient pressure [21]. In addition, pure α-Fe2O3 based catalysts can be promoted with 2 wt % Mg in order to increase the basicity and hence interaction of CO2 with the catalysts. This promotion leads to improved CH4 yield up to 32 % at 8 bar and a GHSV of 10,000 h−1 [22]. The results emphasize that the methanation takes place predominantly on surface carbon and iron carbide species on promoted bulk Fe2O3 catalysts [22]. In general, the high activity of Fe for CO2 activation results from the high reverse-water-gas-shift (RWGS) activity (Eq. (2)) and especially at elevated pressure its further capability of CO hydrogenation via Fischer–Tropsch-Reaction (FTR) (Eq. (3)). (2) CO2 + H2 → CO + H2O (3) nCO + m/2n + 1 H2 → 1/n CnHm + n H2O Lee et al. investigated the CO2 hydrogenation via FTR on Fe catalysts at 1–25 bar and in various H2/CO2 ratios [23]. They found that metallic Fe transforms into mixtures of magnetite and carbides under reaction conditions. Especially in the pressure range of 1–10 bar the increase of pressure leads to an increase of the chain length and higher temperature increases the CO2 conversion as well as CO and CH4 yield. In contrast, the produced H2O from FTR contributes to the equilibrium of the RWGS-reaction that limits the CO2 conversion [23]. In line, on K and S promoted Fe-based catalysts it was shown that the CO2 methanation activity is strongly influenced by the H2/H2O ratio effluent from the reactor [24,25]. It was claimed that conversions increase with increasing H2/CO2 ratio and cannot be further improved than their maximum CO2 conversion of 44 % obtained at 20 bar and a H2/CO2 ratio of 8 [24].With the aim of tailoring Fe-based materials as CO2 methanation catalysts, studies on increasing the C2–C4 fraction in CO-FTR, with CH4 as an undesirable product, provide information on the direction of necessary properties for high CH4 yields: In general, iron carbides are considered as the active phase in FTR and active carbon sites contribute to the chain growth mechanism [26]. In addition, the activity and selectivity is closely related to the particle size of the Fe-based catalysts [26]. Smaller Fe nanoparticles (<7–9 nm) lead to higher CH4 selectivity [27–29]. It was concluded, that low coordinatively unsaturated corner and edge sites are important for CH4 formation, while terrace sites of the bigger Fe particles are responsible for olefin generation [29,30]. Hence, the selectivity of Fe based catalysts for CO2 methanation could be improved by decreasing the Fe particle size. This stands in contrast to the particle size dependency of Ni based CO2 methanation catalysts, which decrease in selectivity if the particle size decrease below 2 nm [31,32].Supporting Fe on zeolites enables a way to produce stable and highly dispersed Fe species. This has been proven by their use as highly stable selective catalytic reduction (SCR) catalysts [33–35]. Despite the high Fe dispersion, zeolites offer additional tailoring possibilities and have shown to positively influence the CO2 methanation performance of Ni-based catalysts [36]. Namely by their compensating cation [37], Si/Al ratio [38] and zeolite framework type such as FAU, BEA, MFI and MOR [39]. Due to the high affinity of zeolites to adsorb water they allow further improvement of catalytic activity by applying so called sorption enhanced conditions whereat H2O is adsorbed by the zeolite and in that way pulled away from the reaction center [40–44]. To the best or our knowledge it was not investigated yet how the combination of Fe supported on zeolites perform in CO2 methanation. In the present study a series of differently loaded Fe on zeolite catalysts are investigated at ambient and elevated pressure up to 15 bar with the aim to increase the CO2 methanation performance. In order to avoid restricting the product spectrum resulting from pore size effects within the zeolite, 13X was selected as zeolite support. On the one hand due to its relatively large and three dimensional pore structure where molecules up to a kinetic diameter of 7.35 Å can form and diffuse freely along all axis including CO, CH4, CH3OH as well as C–C coupled products up to at least C6 compounds. On the other hand a large range of Fe loadings can be theoretically ion exchanged due to the high aluminium content of 13X. As main focus the trends in activity and selectivity with increasing pressure as well as iron loading are carefully analysed and correlated with the properties of the catalysts. This leads to a justification if CO2 methanation on Fe-based catalyst will become feasible as an attractive alternative.1, 5 and 10 wt % Fe/13X catalysts were synthesised by wet impregnation with a 0.05 M Fe(NO3)3 · 9H2O (99 % Sigma-Aldrich) solution in ethanol on commercial Na-13X zeolite (ZEOCHEM, Si/Al = 2.5; Faujasite structure). After ion exchange for 24 h at room temperature under intense stirring ethanol was evaporated in a rotary evaporator. The resulting solids were dried at 80 °C for 12 h and calcined at 400 °C (heating ramp = 5 K/min) for 5 h in a continuous flow of air. The 5 wt % catalyst with collapsed zeolite structure was synthesised by wet impregnation for 30 min with a 0.05 M Fe(NO3)3 ∙ 9H2O aqueous solution on commercial Na-13X. Water was evaporated in a rotary evaporator, the resulting solid dried at 80 °C for 12 h and calcined at 500 °C (heating ramp 5 K/min) for 5 h in a flow of air.The weight loadings of iron of all samples were analysed via Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) on an Agilent 720 ES. The X-ray powder diffraction pattern were measured on a Bruker D8 Advance diffractometer with Ni filtered Cu Kα radiation (λ = 1.5406 Å) and a step size of 0.2° from 2°θ = 20–90. Crystallite sizes of the Fe-particles were calculated according the Debeye–Scherrer equation using the half width of the reflex at 44.7°. UV/vis spectra were collected on a UVVISNIR Lambda 950 spectrometer from Perkin Elmer equipped with a 150 mm integration sphere to analyse the diffuse reflectance of the Fe-zeolites. The spectra where recorded in reflexion mode in a wavelength region of 800–200 nm and a step size of 5 nm. Specific surface area, pore diameter and pore diameter dispersion were analysed by N2 physisorption at 77 K in a Quantachrome Autosorb IQ TPX. All samples were degassed for 12 h in vacuum at 200 °C. The pore diameter and dispersion were analysed according the BJH method from the desorption branch and specific surface area (SSA) by using the BET method. The pressure range for analysis was defined by rouquerol analysis in order to stay in the linear regime of the BET analysis [45]. The microporous surface area was distinguished from the external and mesoporous surface area by the t-plot method. Temperature controlled analysis were performed in the same Quantachrome Autosorb IQ TPX in dynamic mode and with a thermal conductivity detector. For temperature programmed reduction (H2-TPR) all samples were degassed at 400 °C in a flow of N2 for 30 min. Subsequent to the cooling down procedure to 40 °C, TPR was started in a flow of 5 vol% H2 in N2, with a total flow rate of 25 mL/min and a heating ramp of 5 K/min up to 850 °C and isothermally treated at the end temperature for additional 30 min. NH3 was used in order to analyse the acidic properties of the zeolite in the temperature controlled desorption (TPD) experiments. Prior to the analysis all samples were reduced in a flow of 50 % H2 in N2 at 400 °C for 30 min, accordingly to the pre-treatment of the catalytic tests. Residual adsorbed hydrogen was flushed-off from the samples by additional 2 h treatment in N2 at 400 °C. Subsequently, adsorption of 10 % NH3 in N2 was performed at 100 °C and physisorbed NH3 was purged in a flow of N2 at 100 °C for 30 min. TPD was performed in a flow of 25 mL/min N2 and a heating ramp of 10 K/min up to 800 °C. Scanning electron microscopy (SEM) analysis was performed in a Thermo Scientific Phenom XL equipped with a back scattered detector. Concurrent elemental mapping was carried out by using the integrated EDX detector.Methanation tests were performed in fixed bed flow reactor system with an inner diameter of 6 mm at ambient and elevated pressure (5, 10, 15 bar) at a GHSV = 4186 h−1. Prior to the catalytic tests all catalysts were reduced within the reactor in a flow of 50 % H2 in N2 for 30 min at 400 °C and ambient pressure. In a typical run 25 mL/min CO2, 100 mL/min H2 and 12 mL/min N2 as internal standard were supplied by mass flow controller (Bronkhorst, El Flow). During the methanation tests the temperature was raised from 200 to 400 °C in steps of 50 °C and kept constant at reaction temperature for 30 min. The composition of effluent gases from the reactor was monitored by online raman spectroscopy (Kaiser Raman RXN2 spectrometer equipped with AirHead probes). The conversion X, selectivity S and reaction rate of CO2 conversion (r(CO2)) were calculated according Eqs. (4)–(6): (4) X C O 2 = n ˙ C O 2   i n -   n ˙ ( C O 2   o u t ) n   ˙ ( C O 2 i n ) (5) S C H 4 =   n ˙ ( C H 4   o u t ) ∑ n ˙   ( p r o d u c t s ) (6) r C O 2 =   X C O 2   ×   n ˙ ( C O 2 )     n F e c a t   With n ˙ i as the molar flow of component i, and n(Fecat) as the molar amount of Fe in the catalyst bed within the reactor.Catalysts with three different weight loadings (1, 5, 10 wt %) of Fe on 13X were prepared via impregnation. Elemental analysis via ICP-OES confirms the presence of Fe on 13X close to the aimed amounts of Fe on the samples (Table 1 ).Since the zeolite framework is prone to destruction by iron, the integrity of the structure was validated via XRD analysis.The impregnation procedure and calcination temperature strongly influences the stability of the iron impregnated zeolites. Hence, a synthesis optimization was conducted: the zeolite structure stays intact only by avoiding H2O as a solvent and using ethanol as well as decreasing the calcination temperature to 400 °C (Fig. 1 ). Nevertheless, with higher Fe-loading the decrease of intensity of reflexes shows the incipient destruction of the framework even by applying the optimized procedure. Compared to the pure 13X, 1 wt % Fe/13X shows nearly no changes in intensity and all catalysts show reasonable stability. In contrast, the zeolite structure of 5 wt % Fe/13X impregnated in H2O and calcined at 500 °C completely vanishes. For this reason, all catalysts were prepared in ethanol and by calcination at 400 °C and it was avoided to exceed this temperature at any time. As a pretreatment in the catalytic test a reduction of the catalysts in 50 % H2 in N2 at 400 °C for 30 min was performed. Comparison of XRD of ex situ reduced and as calcined catalysts (Fig. 1) ensure that the zeolite structure stays intact for all Fe loadings during the pre-reduction and confirm the Fe-reduction by the raise of the specific reflex of metallic Fe at 2°θ of 44.7° (inset in Fig. 1). According to the Debeye–Scherrer equation extracted Fe crystallite sizes from this reflex are 33 and 23 nm for 5 wt % and 10 wt % Fe/13X, respectively. Solely the reduced 1 wt % Fe/13X does not show this specific reflex. This could be due to two reasons, or a combination thereof: Either the Fe loading is too low for the sensitivity of XRD or the Fe species are highly dispersed within the framework of the zeolite.N2 physisorption analysis confirms the presence of microporosity of all Fe/13X catalysts calcined at 400 °C. Nevertheless, the specific surface area decreases from 612 to 161 m2/g with increasing Fe loading. In line with the decreasing reflex intensity of the zeolite lattice from XRD analysis the micropore area extracted from t-plot analysis decreases from 573 down to 32 m2/g (see Table 1).The dispersion of Fe within the zeolite framework after calcination was analysed with UV/vis spectroscopy. The line shape of the spectra arising from O → Fe3+ charge transfer are rather similar (Fig. 2 ). In all spectra, four distinct peaks are separated by deconvolution (Figs. S1–S3). Two strong bands are found below 300 nm that are assigned to isolated Fe3+ ions. Whereat the band centered at 205 nm attributes to charge transfer from tetrahedral coordinated Fe3+ and the band at 250 nm relates to Fe3+ in higher coordination [34]. The two bands above 300 nm arise from agglomerated Fe-species. Whereby the band from octahedral Fe3+ species in small oligomeric FexOy cluster appears at 350 nm and from large Fe oxide particles as a very broad band at 436 nm. Quantitative analysis of the deconvoluted bands shows that all samples have the same relative amount of Fe3+ in tetrahedral sites. Contrary to this, 1 wt % Fe 13X shows with 55 % of all Fe3+ ions relatively more Fe ions in dispersed and oligomeric octahedral sites. Solely 30 % of the Fe ions agglomerate to particles. In comparison to this, the two higher loaded samples have comparable factional amounts of Fe in all sites and more than 55 % of Fe agglomerate into particles.The reducibility of the Fe/13X catalysts was investigated by H2TPR experiments (Fig. 3 ). In line with the Fe loading of catalyst the intensity of the signals increases and the features of 5 and 10 wt % Fe/13X samples are rather similar. These two samples show a very intense and broad signal between 200–550 °C with a peak maximum that shifts to lower temperatures from 442 to 405 °C with increasing Fe loading from 5 to 10 wt %. In agreement with literature these signals correspond to the reduction of Fe of the agglomerated FeOx particles and dominate the TPR [46]. In addition, two more signals appear at temperatures higher than 550 °C that go in line with the collapse of the zeolite structure.In the TPR of 1 wt % Fe/13X three distinct peaks appear in the temperature region of zeolite’s thermal stability with peak maxima at 375, 424 and 498 °C. According to literature, reduction of Fe3+ within the zeolite structures as well as reduction of Fe2O3 to Fe3O4 from oligomeric and small cluster takes place at lower temperature [47]. The visibility of the fine structure of reduction under the same measurement conditions shows on the one hand that agglomerated FeOx-species are not the main species, and on the other hand, that Fe species coordinated on different sites of the zeolite framework are present in this sample.Temperature programmed desorption of NH3 was performed in order to analyse the influence of the Fe loading on the zeolites acidity (Fig. S4). In line with the decrease of reflex intensity of the zeolite framework in the XRD with increasing Fe loading the total number of acid sites decreases. The main signal in the TPD appear at the same temperature region. Hence, even though the number of acid sites decreases with increasing Fe load, the acid strength as well as nature of acid sites remain constant in all samples. Therefore, it can be excluded to significantly influence the selectivity of the catalysts.All prepared materials were investigated in a temperature region of 200–400 °C and at pressures of 1, 5, 10 and 15 bar.In a first step the two 5 wt % Fe/13X with a collapsed (prepared in H2O and calcined at 500 °C, broken lines in Fig. 4 ) and intact zeolite structure (calcined at 400 °C & exclusion of H2O from the synthesis, solid lines in Fig. 4) were compared by their catalytic performances. In case of the catalysts with a collapsed framework after synthesis a rather low CO2 conversion of 10 % was observed by increasing the temperature up to 400 °C, even at 15 bar. For this reason, the temperature range of the catalytic test was expanded to 550 °C for this catalyst. At 1 bar no significant CO2 conversion was observed up to 550 °C. Likewise all other investigated catalysts, the CO2 conversion increases with temperature and increasing pressure of the catalytic tests. With 5 wt % Fe on collapsed 13X reasonable CO2 conversion was achieved up to 74 % at 550 °C and 15 bar. On this catalyst, CO is the main product at low pressure. With increasing pressure, selectivity towards CH4 increases up to 85 % at 15 bar and 550 °C. No Fischer–Tropsch products were observed under any conditions.On the contrary, 5 wt % Fe/13X with an intact zeolite framework shows already reasonable CO2 conversion of 33 % at 1 bar and 400 °C. CO2 conversion increases with temperature and pressure up to 88 % at 400 °C and 15 bar. Incipient activity is already obtained at 250 °C. Hence, the comparison of these two catalysts clearly demonstrates that an intact zeolite framework is essential to obtain and support high catalytic performances at reasonable temperatures.In order to stay in the kinetic regime the Fe-normalized reaction rates at 300 °C are used to compare the activity of produced catalysts with intact zeolite structure and different Fe loading at pressures from 1 to 15 bar (Fig. 5 ). The catalyst masses included in the reactor and corresponding Fe-content from ICP analysis were used to calculate the Fe molar reaction rates. The two catalysts with 5 and 10 wt % Fe/13X show similar and increasing reaction rates at increasing pressure of up to 12 and 8 mmol(CO2)/(mol(Fe)∙s), respectively. This points out that the main active sites are the same in these two catalysts. In opposition to these results, 1 wt % Fe/13X shows much higher reaction rates at all investigated pressures up to 42 mmol(CO2)/(mol(Fe)∙s) at 10 bar. These trends show in correlation to the UV/vis analysis that finely dispersed Fe-species, which are the main species in 1 wt % Fe/13X, have a much higher catalytic activity than the agglomerated Fe-species, that are the main species of the 5 and 10 wt % loaded Fe/13X catalysts. Nevertheless, the reaction rate decreases upon further increase of the pressure from 10 to 15 bar against the principle of Le Chatelier. This might be either due to hampered desorption of one of the products from the catalyst surface or reconstruction of the Fe-species or zeolite framework at elevated pressure.The catalytic performances at 350 °C are used to compare the variation of product selectivity of different Fe loading and at varying pressures (Fig. 6 ).At 1 bar the 10 wt % Fe/13X catalyst shows a very high selectivity towards CO (S(CO) = 97 %) and minor selectivity to CH4 (Fig. 6a). With increasing pressure to 15 bar the selectivity towards C–C coupled products and CH4 increases monotonously. CH4 becomes the main product at 10 bar and reaches its maximum selectivity of 61 % at 15 bar. The selectivity to C–C-coupled products increases up to 22 %, while the selectivity towards CO decreases to 15 %.Likewise, 5 wt % Fe/13X catalyst shows the same trend with increasing pressure (Fig. 6b). At high pressures it increases its selectivity towards the desired product CH4 up to 68 %, while the selectivity towards CO (S(CO) = 14 %) and CC-coupled products (S(CC) = 17 %) stays relatively low.In contrast to the behavior of the two higher loaded Fe catalysts 1 wt % Fe/13X shows at 1 bar already significant CH4 selectivity of 22 % (Fig. 6c). The CO selectivity of 21 % at 350 °C and 1 bar is relatively low and selectivity towards C–C coupled products is at 56 % and therefore surprisingly high in that sequence. In opposition to the trend of 5 and 10 wt % Fe/13X as well as literature [26] on Fe-based Fischer–Tropsch catalysts, the selectivity towards C–C-coupled products decreases with increasing pressure on 1 wt % Fe/13X. The main product is CH4 from 5 to 15 bar with a selectivity up to 76 % at 10 bar and 350 °C. Comparable product selectivites of S(CO) = 11 % and S(CC) = 14 % are observed at 10 and 15 bar. This trend of decreasing selectivity towards C–C coupled products and increasing CH4 selectivity with increasing pressure is opposed to the general trend of Fe-based Fisher–Tropsch catalysts reported in literature [26], in which Fe3C is regarded as active species. But it stands in line, that more coordinative unsaturated Fe species have a higher tendency to produce CH4 [29,30].XRD analysis of the catalyst after the catalytic tests shows a decrease of the reflexes from the zeolite framework for all samples (Fig. 7 ). Nevertheless, 1 wt % Fe/13X shows considerably high intensities. Hence, the integrity of the framework is still given in the mayor fraction of the sample even though a small fraction of the zeolite framework collapses. In the XRD of this sample no other reflexes from other phases than the 13X framework are visible. In contrast to the results of 1 wt % Fe catalysts 5 and 10 wt % Fe catalyst show significant decrease and a full depletion of the reflexes from the zeolite. Additionally, reflexes from a Fe3C phase appear in the diffractograms of both catalysts after the methanation experiments. Given by the sharp shape of metallic Fe reflex in the 5 and 10 wt % Fe/13X catalysts the crystallite size of Fe increases during the catalysis. On the low loaded 1 wt % Fe/13X still no reflexes origin from metallic Fe or Fe3C, respectively, under reaction conditions. Hence, the included Fe is very stable within the framework of the zeolite.The collapse of the zeolite with high Fe loading is visible in the SEM micrographs as well. The spherical shape of the zeolite crystallites is still visible in the used 5 wt % Fe/13X catalysts (Fig. S5). This shape nearly vanishes completely on the 10 wt % Fe/13X catalyst after operation. Larger fragments with different morphology, consisting of Al and Si, become obvious instead (Fig. 8 bottom, middle & right). In addition to this, the formation of larger Fe particles is visible, too. In comparison to SEM images of the reduced catalysts prior to the catalytic testing it seems that Fe migrates out of the zeolite particles and forms, together with deposited carbon, an outer shell around the support (Fig. 8 bottom). In the case of the used 10 wt % Fe/13X catalyst the EDX mapping indicates that residual zeolite particles do not contain Fe at all, while the concentration of Fe is comparably high on the amorphous fragments (Fig. 8). As opposed to this the spherical shape of the 1 wt % Fe/13X zeolite catalyst particles appears to be unchanged after the reaction in the SEM micrographs (Fig. 8 top). In addition, EDX analysis shows a homogeneous dispersion of Fe over the sample and no larger particle agglomerations of Fe or creation of a common Fe–C shell/layer is visible. Hence, the SEM micrographs confirm together with XRD analysis the destruction of the higher loaded zeolite during the catalytic run under formation of a Fe3C shell at the outer layer of the catalysts.The hydrogenation of CO2 towards CH4 on differently loaded Fe/13X catalysts was investigated at ambient and elevated pressure (5–15 bar). Comparison of the catalytic performances with a catalyst with collapsed zeolite framework shows, that an intact zeolite structure and hence high dispersion of Fe within the catalyst is essential for high CO2 conversion at temperatures below 400 °C and all investigated pressures.Catalytic tests on 10, 5 and 1 wt % Fe/13X catalysts with intact zeolite structure revealed a different reactivity of the two higher loaded catalysts compared to the 1 wt % Fe/13X catalyst. Higher Fe-loading leads to relatively low reaction rates of up to 12 mmol(CO2)/(mol(Fe)∙s) at 15 bar. CO is the main reaction product at low pressures of 1 and 5 bar. With increasing pressure the selectivity towards CH4 as well as C–C-coupled products increases. The low Fe-loading of 1 wt % leads to a significant increase of the molar reaction rate at all investigated pressures up to 42 mmol(CO2)/(mol(Fe)∙s) at 300 °C and 10 bar. In contrast to both higher Fe-loadings, the lower Fe-loading leads to high selectivity for C–C-coupled products of 56 % at 1 bar. The selectivity towards desired CH4 increases up to 76 % with increasing pressure at the expenses of the formation of CO and C–C-coupled products.Physico-chemical characterization before and after the catalytic run show on the one hand that in 5 and 10 wt % Fe catalysts, Fe is mainly present as agglomerated particles. This leads to a destabilization of the zeolite and further agglomeration of Fe under reaction conditions with simultaneous formation of Fe particles embedded in a Fe3C-phase as an outer shell layer. On the other hand, in 1 wt % Fe/13X, Fe is mainly present as octahedrally coordinated dispersed and oligomeric species. This leads to a higher hydrothermal stability of the catalysts and neither formation of larger Fe agglomerates nor Fe3C-phase formation under operation. The high dispersion of Fe within the material suppresses CC coupling reactions at higher pressure due to confined neighboring Fe sites and this in turn supports the hydrogenation of CO2 to methane. At 15 bar the selectivity towards CO is limited down to 8 %. Even though the performance is not yet fully optimized, the presented results show, that the utilization of Fe-based catalysts as alternative to more expensive and especially hazardous Ni-catalysts for e.g. biogas upgrading and feed into the natural gas grid becomes considerable and provides essential prerequisites for the direction of further catalyst optimization.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Tanja Franken: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Writing - original draft, Writing - review & editing, Visualization, Project administration, Funding acquisition. Andre Heel: Conceptualization, Methodology, Writing - review & editing, Supervision, Project administration, Funding acquisition.The authors kindly acknowledge funding of the SmartHiFe Project by the Swiss Federal Office of Energy (grant number: SI/501754-01). Additionally the authors kindly thank Michal Gorbar and Dr. Roman Kontic for their support during the SEM and XRD analyses.Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jcou.2020.101175.The following are Supplementary data to this article:

The raise of regenerative but unsteadily produced energy demands a highly flexible way to store the energy for time periods when less energy is produced than consumed. In the current study, it is investigated if catalysts based on environmentally more attractive and less hazardous to health Fe might be able to be considered as an alternative to Ni catalysts in the CO2 methanation at elevated pressure. For this a set of catalysts with 1–10 wt % Fe supported on the zeolite 13X is analysed in CO2 methanation at 1–15 bar. The trends of activity as well as selectivity with varying Fe loading and pressure are presented. Correlation with thorough characterization of the materials shows that a very high dispersion of Fe in octahedral sites within the zeolite is necessary to generate CH4 as the main reaction product and suppress the Fischer–Tropsch activity towards CC coupling reactions at elevated pressure. Especially with low Fe loading such as 1 wt % high reaction rates of 42 mmol(CO2)/(mol(Fe)∙s) with a CH4 selectivity of 76 % at 300 °C and 10 bar are obtained. In contrast to that, highly Fe loaded catalysts tend to form increasing amounts of Fischer–Tropsch products at increasing pressure. In addition, highly Fe-loaded catalysts are much more susceptible to destruction of the zeolite under reaction conditions. At the same time, highly loaded catalysts form a Fe3C shell around the remaining support. Hence, avoiding the formation of a Fe3C phase is crucial for high CH4 selectivity. The results presented here therefore show that catalysts with a very high Fe-dispersion in particular can gain considerably in importance as alternatives to Ni-methanation catalysts at elevated pressure.

Data will be made available on request.Due to the growing concern of the greenhouse gas reduction and the discovery of surplus shale gas reservation, the dry reforming of methane (DRM) reaction has attracted widespread attention [1–7]. The dry reforming of methane (DRM) is a prospective process to convert two major greenhouse gases, carbon dioxide (CO2) and methane (CH4), into syngas. The methane consumption in DRM is half of the steam reforming and partial oxidation reforming of methane [5]. It could potentially alleviate the adverse influence of these pollutants while supplying widely consumed chemicals [8–10]. The syngas produced in the reaction has a stoichiometric ratio of molecular hydrogen (H2) to carbon monoxide (CO) of 1:1, so to obtain a more favorable ratio, such as that for Fischer–Tropsch synthesis (2:1), we must cope by implementing such processes as steam reforming (3:1) or autothermal reforming (2.5:1) [11]. The problem is that these two processes have a very negative carbon footprint, and it can be more efficient to trap carbon and raise the H2:CO ratio by taking advantage of the reactivity of CO [12,13]. Moreover, the ratio can be lower with a subsequent reverse water-gas shift reaction. The rest of the reactions involved in the process are the Boudouard and CH4 decomposition [14]. One of the critical elements in this reaction is the catalyst stability, which depends on various factors, including coking [15,16]. The stability of the catalyst plays a role that could be more important than the activity per se [17].Among the long list of catalysts used and tested for this process, nickel (Ni)-supported catalysts stand out for their balanced cheap price and high activity from an industrial point view [18]. However, these catalysts greatly suffer from severe coking and thermal sintering due to their low Tammann temperature [19–22]. Hence, the development of Ni-based catalysts with superior activity and stability sparked numerous catalysis studies. Many factors impact the sintering and coking resistance of the Ni-based catalysts, such as particle size, the strong metal support interactions (SMSI), surface oxygen species and lattice oxygen in the support, surface carbonate species and the formation of alloy. Reducing the particle size will provide more active sites and suppress the carbon deposition [23]. The SMSI highly improves the dispersion of Ni particles and alleviate the possibility of sintering, leading to the enhancement of the activity and preventing coking [24–26]. Significant studies by Kawi’s group have demonstrated that both surface and lattice oxygen species were investigated to play a key role in activating the C-H bond of CH4 molecule and carbon suppression [27–29]. The lattice oxygen could react with CO2, forming monodentate and bidentate carbonate, which alleviates the carbon deposition by oxidation of surface carbon and exhaust as CO [30]. In addition, alloying with a second metal, altering the geometric and electronic structure of Ni active site, will achieve an optimized synergetic effect, enhancing the activity, selectivity, and stability [31]. Due to the inherent scientific significance and crucial role in technology view, bimetallic alloys have obtained widespread appeal [32]. In this case, the goal is to change the electron density of the Ni atoms by inducting additional metal, affecting the typically considered rate-determining step: CH4 dissociation [33]. The alloy formed between the Ni and second metal can improve the stability, alleviating the coke deposition (e.g., noble metals are known to enhance the stability of Ni-based catalysts) [34]. A promising approach to enhance the performance of Ni-supported catalysts without affecting the price that much is to use a secondary metal, such as iron (Fe), cobalt (Co), copper (Cu), or molybdenum (Mo) [31,35]. Moreover, Fe is very interesting due to its low cost and synergistic effect upon intimate interaction with Ni [36,37].The support and its integration with the metal also play a crucial role in catalyst activity, selectivity, and stability [38]. Conventionally, the synthesis, such as wet impregnation or vapor deposition, lacks control of particle size, dispersion, morphology, and metal-support interaction, leading to faster sintering and coking [39–41]. The in situ exsolution on perovskites overcomes these problems, improving the metal-support interactions and stabilizing the Ni exsolved particles [42–46]. In this method, the transition metal cations partially substitute the perovskite oxide (ABO3) B-site cations, then migrate (exsolve) from the host lattice and agglomerate in the form of nanoparticles under reduction conditions [47,48].Among the investigated perovskites in the exsolution concept, A-site ordered double-layer perovskite system PrBaMn2O5+δ has drawn significant attention in solid oxide fuel cells due to its thermal and chemical stability, high oxygen diffusion rate, appreciable catalytic activity in hydrocarbon oxidation and the high flexibility to regulate first-row transition metal in the B sites as the active sites [49,50]. All these features benefit the performance of the catalyst under harsh DRM conditions. In DRM, several studies have investigated this type of perovskite with Ni with an 11 % CH4 conversion at 800 °C and 5 h on stream with little coke deposition (0.017 gcoke gcat ‑1 h‑1) [51]; or with Co-Mo with a CO2 conversion higher than 95 % at 800 °C and 24 h on stream [52]. To avoid most metal cations remaining embedded in the host bulk, resulting in low metal utilization, Joo et al. [53] performed systematic research on the perovskite by adopting the topotactic exsolution approaches in which guest cation was first supported on the matrix perovskite and then ion exchanged with the exsolved cation. The multi-step strategy improved the B-site transition metal exsolved fractions compared to typical conditions, which was applied to catalyst systems such as Ni-Fe [54], Co-Fe [55], and Co-Ni-Fe [56], to facilitate more exsolved nanoparticles, all of which displayed enhanced catalytic activity. However, novel strategies need to be explored and developed aiming a larger fraction of exsolved metal in a facile way. In addition, the enhancements of the stability and fundamental understanding of the deactivation mechanisms over these exsolved catalyst systems need to be clarified too.Firstly, we aim to explore and develop new methods to exsolve a high fraction of Ni and Ni-Fe alloy and make uniform-sized metal nanoparticles anchored into a stoichiometric PrBaMn1.6Ni0.4–2xFe2xO5+δ double-layer perovskite. Secondly and given the intimate interaction between the nanoparticles and the support, our goal is to understand the reasons behind the potential higher stability (slower coking and sintering) in worst case scenarios: long term reactions and at high pressure. To this aim, we will combine several catalyst formulations with variable proportions of Ni and Fe in PrBaMn1.6Ni0.4–2xFe2xO5+δ matrix with fixed ratios of Pr, Ba, and Mn, preparation methods, and characterization techniques, including x-ray absorption spectroscopy, dry reforming reactions and ab initio calculations. We will determine the fine structure of the catalyst responsible for the potential higher stability of the catalyst.A series of Pr0.5Ba0.5Mn0.8Ni0.2−xFexO3 (x = 0, 0.05, 0.1, 0.2), defined as P-Ni0.2−xFex was synthesized using the improved sol-gel method. Stoichiometric Pr(NO3)3·6 H2O (Aldrich, 99.9 %, metal basis), Ba(NO3)2 (Aldrich, 99 %), Mn(NO3)2·4 H2O (Aldrich, 98 %), Ni(NO3)2·6 H2O (Aldrich, 98.5 %), and Fe(NO3)3·9 H2O (Aldrich, 98 %) were dissolved in distilled water. The appropriate amounts of citric acid (Aldrich 99.5 %) and ethylene glycol (Aldrich) were added into the solutions as complexation agents, adjusting the mole ratio of metal ion to citric acid to ethylene glycol as 1:3:1.5. The pH value of the solution was maintained at around 8 by adding ammonium hydroxide. The resulting aqueous solution was continuously stirred at 85 °C forming a uniform gel, which was heated at 350 °C to decompose slowly and completely. Then, the precursor powder was ground and calcined at 950 °C for 4 h in air. After reduction pretreatment, the single layer perovskite transformed to double layered PrBaMn1.6Ni0.4–2xFe2xO5+σ, defined as E-Ni0.2- xFex. A sequence of Pr0.5Ba0.5Mn1−xNixO3 (x = 0, 0.1, 0.3), defined as E-Nix, was also prepared with the same procedure to investigate the effect of the Fe promoter. In contrast, the corresponding Ni-impregnated Pr0.5Ba0.5MnO3 was prepared via the wetness impregnation method. Regarding the wetness impregnation method, a Pr0.5Ba0.5MnO3 perovskite support, prepared with the same process mentioned above, was impregnated with a proper amount of Ni(NO3)2·6 H2O (Aldrich, 98.5 %) solution. And then the slurry was dried overnight at 80 °C, calcined at 350 °C for 2 h and at 950 °C for 4 h, respectively to achieve the Pr0.5Ba0.5MnO3/Imp Nix (x = 0.1, 0.2, 0.3) samples, defined as Syn-I-Nix. After reduction pretreatment, the perovskite support in Pr0.5Ba0.5MnO3/Imp Nix transformed to double layered perovskite in PrBaMn2O5+σ/Imp Nix, defined as I-Nix. The chemical composition of the prepared materials and their abbreviations are presented in Table S1. The nominal and actual loading of different catalysts are presented in Table S2.The XRD was performed using a Bruker D8 Advanced A25 diffractometer in Bragg–Brentano geometry equipped with a Cu Kα target (λ = 1.54056 Å) at 40 kV and 40 mA in the range of 10º to 80° under continuous scanning mode. The N2 adsorption-desorption measurement was performed to analyze the specific surface area with a Micromeritics ASAP 2020 surface area and porosity analyzer by collecting the nitrogen sorption isotherms at 77 K, from which the specific surface area was calculated according to the Brunauer–Emmett–Teller equation.Thermogravimetric analysis and differential thermogravimetric measurements of the fresh samples were conducted in 5 % H2/N2 from 25 °C to 800 °C at a heating rate of 10 °C min–1 using the Mettler–Toledo Star system. The thermogravimetric analysis mass spectrometry was conducted under temperature-programmed oxidation to study the amount and type of C deposited on the catalysts. The used catalysts were heated to 800 °C at 10 °C min‑1 in 10 % molecular oxygen (O2)/argon (Ar).The reducibility of the perovskites was investigated by the H2 temperature-programmed reduction using a Micromeritics AutoChem II 2920 chemisorption analyzer. Before the measurement to remove the absorbed impurities, 100 mg sample was pretreated at 300 °C for 1 h with Ar and then cooled to 50 °C. After switching the gas flow to 10 % H2/Ar, the H2 consumption was monitored with a thermal conductivity detector during heating the temperature from 150 °C to 850 °C at 10 °C min–1. Moreover, a cold trap made of ice was set between the sample and detector to remove the water formed during the process.The oxygen temperature-programmed desorption (O2-TPD) was also conducted with the Micromeritics AutoChem II 2920 chemisorption analyzer. The samples were first pre-reduced at 800 oC with 10 % H2/Ar for 6 h and then in-situ reduced for 30 min in AutoChem. After flushing the sample with He at the sample temperature for 30 min, the system cooled down to 50 °C in He. Then the sample was treated with 20 % O2/N2 at 50 °C for 30 min. After flushing the sample with He for 1 h, the O2-TPD was performed by increasing the temperature from 50 °C to 800 °C in 40 mL min–1 He with a ramp of 10 °C min–1.The morphology of the samples was observed using scanning electron microscopy (SEM; FEI Teneo VS) at an accelerated voltage of 5 kV. The sample was deposited on graphite and sputtered with a gold (Au) conductive layer. The sample TEM was performed using a Cs-Probe corrected Titan microscope from Thermo Fisher Scientific. It operated at the accelerating voltage of 300 kV and with a 0.5–0.8 nA beam current. Dark-field imaging was performed by STEM coupled with a HAADF detector. The STEM-HAADF data were acquired with a convergence angle of 21.4 mrad and a HAADF inner angle of 49 mrad.Furthermore, an x-ray energy-dispersive spectrometer (FEI SuperX, ≈0.7 sR collection angle) was also employed with dark-field STEM imaging to acquire STEM-EDS spectrum-imaging datasets (dwell time 2.5 µs). A corresponding EDS spectrum was obtained during the acquisition of these datasets to generate the elemental maps at every image pixel. After background subtraction, the elemental maps for Ni, Fe, Mn, Pr, Ba, and O atoms were computed using the extracted intensity of their respective Kα or Lα lines. The generated maps were slightly post-filtered by applying a Gaussian filter (sigma = 0.5). The STEM electron energy loss spectroscopy (EELS) analysis was performed by operating the microscope at the accelerating voltage of 300 kV using a convergence angle of 17 mrad and an effective collection angle of 36 mrad. The spectrum-imaging dataset included the simultaneous acquisition of zero-loss and core-loss spectra (DualEELS) using a 0.5 eV/channel dispersion. It was recorded using a beam current of 0.2 nA and a 5 ms pixel–1 dwell time. The Fe L2,3-edge, Ni L2,3-edge, and O K-edge were selected to build the chemical maps. Plural scattering was removed from the Fe and Ni L-edges using Fourier-ratio deconvolution with prior energy shift correction and background subtraction (power-law model). The contribution of transitions from the 2p3/2 and 2p1/2 initial states to the continuum states must be considered to acquire the white-line intensities of the L3 and L2 edges. The latter was conducted through a classical normalization with the Athena softwar [57]. Then, the L-edge spectra were modeled with a double arctangent step function and two split Lorentzian functions to account for the peak asymmetry [58]. The white-line intensities were finally computed from the area of the Lorentzian peaks.The x-ray absorption spectroscopy of the prepared and reduced catalysts was performed at the 1W1B beamline at the Beijing Synchrotron Radiation Facility. The catalysts were ex-situ reduced for XAS analysis. The fresh catalysts were first pressed into pellets and reduced under the corresponding gas conditions. And then, the reduced pellets were collected and sealed in a glove box. Afterward, the sealed pellets were tested. The data were collected in the transmission mode via a Si (111) double crystal monochromator, detuned to reject higher harmonics. In addition, the Ni (8333 eV) and Fe (7113 eV) standard foils were applied for energy calibration. The Ni foil, NiO, Fe foil, FeO, and Fe2O3 were used as references to analyze the XANES and EXAFS. The as-obtained XAFS data were processed using the Athena software program.The XPS was conducted to investigate the chemical surface composition using a Kratos Axis Ultra DLD spectrometer equipped with a monochromatic aluminum Kα x-ray source (hν = 1486.6 eV). The XPS spectra were referenced to the C1s binding energy of 284.6 eV. The fitting of the XPS peaks was processed using XPSPEAK software.The Raman analysis of the used catalysts was performed on an RXN1 Raman spectrometer (Kaiser Optical Systems) fitted with a 532 nm laser operating at 40 mW. In addition, inductively coupled plasma-optical emission spectrometry was performed to analyze the exact elemental content using an Agilent 5100 instrument. Before the measurement, the materials were digested in an ETHOS1 microwave digestion milestone.In-situ diffuse reflectance infrared Fourier transform spectroscopy (in-situ DRIFTS) was conducted in a Nicolet 6700 IR spectrophotometer (Thermo Scientific) equipped with a Harrick Praying Mantis DRIFTS gas cell. Before the measurements, the catalysts were pre-reduced at 800 °C for 6 h in a quartz tube and then reduced in situ in the DRIFTS cell at 450 °C for 1 h, followed by flushing with He flow. The background was collected under a He flow at 450 °C. Gas-switching experiments (step 1 (He flow after CH4) → step 2 (CO2 flow) → step 3 (mix gas flow: CH4/CO2/N2 =33/34/33) were carried out at 450 °C to unravel the evolution characteristics of surface species on the catalysts. The time-resolution IR spectra were recorded in a range of 400–4000 cm−1 at 32 scans per spectrum and 4 cm−1 resolution with an interval of 30 sCatalytic tests were performed in a four-channel Flowrence XD platform from Avantium. The reactors are 300-mm-long quartz tubes, of which the outside and inside diameters are 3 and 2 mm, respectively. One of the reactors was adopted as the blank without a loading catalyst among the four channels. Typically, 10 mg catalyst was loaded in a quartz reactor, and the gas flow was 5 mL min–1 for each reactor. The catalysts were pelletized and sieved to achieve a powder with sizes between 150 and 250 µm. The proper amount of catalyst and reactant mixture gas flow was used to maintain the gas hourly space velocity (GHSV) per channel at 30,000 mL gcat –1 h–1 under atmospheric pressure and 12,000 mL gcat –1 h–1 under high pressure at 14 bar, respectively. The composition of the reactant mixture gas is CH4:CO2:N2 = 33 %:34 %:33 %. Prior to feeding the reactant gas, the catalysts were reduced in situ in a 10 % H2/Ar atmosphere for 6 h at 800 °C. The reactants and products were continuously monitored with an online micro gas chromatograph (Agilent 7890B).The conversions of i (CH4 or CO2) were calculated as: (1) X i = F i inlet – F i outlet / F i inlet Where F i inlet and F i outlet, denote the inlet and outlet molar flow rate of i. The apparent rates of reaction are calculated as follows: (2) -r i |app = F i inlet X i /(WNi) Where W is the catalyst loading and WNi is the nickel loading. The H2/CO ratio is defined as: (3) H2/CO = FH2 outlet / FCO outlet The apparent coke formation rate (rcoke|app in mmol gcat −1 s−1) was evaluated via temperature-programmed oxidation (TPO) combined with MS experiments. The coke formation rate was calculated based on Eq. (4). (4) rcoke|app = fcoke/t Where fcoke represents the amount of coke formed on the catalyst in mmol gcat −1 considering coke as pure carbon, at a given time on stream (t).The thermodynamic equilibrium conversions of CH4 and CO2 and the H2/CO ratio were determined by Aspen Plus software.Adsorption energy calculations were performed with the first-principles DFT using the Vienna Ab Initio Simulation Package (VASP). The electron exchange and correlation interactions were modeled using the generalized gradient approximation with the Perdew–Burke–Ernzerhof functional. The electron-ion interactions were defined using the projector-augmented wave method. A plane-wave basis set was used to describe the valence electrons with an energy cut-off of 400 eV. The Brillouin zone, sampled at the Monkhorst-Pack 3 × 3 × 1 k-point grid, was used as the Ni (111) and Ni4Fe1 (111) model. The Ni (111) and Ni4Fe1 (111) surfaces were modeled as a four-layer slab using a 5 × 5 supercell with 15 Å of vacuum between the slabs. All geometries were optimized until the convergence reached 1.0·10−6 eV, and the atomic forces were smaller than 0.05 eV Å−1. For the gas phase molecule, a cubic box of 15 × 15 × 15 Å3 was used. The climbing image nudged-elastic band method was used to identify the transition state structures of the elementary reactions involved in the reaction mechanisms. The following equation was used to calculate the binding energy of species present in the reaction media: (5) EBind = Eadsorbate+surface -Eadsorbate -Esurface, where E adsorbate+surface is the total energy of the adsorbate on the metal surface, and E adsorbate and E surface denote the total energy of adsorbate in the gas phase and the bare metal surface, respectively. A more negative binding energy value refers to the species adsorbed stronger on the metal surface (or a stronger interaction between the adsorbate and the metal site of the surface), and vice versa.The crystalline structures of the parent PrBaMn2O5+δ perovskite (P) and the corresponding exsolved counterparts P-Ni0.2−xFex (x = 0, 0.05, 0.1, 0.2) with different Ni/Fe loading were analyzed using x-ray diffraction (XRD) before and after reduction ( Fig. 1). Moreover, the crystalline structures of varying Ni substitution amounts for Mn have also been investigated (Fig. S1). The parent and exsolved materials comprise a perovskite structure with a mixture of hexagonal and cubic phases (Fig. 1a). However, the reduction induced the transformation of the original perovskite into a layered perovskite in the tetragonal phase (Fig. 1b), which might increase the specific surface area due to the formation of metal (Ni or Ni-Fe alloy) nanoparticles on the support surface and defects in the lattice (Fig. S2) [49].For P-Ni0.2, the diffraction peak located at 44.5° is attributed to metallic Ni due the exsolution [59,60]. Concerning P-Ni0.15Fe0.05, the similar diffraction peak shifts slightly to a lower diffraction degree, assigned to the diffraction peak of Ni-Fe alloy (Fig. 1c), deriving from Fe dissolving into the Ni lattice, revealing that the exsolved metal forms a binary Ni-Fe alloy [54]. Upon reduction, the exsolved nanoparticles in B sites are MnO for P, metallic Ni for P-Ni0.2, and Ni-Fe alloy for P-Ni0.15Fe0.05 and P‑Ni0.1Fe0.1. The composition of the exsolved Ni-Fe alloy varies according to the Fe substitution since the increase of the Fe content in the matrix relatively decreases the exsolved Ni amount. Combined with the following TEM analysis, from E-Ni0.15Fe0.05 to E-Ni0.1Fe0.1, the exsolved Ni-Fe alloy composition varied from Ni4Fe1 to Ni3Fe1. In contrast, no exsolved Fe phase is evident for P-Fe0.2 (Fig. 1c). The same trend of transition metal exsolution in the layered perovskite was previously observed, indicating that Ni exsolves more efficiently to the surface than Mn and Fe [48]. The results demonstrate that, although Fe alone hardly exsolves, the existence of Ni in the B site promotes the exsolution of Fe, forming a Ni-rich Ni-Fe alloy nanoparticle.The dynamic exsolution process was evaluated by analyzing the weighted loss profiles (through thermogravimetry, Fig. 2a and b), H2 consumption (through the thermal conductivity detector, Fig. 2c) and SEM (Fig. S3) during the reduction and exsolution processes. For the P-Ni0.2−xFex materials, three peaks are observed at (i) 250–450 °C, attributed to the loss of oxygen in the PrOx plane; (ii) 450–650 °C, attributed to the partial escape of the intensely bonded lattice oxygen atoms from the perovskite [61]; and (iii) above 650 °C, attributed to the reduction of metal cations located at the B site and the exsolved process of Ni or Ni-Fe to the double-layer perovskite surface. A higher Ni content in the perovskite triggered a lower reduction temperature of the first peak due to the higher electronegativity of Ni (1.75) compared to Mn (1.69) [62]. As Ni is more reducible than Fe, from P-Ni0.2 to P-Fe0.2, the reaction peak shifts to a higher temperature, and the hydrogen consumption decreases in sequence with the increase in the Fe.The exsolution of the nanoparticles was observed from the morphology of the P-Ni0.15Fe0.05 material. High-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) and energy-dispersive x-ray spectroscopy (EDS) elemental mapping were performed to confirm the composition of the exsolved nanoparticles and support. The pristine P-Ni0.15Fe0.05 material has a homogeneous porous surface ( Fig. 3a), whereas numerous nanoparticles emerged on the surface after H2 treatment (Fig. 3b). A detail of an exsolved nanoparticle that was partially socketed in the support is shown Fig. 3c. The results shown in Fig. 3d indicate the spherical shape of the Ni-Fe particles anchored to the surface of the double-layer perovskite. The interplanar spacing of the nanoparticles is 2.05 Å, which is consistent with the d‑spacing of the (111) plane of the Ni-Fe alloy phase (Fig. S4) [37]. The particle size analysis of the E-Ni0.15Fe0.05 catalyst indicates an average of 21 ± 5.6 nm. The EDS elemental mappings highlight that only Ni and Fe atoms were found in the emergent nanoparticles. The quantitative analysis of the EDS data operated on more than 30 nanoparticles indicated a molar Ni/Fe ratio of 3.6 ± 1.0. The chemical elements Pr, Ba, Mn, and O are homogeneously distributed across the support. Only traces of Ni atoms remained in the perovskite, whereas a substantial number of Fe atoms were still present, in agreement with the calculations by Kwon et al.[48] The same exsolution phenomenon of Ni-Fe and Ni nanoparticles was also observed for E- Ni0.1Fe0.1 (Figs. S5 and S6) and E-Ni0.2 (Fig. S7), respectively.The local coordination environment and chemical state of the exsolved nanoparticles were further analyzed using x-ray absorption fine structure spectroscopy (XAFS) at the Ni and Fe K-edges of E-Ni0.2 and E-Ni0.15Fe0.05 catalysts. Even though the reduced pellets were handled carefully during the transfer and test process, it could not be excluded that slight surface oxidation could occur. In the normalized Ni K-edge x-ray absorption near-edge structure (XANES) spectra ( Fig. 4a), the Ni K-edge spectra almost fit with the Ni foil reference concerning the energy position and pattern, indicating that Ni is prevailingly reduced to Ni0 in the two catalysts. In addition, the pre-edge peak slightly shifts to higher energy from E-Ni0.2 to E-Ni0.15Fe0.05 (an enlarged region in Fig. 4a), demonstrating the minor increase in the Ni average oxidation state, indicating the formation of the Ni-Fe alloy [56].Moreover, the white line (7131 eV) of the Fe K-edge XANES spectrum of the E-Ni0.15Fe0.05 catalyst is more intense than that of Fe foil in Fig. 4b, indicating that part of the Fe is oxidized [63]. The intensity of the pre-edge peak of E-Ni0.15Fe0.05 is stronger than that of the Fe2O3 reference. Thus, it is deduced that part of the Fe is reduced to metallic Fe0, whereas another part remains oxidized in the double-layer perovskite support. The wavelet transform signals of the Ni-metal bond were observed around 8 Å−1 in the contour plots of the E-Ni0.15Fe0.05, E-Ni0.2, NiO reference, and Ni foil. In contrast, the Ni-O bond signals were absent except for the NiO standard (Fig. 4c and Fig. S8). In addition, Fig. 4d and e exhibit the extended XAFS (EXAFS) spectra in the k-space and the corresponding Fourier transform in the R space at the Ni K-edges. The monometallic E-Ni0.2 reveals the same oscillations as those of the Ni foil, whereas in terms of the bimetallic E-Ni0.15Fe0.05, the changes minor shift to a smaller k-value. The slight k-value shifted for the bimetallic Ni-Fe system is reported by previous work [36]. The EXAFS spectra of both E-Ni0.2 and E‑Ni0.15Fe0.05 have similar characteristics to those of Ni foils, with the Ni-Ni coordination at ∼2.18 Å (Fig. 4e). This result confirms the existence of Ni in the metallic state [64].The linear combination fitting (LCF) method was adopted based on the identifiable features of each reference in the XANES spectra to quantify the distribution of different Ni and Fe oxidation states. The LCF analysis of the Ni K-edge XANES spectrum reveals that about 94.0 % of Ni is reduced to the metallic state, which is almost double the other reported results (58 % Ni exsolved) [48], and only 6.0 % of Ni remains in the oxidation state in E-Ni0.15Fe0.05 (Fig. 4f). However, the distribution of Fe in E‑Ni0.15Fe0.05 is approximately 25.1 % Fe0 and 74.9 % in the Fe oxidation state (Fig. 4g).The x-ray photoelectron (XPS) spectra of the fresh catalyst ( Fig. 5a) indicate that only divalent Ni2+ (∼854.5 eV, 856.6 eV) and satellite peaks (∼860.8 eV) are detected in P‑Ni0.15Fe0.05 [65]. The metallic nickel Ni0 peaks (∼852.3 eV) appeared after reduction (Fig. 5b), accounting for only a small fraction of the Ni elements, most likely due to oxidation during ex situ movement. Similar to Ni, the Fe element exhibits mixed oxidation states consisting of Fe2+ (∼709.6 eV) and Fe3+ (∼710.8 eV) before reduction (Fig. 5d), a small part of which is reduced to Fe0 (706.7 eV) after reduction treatment (Fig. 5e) [66]. The XPS spectra indicate that metallic Ni0 and Fe0 coexist in the E‑Ni0.15Fe0.05 catalyst, further implying the formation of Ni-Fe alloy. The oxygen species consists of lattice oxygen (∼528.5 eV) and adsorbed oxygen species (∼531 eV) as shown in Fig. S9a and b. The peak ratio of the adsorbed/lattice oxygen increases on the E-Ni0.15Fe0.05 surface with respect to that on the P-Ni0.15Fe0.05 surface [67], which is ascribed to the formation of oxygen vacancies along with the Ni-Fe alloy exsolution process [68].Thus, combined with the XRD (Fig. 1), TEM (Fig. 3), and XAFS analysis (Fig. 4), we infer that Fe was partially reduced to the metallic phase and exsolved to the surface of the perovskite matrix, forming an alloy with Ni [59].The catalytic activity and stability of the preceding impregnated (I-) and exsolved (E-) catalysts were evaluated in dry reforming at 800 °C under atmospheric pressure and high pressure (14 bar). The activity of the catalysts is compared based on the apparent reaction rates of CH4 and CO2. The amount of Ni in the normalized activity is obtained by ICP. This term is considered “apparent” because it is derived from any value of conversion, including the ones of the integral reactor (X > 10 %). Although it is not convenient to refer to it as an “intrinsic reaction rate”, it provides us with a parameter that can be used to compare activity across our catalysts and those of the literature.The initial apparent reaction rate of CH4 for exsolved catalysts (E‑Nix, x = 0.1, 0.2, and 0.3) are much higher than those of their impregnated counterparts (I-Nix; Fig. 6a). Moreover, unlike the fast deactivation of the impregnated catalyst, even the most stable I-Ni0.2 catalyst in the impregnated catalysts dropped by 11.5 % within 12 h (Fig. 6a), the E-Nix catalysts remain stable for the CH4 reaction rate throughout 40 h on stream, mirroring the same stability as the H2/CO ratio (Fig. S10). As for the effect of the Ni loading, the apparent reaction rate of E-Nix catalysts decreased as the amount of Ni increases, despite the conversion delivered opposite trend (Fig. S10).Based on the results, E-Ni0.2 is further modified with the dopant of Fe in the B site (E-Ni0.2−xFex (x = 0.05, 0.1, 0.2)). The CH4 apparent reaction rate of the E-Ni0.15Fe0.05 catalyst is slightly lower than that of the E-Ni0.2 (Fig. 6a and b). The CH4 apparent reaction rate of the E-Ni0.1Fe0.1 catalyst is not stable during 40 h on stream. For E-Fe0.2, its catalytic performance results show negligible CH4 and CO2 apparent reaction rate, indicating that both the perovskite substrate and Fe nanoparticles are not active site for the conversion of CH4 (Figs. 6b and S11), in agreement with the previous reports [69]. The CO2 conversions are slightly higher than the corresponding CH4 conversions for all catalysts, independent of the metal loading and the Ni/Fe ratio ( Figs. 6 and 7), probably originating from the reverse water-gas shift reaction (RWGS) [56,59]. RWGS usually occurs as a side reaction in DRM, which results in higher CO2 conversion than CH4 conversion and makes the H2/CO ratio lower than 1 since the reaction produces more CO and consumes H2 [70]. The RWGS is observed in our catalysts in the results presented in Figs. S10–15 and corroborated by previous works [32,70]. Besides, compared to the exsolved catalyst, the unreduced E-Ni0.15Fe0.05 catalyst without the generation of exsolved metal species synthesized by the same method exhibited relatively low activity and deactivated fast within 15 h (Fig. S12). Similarly, the E-Ni0.15Fe0.05 catalyst reduced under pure H2 at 800 °C also displayed low performance, probably due to the perovskite decomposition under pure H2 (Fig. S13).The long-term stability test for the E-Ni0.15Fe0.05 and E-Ni0.2 catalysts is presented in Fig. 7a and Fig. S14. After around 135 h on stream, the stability test of E-Ni0.2 ceases because the catalyst bed is congested with coke (the pressure drop rises and the flow rate decreases). The E-Ni0.15Fe0.05 catalyst displays a stable apparent reaction rate with no noticeable deactivation for 260 h on stream at 800 °C. This result indicates that Fe plays a critical role in stabilizing the catalyst. Besides, the E-Ni0.15Fe0.05 catalyst shows 100 h stability under undilute gas conditions with CH4:CO2 = 1:1 as feed gas without deactivation (Fig. S15). Industrial syngas must be compressed, such as compression from 1 to 10 bars, for utilization, which costs more than 85 % of the total capital investment and 60 % of the operational costs [71]. Additionally, the high-pressure operation increases the production capacity, as well. However, coking is highly favored in high-pressure conditions, which is the central dilemma to address. Both E-Ni0.15Fe0.05 and E-Ni0.2 exhibited 40 h stability in reaction conditions at 14 bar (Fig. 7b). Compared to a previous study [71], E-Ni0.15Fe0.05 exhibits similar CH4 conversion, lower CO2 conversion, and a much higher H2/CO ratio (Fig. S16), even without 10 % H2O feeding, indicating that E-Ni0.15Fe0.05 is also a promising catalyst for the high-pressure dry reforming of CH4.To elucidate the role of Fe on the carbon resistance of Ni-based catalysts under reaction conditions, thermogravimetric techniques were applied to measure the carbon deposition on the used catalyst. The CO2 signal was analyzed in a mass spectrometer during combustion. As depicted in Fig. 7c, the apparent coke formation rate on the E-Ni0.2 catalyst is 2.57·10–5 mmol gcat −1 s−1 during the 135 h on stream, whereas the apparent coke formation rate significantly drops to 4.71·10–8 mmol gcat −1 s−1 for the E-Ni0.15Fe0.05 catalyst during the 260 h on stream. Compared with approximately 140 catalysts from the state-of-the-art references, the proposed E-Ni0.15Fe0.05 catalyst has the slowest apparent coke formation rates of the relatively extensive catalyst portfolio (Fig. 7e and Table S3). Many studies in the literature do not assess coke fouling on the catalysts for the following reasons: (i) insufficient time on stream, (ii) unrealistically mild reaction conditions, or (iii) inappropriate excess of catalyst and the subsequent underestimation of coke formation. Per the thermogravimetric results, contrary to the severe coking on the used E-Ni0.2 catalyst, the Raman spectra indicated no coke accumulation on the used E-Ni0.15Fe0.05 catalyst (Fig. S17). The minor coking deposition convincingly demonstrates the coking-resistant effect of Fe, coinciding with previous reports that Fe substitution improves the coking resistance of Ni-based catalysts in reaction conditions [72]. In addition, the used catalyst after high-pressure conditions at 14 bar is also characterized and shown in Fig. S18. Generally, the high pressure favored the coking formation. The weight loss of the used E-Ni0.15Fe0.05 is only 2 % after 40 h on stream. Compared to the used E-Ni0.2, the D and G band intensity of the used E-Ni0.15Fe0.05 is lower, further indicating an improvement of coking resistance with the introduction Fe even at high pressure.The particle size distribution analysis revealed that the used E-Ni0.15Fe0.05 catalyst possessed exsolved Ni-Fe nanoparticles with an average size of 30.1 nm after 260 h on stream, slightly larger than the nanoparticles of the E-Ni0.15Fe0.05 before the reaction (21.0 nm; Fig. 7d). The exsolved nanoparticles of the used E-Ni0.15Fe0.05 catalyst anchor partially in the support without any observable coke with a morphology similar to the pristine one (Fig. S19a and c). However, the counterpart used I-Ni0.2 catalyst exhibited severe sintering of Ni particles from 35.2 to 56.6 nm only after 12 h on stream with apparent filamentous coke on the catalyst surface (Fig. 7d, S19b and d). Hence, the exsolved E-Ni0.15Fe0.05 catalyst exhibited improved sintering resistance compared with the I-Ni0.2 catalyst, which is closely associated with their stability (Figs. 6 and 7).Structural changes in the E-Ni0.15Fe0.05 catalyst between the reduced state and after the reaction were further investigated using STEM coupled with electron energy loss spectroscopy (EELS). Spatially resolved EELS spectra at the L-edges were used to analyze the metal oxidation state in the core and particle surfaces. For 3d metals, a typical L-edge EELS spectrum includes a pair of strong white lines corresponding to 2p3/2 → 3d (L3-edge) and 2p1/2 → 3d (L2–edge) transitions and two edge jumps corresponding to 2p → continuum transitions. The two white lines are separated by the spin–orbit interaction of the 2p core states. A one-electron excitation theory usually fails to interpret the spectral fingerprint (e.g., branching ratio and multiple interactions) because the 2p- and 3d-hole have radial wave functions overlapping significantly [73].In this work, we restrict the analysis of the L2,3-edges by considering only the total intensity of the white lines. Previous authors have demonstrated that the total number of 3d holes is proportional to the integrated L2,3-edge peaks (see experimental method section), which is a useful feature to determine the metal oxidation state [74–77]. Fig. 8 presents the Ni and Fe L-edge spectra of the metal and metal monoxide standards, with spectra taken in the core and shell of the Ni-Fe nanoparticles before and after the catalytic reaction. The comparison of the total intensities of the white lines with the Fe and Ni standards (I(Fe) = 21 and I(Ni) = 10) indicated that the nanoparticle core remains in a metallic state throughout the reaction (I(Fe) = 20 and I(Ni) = 10–11). This result is in agreement with the XPS spectra of the used catalyst (Fig. 5c and f). For the oxidized shell of the used catalyst, the Ni atoms were in a mixture of Ni0 and Ni2+ states (I(Ni) = 16 vs. I(Ni) = 23 in NiO), whereas the Fe atoms were predominantly in the Fe2+ state with a probable minor presence of the Fe3+ state (I(Fe) = 32 vs. I(Fe) = 29 in FeO).The same observations were made for the oxidized shell of the reduced catalyst. Thus, the presence of the latter oxide shell on metal nanoparticles was essentially due to handling the catalyst in the air prior to the TEM analysis. To eliminate these effects, the quasi in-situ TEM is performed in reaction conditions and displayed in Fig. S20. This result shows the partial redistribution of Fe and a relatively low concentration of oxygen layer surrounding the Ni-Fe nanoparticles in the E-Ni0.15Fe0.05 catalyst. Thus, we can consider that as the initial status of the catalyst (before the reaction). In contrast, FeOx species are formed on the outer layer of the Ni-Fe alloy nanoparticles during the dry reforming conditions [78]. Besides, the average Ni/Fe molar ratio of 3.6 significantly increased to 9.2 after the reaction with a much larger standard deviation (from 1.0 to 5.3, measured on 38 particles). This evidence that part of the metallic iron was redistributed on the support during the reaction agrees with previous observations of Coperet et al. [63].The DFT calculations were performed to compare the adsorption energy of the key intermediates on Ni4Fe1 (111) models to that on monometallic Ni (111) models to elucidate the improved coking resistance of the Ni-Fe bimetallic catalyst at the atomic resolution level. These surfaces were considered as they were identified experimentally using XRD (Fig. 1c), high-resolution TEM (Figs. S4 and S7), EDX analysis (Fig. 3d), EELS analysis (Fig. 8). Even though the oxygen vacancy defects in the perovskite matrix will assist increasing ratios of singlet oxygen species on the surfaces for removing carbon species, the TGA and TPR results in a reduction atmosphere (as shown in Fig. 2) indicated that the amount of the oxygen vacancy originating from metal exsolution of P-Ni0.2 is slightly larger than that of P-Ni0.15Fe0.05. However, the coking resistance of E-Ni0.15Fe0.05 is much better than that of E-Ni0.2, indicating that in these catalysts, the oxygen vacancy facilities carbon removal, but it is not the main factor. Therefore, the effect of the perovskite support is not considered in the DFT calculations, which focused on elucidating the improvement of the coking resistance of E-Ni0.15Fe0.05. Thus, only the monometallic Ni and bimetallic Ni-Fe alloy were considered for the slab model (Figs. S21-S23). By calculating the effective barriers for the C and CH oxidation pathways on Ni and Ni4Fe1 (111) surface (Fig. S24), it is considered that the O* originating from the CO2 dissociation directly oxidizes the intermediate CH* is the dominant oxidation pathway on both Ni and Ni4Fe1 (111) surfaces, which is consistent with other reports [36,81]. Although the slab model cannot conclude the fine structure of the Ni-Fe binary alloy, a reasonable understanding of the reaction mechanism was gained throughout the DFT.The binding energies of the critical intermediates on Ni4Fe1 (111) were compared with those on Ni (111), as presented in Table S4. Despite the same binding energy of CH4 between Ni (111) and Ni4Fe1 (111), the binding strength of carbon-containing intermediates, including CH3 * , CH2 * , CH* , C* , and CO* , are all weaker on the Ni4Fe1 (111) surface than that on the Ni (111) surface, on which coke is likely prone to form due to the higher C* binding energy [82]. In addition, due to the stronger Fe-O bond, the adsorption of the oxygen-containing species, such as O* and CHO* , on the Ni4Fe1 (111) surface is stronger than that on the Ni (111) surface, in agreement with the time-resolution DRIFTS spectra under switching gas conditions (Figs. S25 and S26) and previously reported results [36]. The O* and CHO* adsorption strengths on Ni4Fe1 (111) surface are 0.15 and 0.08 eV higher than the pure Ni (111) surface. To observe the effects of Fe on the catalyst, the chemisorption energy of O, CO and C on pure Ni (111), Ni4Fe1 (111) and pure Fe (111) are plotted in Fig. 9a, b and c, respectively. The adsorption energy of O increases along with the increase of Fe composition, implying the oxophilic nature of Fe (Fig. 9a), consistent with the O2-TPD results (Fig. S27). In contrast, the adsorption energy of CO has no obvious correlation with the Fe composition (Fig. 9b). These trends indicate that the difference of CO2 dissociation energy for various catalysts mainly originates from the adsorption capability of O* species, instead of that of CO* species (Fig. 9a and b). Besides, as shown in Fig. 9c, the adsorption energy of C closely connected with the Ni composition on the active site, further implying that coke prefers to form on pure Ni site other than Fe site.In DFT calculations, we are mainly focusing on the coke formation reactions. The energy barriers of some key elementary steps of the reaction on Ni (111) and Ni4Fe1 (111) surfaces are displayed in Figs. S24 and S28 and Table S5. The dissociative adsorption energies of CH4 and CO2 are presented in Fig. 9d and e. The CH4 dissociation energy of Ni4Fe1 is 0.04 eV higher than that of Ni (111), leading to less CHx * (x = 0, 1, 2, 3) intermediate species on the Ni4Fe1 (111) sites. Moreover, the CO2 dissociation energy of Ni4Fe1 (111) is 0.06 eV lower than that of pure Ni (111). Due to the oxophilic nature of Fe, adding Fe to Ni-based catalyst enhances the adsorption of O* species and reduces the CO2 dissociation energy (exhibiting a slightly inferior CH4 dissociation energy compared to Ni). The higher concertation of O* species on the surface of the Ni-Fe binary alloy catalyst reacts with C* species and lower the coking rate, contributing to the atypical coking resistance of this catalyst.The direct exsolution of Ni and Ni-Fe in a single reduction step leads to well dispersed, anchored and alloyed (in the case of the bimetallic sample) nanoparticles on PrBaMn1.6Ni0.4–2xFe2xO5+δ double-layer perovskite. We synthesized these catalysts together with counterparts prepared by impregnation, characterized and tested them in the dry reforming of methane. The exsolved Ni-based catalyst has a significantly superior performance and longer stability due to enhanced metal-support interaction. The exsolved Ni-Fe alloy catalyst shows slightly slower reaction rates but a significantly longer lifetime: with negligible coke depositions at 800 °C during 260 h on stream under 1 bar or 40 h on stream under 14 bar (more relevant for industrial implementation). Our main objective has been to understand the reasons behind the higher stability of this Ni-Fe catalyst by characterization, ab initio calculations, and dry reforming reactions. Our results show that Fe (in the exsolved Ni-Fe catalyst) stabilizes O* species, helps in the CO2 dissociation, and facilitates the reactions of C* species as its adsorption is weakened. At the same time, the stronger metal-support interaction in this catalyst leads to slower sintering. These combined effects are the reasons behind the atypical more extended stability of the exsolved Ni-Fe alloy catalyst. Xueli Yao: Conceptualization, Investigation, Methodology, Data curation, Writing – original draft, Writing – review & editing, Visualization. Qingpeng Cheng: Conceptualization, Methodology, Data curation, Writing – review & editing. Yerrayya Attada: DFT calculations and Formal analysis. Samy Ould-Chikh: Data curation, Formal analysis and Writing – review. Adrian Ramírez: Data curation and Formal analysis. Xueqin Bai: Investigation. Hend Omar Mohamed: Formal analysis. Guanxing Li: Data curation. Genrikh Shterk: Formal analysis. Lirong Zheng: Data curation and Formal analysis. Jorge Gascon: Formal analysis and review. Yu Han: Formal analysis and review. Osman M. Bakr: Formal analysis and review. Pedro Castaño: Funding acquisition, Project administration, Resources, Supervision, Formal analysis, Writing – review & editing.The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Pedro Castano reports financial support was provided by King Abdullah University of Science and Technology. Pedro Castano has patent pending to King Abdullah University of Science and Technology (KAUST).This work was conducted thanks to the financial support of the King Abdullah University of Science and Technology (KAUST, BAS/1/1403).Supplementary data associated with this article can be found in the online version at doi:10.1016/j.apcatb.2023.122479. Supplementary material .

Dry reforming of methane simultaneously achieves several sustainability goals: valorizing methane-activating carbon dioxide while producing syngas. The catalyst has an enormous influence on the process viability by controlling activity, selectivity, and stability. A catalyst with uniform-sized Ni-Fe alloy nanoparticles anchored into PrBaMn1.6Ni0.3Fe0.1O5+δ double-layered perovskite is assembled via a facile one-step reduction strategy. Our method attains more exsolved Ni nanoparticles (94 %) than the common conditions. The exsolved Ni0.15Fe0.05 catalyst shows exceptional stability in 260 h tests at 800 °C, with one of the slowest coke formation rates compared with the state-of-the-art catalysts. Besides, no deactivation was observed during 40 h operation at more demanding and coking conditions (14 bar) where this process is more likely to operate industrially. Via experimental characterizations and computational calculations, the stability of the robust exsolved Ni-Fe catalyst is demonstrated by its unique balance of adsorbed species, which inhibits coking.

Ethylene (C2H4) is regarded as the most important petrochemical platform molecule to produce diverse commodity chemicals such as polyethylene, ethylene oxide, vinyl chloride, and polystyrene, with global demand of 153 million tons in 2016 and net added demand of about 5.2 million tons every year (Xu, 2017). Nowadays, its universal production in industry is based on the steam cracking of oil-based naphtha. However, the oil resource is increasingly dwindling, and thus it has turned out to be a hotspot in modern industries to pave the way for efficient and ecofriendly utilization of the nonoil resources (e.g., natural gas, coal, and renewable biomass) to produce ethylene with the aid of effective catalytic processes. Ethane (C2H6) is abundant in natural gas, and in particular, the shale gas revolution in recent years greatly enriches ethane resources (Sattler et al., 2014). Therefore, ethane-to-ethylene conversion (in terms of oxidative dehydrogenation of ethane [ODE], catalytic dehydrogenation, and steam cracking) tantalizes global enthusiasm. The latter two suffer from their thermodynamic constraints and high operation temperature (>700°C), and the ODE is thus more competitive, benefitting from its oxidative feature that can cast off the thermodynamic limitation and allow lower operation temperature (350°C–550°C) (Heynderickx et al., 2005).However, controlling the ethylene selectivity for ODE reaction represents the grandest challenge because the excessive oxidation of ethylene to carbon dioxide is thermodynamically and kinetically favorable. Therefore, developing a qualified catalyst with high activity and selectivity is the goal of most efforts for this reaction. To date, various catalysts have been explored (such as alkaline-/rare-earth metal oxides, Mulla et al., 2001, Gaab et al., 2003; noble metals, Fu et al., 2013; and transition metal oxides, Liu et al., 2003; Nakamura et al., 2006), and NiO-based catalysts are the most attractive owing to its low operation temperature, simple preparation, and low cost (Heracleous and Lemonidou, 2006, 2010; Savova et al., 2010; Zhu et al., 2012). However, NiO alone mainly yields carbon dioxide due to the large amount of electrophilic (unselective) oxygen species (Heracleous and Lemonidou, 2006, 2010; Savova et al., 2010; Zhu et al., 2012). Many kinds of oxides were doped into NiO to tune the oxidative properties of oxygen species. Lemonidou et al. explored a series of alter-valent cations such as Li, Mg, Al, Ga, Ti, Ta, and Nb (Heracleous and Lemonidou, 2010), and the unselective oxygen amount on NiO surface declines along with the increase in dopant cations' valence. Accordingly, the Nb2O5-doped catalyst offers the highest ODE performance such as 78% ethylene selectivity and 33% ethane conversion at 350°C (Savova et al., 2010). They further proposed that Nb doping into NiO lattice by filling the cationic vacancies on defective non-stoichiometric NiO surface and/or substituting Ni atoms reduces the amount of unselective oxygen (Zhu et al., 2012; Heracleous and Lemonidou, 2006). However, such Nb2O5-NiO catalysts suffer from poor stability due to their sintering deactivation (Heracleous and Lemonidou, 2006, 2010; Savova et al., 2010; Zhu et al., 2012).Despite the above-mentioned interesting advances, the real-world use of these catalysts still remains a challenge as their poor thermal conductivity is detrimental to rapid dissipation of reaction heat released in this strongly exothermic ODE reaction (ΔH = −104 kJ mol−1), which causes severe hotspots in the catalyst bed and therefore leads to the ethylene excessive oxidation while releasing more heat. Recently, the development of structured catalyst based on the monolithic metal-foam has been attracting great interest in heterogeneous catalysis because of the intensified heat transfer, which is favorable to tailor catalysts for strongly exothermic reactions (Chen et al., 2019; Zhao et al., 2016; Zhang et al., 2018a, 2018b). However, the main issue is how to make these promising metal-foam qualified catalysts, or more concretely, how to fabricate the highly active and selective NiO-based nanocomposites onto foam surface.Herein, we demonstrate the remarkable improvement of the Nb2O5-NiO/Ni-foam catalyst performance for ODE reaction, by finely tuning the Nb2O5-NiO interaction by morphology-controllable growth of NiO-precursors onto Ni-foam.First, three kinds of NiO-precursors with different morphologies (i.e., clump for Ni(OH)2, rod for NiC2O4, nanosheet for nickel terephthalate (Ni-Tp), identified by X-ray diffracxtion [XRD] in Figure S1) were controllably and endogenously grown onto a Ni-foam (100 pores per inch). Against the smooth surface of Ni-foam (Figures 1A–1C), clearly, the in situ growth of three morphology-different NiO-precursor layers on the foam struts succeeds clump with dense stacking for Ni(OH)2 layer by ammonia evaporation method (Figures 1D, 1G, and 1J), rod with diameter of about 450 nm for NiC2O4 layer by hydrothermal method (Figures 1E, 1H, and 1K), and nanosheet of thickness 30 nm for Ni-Tp layer by solvothermal method (Figures 1F, 1I, and 1L). Moreover, unlike the dense layer feature of the Ni(OH)2 clump and NiC2O4 rod, the Ni-Tp nanosheets stand upright and irregularly cross-link each other to form honeycomb-like porous layer. Not surprisingly, the Ni-Tp/Ni-foam delivers a specific surface area (SSA) of 6.3 m2 g−1 much higher than 1–2 m2 g−1 for the Ni(OH)2/Ni-foam and NiC2O4/Ni-foam (Table 1 ).Subsequently, niobium ammonium oxalate was wet-impregnated onto the above-obtained Ni(OH)2/Ni-foam, NiC2O4/Ni-foam, and Ni-Tp/Ni-foam at a Nb2O5 content of 5 wt. % (including the Ni-foam mass), followed by drying overnight and calcining in air at 450°C, to form Ni-foam-structured Nb2O5-NiO catalysts (Figures 1M–1U). These catalysts are denoted as Nb2O5-NiO/Ni-foam-C (clump), Nb2O5-NiO/Ni-foam-R (rod), and Nb2O5-NiO/Ni-foam-NS (nanosheet), which all possess equivalent NiO content (∼21 wt. %, including Ni-foam mass; Table 1). The NiO and Ni (from Ni-foam) phases are clearly detected by XRD for all three catalysts, whereas no Nb2O5 diffraction peaks are observed, indicating its high dispersion or amorphous structure (Figure S2) (Liu et al., 2016). Notably, the Ni(OH)2-, NiC2O4-, and Ni-Tp-derived nano-NiO aggregations show well-preserved clump-, rod- and nanosheet-morphologies regardless of Nb2O5 introduction (Figures 1M–1O). In addition, the Nb2O5-NiO ensembles show porous feature in association with the thermolysis of their precursors (Figures 1P–1R) thereby leading to a visible increase in their SSA (Table 1).Interestingly, the Nb2O5-NiO/Ni-foam-NS achieves an SSA of 20.8 m2 g−1, much higher than 12–13 m2 g−1 seen with the other two catalysts (Table 1). The enhanced surface area can be related to the fact that the nanosheet-like morphology of Ni-Tp/Ni-foam not only favors the formation of catalyst with high SSA (see NiO/Ni-foam-NS, Table 1) but also is helpful for highly dispersing Nb2O5-precursor onto the Ni-Tp nanosheet to hinder the crystallization of NiO during the calcination process (Solsona et al., 2011, 2012) (Table 1). Not surprisingly, the Nb2O5-NiO/Ni-foam-NS catalyst provides an average NiO size of 13.5 nm, smaller than that of ∼20 nm for the Nb2O5-NiO/Ni-foam-C and Nb2O5-NiO/Ni-foam-R (Table 1). Nevertheless, the NiO/Ni-foam-NS obtained by calcining the Ni-Tp/Ni-foam in air at 450°C offers an average NiO size of ∼20 nm, being compatible to that seen with the ones derived from Ni(OH)2/Ni-foam and NiC2O4/Ni-foam. This observation reveals that Nb2O5 introduction favors the decomposition of Ni-Tp nanosheets, rather than Ni(OH)2-clump and NiC2O4-rod, to form smaller NiO nanoparticles. Moreover, the Nb2O5-NiO/Ni-foam-NS achieves more homogeneous NiO-Nb2O5 composites than the other two catalysts (Figures 1S–1U).The Nb2O5 modification dramatically improves the ethylene selectivity and slightly the ethane conversion while leading to a remarkable increase in the turnover frequency (TOF) for ethylene formation from ∼0.62 h−1 for the Nb2O5-free samples to 0.91–0.96 h−1 at 300°C (Table 1 and Table S1; the detailed calculation method in the Supplemental Information). As shown in Figure 2 , three Nb2O5-free samples all achieve almost identical ethane conversion and ethylene selectivity in the whole temperature range studied. In contrast, the Nb2O5-NiO/Ni-foam catalysts exhibit different ODE performance under identical reaction conditions, showing the NiO-precursor morphology dependence; the Nb2O5-NiO/Ni-foam-NS is obviously superior to the Nb2O5-NiO/Ni-foam-C and Nb2O5-NiO/Ni-foam-R catalysts (Figure 2), achieving a 58.4% ethane conversion and 75.4% ethylene selectivity at 425°C. In addition, compared with the very low productivity of only 0.18 gethylene gcat. −1 h−1 over the Nb2O5-free NiO/Ni-foam catalysts, Nb2O5 modification gets the ethylene productivity doubled even more. The Nb2O5-NiO/Ni-foam-NS achieves the highest ethylene productivity of 0.46 gethylene gcat. −1 h−1 (Figure S3).To reveal the underlying origin of the NiO-precursor morphology-dependent ODE catalysis on the above Nb2O5-NiO/Ni-foam catalysts, the amount and type of oxygen species were collaboratively probed by H2-temperature-programmed reduction (H2-TPR) and O2-temperature-programmed desorption (O2-TPD) (Zhu et al., 2012; Zhang et al., 2018a, 2018b). Clearly, whereas the Nb2O5-free NiO/Ni-foam samples show quite different NiO morphologies (Figure S4), they all possess identical reducibility (by H2-TPR) and properties of surface oxygen species (by O2-TPD), solidly evidenced by their almost same H2-TPR and O2-TPD profiles (shape, peak area, and peak temperature; Figures 3A and 3B, profiles 1–3). It is thus not surprising that they achieve NiO-precursor morphology-independent ODE performance (Figure 2). In combining this information with the observation of NiO-precursor morphology dependences of distinct ODE performance after Nb2O5 modification, it is safe to say that the NiO-Nb2O5 interaction is sensitive to NiO-precursor morphology, which in nature is responsible for the distinct ODE performance for the Nb2O5-NiO/Ni-foam catalysts.Indeed, the reducibility and properties of the surface oxygen species of the Nb2O5-NiO/Ni-foam catalysts show strong NiO-precursor morphology dependence (Figures 3A and 3B, profiles 4–6). The Nb2O5-NiO/Ni-foam-C offers a single H2-TPR peak at 340°C with an 8°C delay compared with the NiO/Ni-foam, likely due to the weak Nb2O5-NiO interaction. The Nb2O5-NiO/Ni-foam-R delivers a main peak at 332°C and a weak shoulder at 358°C, suggesting the very limited local occurrence of moderate NiO-Nb2O5 interaction. In contrast, the Nb2O5-NiO/Ni-foam-NS provides a main peak at 371°C and a very weak one at only 297°C. It should be noted that the H2 consumption is attributed exclusively to the NiO reduction because Nb2O5 reduction cannot occur under such conditions (Zhang et al., 2018a, 2018b). Particularly, the NiO size of the Nb2O5-NiO/Ni-foam-NS is 13.5 nm, smaller than 20 nm for the others. In general, the lattice oxygen of the smaller NiO nanocrystallites diffuses more efficiently than the larger ones (Zhu et al., 2012). So, the weak peak at 297°C is assignable to the small NiO species that interacted weakly with Nb2O5, whereas the main peak at 371°C is ascribable to the comprehensive occurrence of strong NiO-Nb2O5 interaction.All catalysts with and without Nb2O5 modification deliver dual-peak O2-TPD profiles, in which the peak at 342°C is assigned to O2 - and the one at 543°C is assigned to O− - (Figure 3B) (Wu et al., 2012; Iwamoto et al., 1976). The O2 - species have strong oxidizing electrophilicity and thus are considered to be non-selective oxygen species that favor the deep oxidation of product (Wu et al., 2012; Iwamoto et al., 1976). The amount and desorption behavior of O2 - and O− - species are tuned markedly by Nb2O5 modification, showing clear NiO-precursor morphology dependence (Table S2 and Figure 3B). The desorbability of such two types of surface oxygen species is almost unchanged for the Nb2O5-NiO/Ni-foam-C and Nb2O5-NiO/Ni-foam-R, whereas their non-selective O2 - amounts are markedly reduced in association with a slight decline of the O− amount, when compared with the Nb2O5-free samples (Table S2 and Figure 3B). For Nb2O5-NiO/Ni-foam-NS, most notably, the Nb2O5 modification makes the non-selective O2 - species almost disappear, but slightly decreases the selective O− - species, whereas lowers the desorption temperature of O− - species to 520°C by 23°C (Table S2 and Figure 3B). According to the Mars van Krevelen mechanism (Figure S5) (Zhu et al., 2012), the types of oxygen species determines the further reaction of ethyl radical to form ethylene (β-elimination) or CO2 (C-C bond cleavage). It is not surprising that Nb2O5 modification and thinning the NiO-precursor thickness are inclined to reduce the non-selective O2 - species amount and form the ethylene via β-elimination.In nature, tuning the NiO-precursor morphology from dense Ni(OH)2 clump and NiC2O4 rod (450 nm) to Ni-Tp nanosheet (30 nm thickness) strengthens the NiO-Nb2O5 interaction thereby leading to almost elimination of the non-selective O2 - species and meanwhile improving the mobility of the highly selective O− - species. Improved mobility of the O− - species (lowered desorption temperature, Figure 3B) (Wu et al., 2012; Skoufa et al., 2014) makes it more active than the other two catalysts, which in turn compensates the activity loss caused by the reduction of non-selective O2 - and selective O− - species (Zhu et al., 2016). That is the reason why the Nb2O5-NiO/Ni-foam-NS catalyst always achieves higher conversion especially above 375°C (Figure 2A).To further gain insight into the O2 - reduction caused by Nb2O5-NiO interaction, the surfaces of the NiO/Ni-foam and Nb2O5-NiO/Ni-foam catalysts were probed by X-ray photoelectron spectroscopy (XPS). Figure 3C shows the Ni2p spectra of the catalyst samples. Three peaks are detected: main peak at binding energy (BE) of 853.8 eV for Ni2+ in NiO; satellite peak at 855.8 eV S(I) for Ni3+ in Ni2O3, Ni2+-OH species, and Ni2+ vacancies; and the other satellite peak at 861.3 S(II), involving a ligand-metal charge transfer (Salagre et al., 1996; Veenendaal and Sawatzky, 1993). The intensity ratio of S(I) to the main peak at 853.8 eV has been used to present the surface and/or structural density of defect sites (Solsona et al., 2012; Zhu et al., 2015), offering the information about the non-stoichiometric (or non-selective) property of NiO. Notably, this ratio declines from 4.0 for the NiO/Ni-foam-NS to 1.9 for the Nb2O5-NiO/Ni-foam-C, to 1.7 for the Nb2O5-NiO/Ni-foam-R, and further to 1.1 for the Nb2O5-NiO/Ni-foam-NS (Table S3). Clearly, Nb2O5 modification provides the ability to markedly reduce the non-stoichiometric Ni3+ (responsible for the non-selective O2 - species), whereas the nanosheet NiO-precursor morphology synergistically promoted such Nb2O5 modification effect. This observation is in good agreement with the O2-TPD results (Figure 3B). Figure 3D shows the XPS spectra in Nb3d region for the Nb2O5-NiO/Ni-foam catalysts. Taking the Nb5+ in pure Nb2O5 (207.4 eV) as reference (Liu et al., 2016), the BE of Nb5+ shifts to 207.2 eV for the Nb2O5-NiO/Ni-foam-C, 207.1 eV for the Nb2O5-NiO/Ni-foam-R, and then 206.9 eV for the Nb2O5-NiO/Ni-foam-NS. This trend is consistent with the increasingly stronger NiO-Nb2O5 interaction (Zhu et al., 2012).As aforementioned, the nanosheet Ni-Tp precursor is much thinner than Ni(OH)2 clump and NiC2O4 rod and is irregularly aligned to form a porous layer (Figures 1F, 1I, and 1L). This morphology undoubtedly gives higher SSA, which is helpful for highly dispersing Nb2O5 into the NiO matrix (Figures 1S, 1T, and 1U), leading to the lower Ni/Nb ratio in catalyst surface (Table S3); furthermore, as indicated by the high-angle annular dark-field scanning transmission electron microscopy images and elemental maps in Figures 4A–4F, the Nb2O5-NiO/Ni-foam-NS achieves the contacting of NiO with Nb2O5 more sufficient than the Nb2O5-NiO/Ni-foam-C and Nb2O5-NiO/Ni-foam-R. On the other hand, the thinner nanosheet feature of Ni-Tp facilitates the incorporation of Nb ions into NiO during calcination treatment. Indeed, the lattice constant obtained by XRD (Solsona et al., 2012) reveals that the NiO lattice constant in the Nb2O5-NiO/Ni-foam-NS is 4.1724 Å, smaller than 4.1767 Å for the NiO/Ni-foam and 4.1752 Å for both the Nb2O5-NiO/Ni-foam-C and Nb2O5-NiO/Ni-foam-R (Table 1). This observation evidences that Nb ions are, at least partially, incorporated into NiO to the most extent for the Nb2O5-NiO/Ni-foam-NS (Solsona et al., 2012; Zhu et al., 2012).Last but not the least, according to the foregoing findings, we are confident that the ODE performance of Nb2O5-NiO/Ni-foam catalyst can be improved further if the NiO-precursor nanosheet is able to be thinned further. Indeed, the Ni(OH)2 nanosheet (∼20 nm) is successfully structured onto the Ni-foam by hydrothermal treatment in an aqueous solution of NH4F (denoted as Ni(OH)2/Ni-foam-F, Figures 5 A–5C and S6), and therefore, a Nb2O5-NiO/Ni-foam-F catalyst was obtained by subsequent Nb2O5 modification. As expected, such catalyst shows much higher activity and selectivity than the Nb2O5-NiO/Ni-foam-C; when compared with the Nb2O5-NiO/Ni-foam-NS it achieves comparable activity but markedly improved selectivity (Figure S7). Notably, our Nb2O5-NiO/Ni-foam-F catalyst yields better performance (especially the selectivity, stability, and TOF) than the NiO-based catalysts (Tables 2 and S4) and powdered Nb2O5/NiO (5/21, w/w) catalyst literature (Table S5). Moreover, the ethylene yield (ethane conversion times ethylene selectivity) for such catalyst is comparable to the costly MoVTeNbO catalyst when it is tested at 2,120 cm3 g−1 h−1, but our catalyst runs at much higher reactor capacity (GHSV) of 9,000 cm3 g−1 h−1 (Table 2).In addition, it is not surprising that the Nb2O5-NiO/Ni-foam-F exhibits highly enhanced Nb2O5-NiO interaction (Figures 5D–5F) by further thinning the NiO-precursor, which results in a further reduction of the NiO lattice constant (Table 1), the NiO nanoparticle size (Table 1 and Figure S6), and especially the non-selective O2 - amount as well as the NiO reducibility (Figure S8), compared with the ones using Ni(OH)2/Ni-foam-C (dense clump of Ni(OH)2) and Ni-Tp/Ni-foam-NS (∼30 nm Ni-Tp nanosheet). This is undoubtedly responsible for the further catalytic performance improvement observed on the Nb2O5-NiO/Ni-foam-F catalyst. Most notably, this catalyst exhibits favorable stability, being stable for at least 240 h at 400°C with ∼44% ethane conversion and ∼82% ethylene selectivity (Figure 5G), which shows great superiority when compared with the previously reported Nb2O5-NiO catalysts (Table S4). This is benefited from the high Nb2O5-NiO sintering resistance (evidenced by the well-preserved SSA and particle size of NiO for the used catalyst, Table 1), as a result of the strong interaction between NiO and Nb2O5 (Solsona et al., 2011, 2012) in combination with the enhanced heat transfer of the Ni-foam-structured designing that could rapidly dissipate the large quantity of reaction heat from the ODE reaction (Table S5) (Li et al., 2015; Zhao et al., 2016; Zhang et al., 2018a, 2018b).In summary, a low-temperature active, highly selective, and highly stable Nb2O5-NiO/Ni-foam catalyst has been developed for the ODE reaction, by carefully tuning the NiO-precursor morphology-dependent Nb2O5-NiO interaction. The Nb2O5-NiO interaction can be markedly improved by thinning the NiO-precursors endogenously grown onto the Ni-foam substrate, especially leading to significant elimination of the nonselective O2 - species and, meanwhile, remarkable improvement of the mobility of selective O− species. This work provides an interesting clue to tailor high-performance ODE catalyst via morphology modulation strategy.The ammonium niobium oxalate is a little bit costly.All methods can be found in the accompanying Transparent Methods supplemental file.We acknowledge the financial supports from the Key Basic Research Project (grant 18JC1412100) form the Shanghai Municipal Science and Technology Commission, the National Natural Science Foundation of China (grants 21773069, 21703069, 21703137, 21473057, U1462129, 21273075), and the National Key Basic Research Program (grant 2011CB201403) from the Ministry of Science and Technology of the People's Republic of China.Y. Lu, Z.Z., and G.Z. conceived the idea for the project and designed the experiments. Z.Z., G.Z., Y. Liu, and Y. Lu carried out the interpretation and wrote the manuscript. Z.Z. conducted the material synthesis, characterizations, and catalytic tests. W.S. drew the structure of Ni-foam in Figure 1. All authors discussed and commented on the manuscript. Y. Lu directed the research.Y. Lu, Z.Z., G.Z., and Y. Liu have a patent application related to this work filed with the Chinese Patent Office on October 15, 2017 (201710956118.5). The authors declare that they have no competing interests.Supplemental Information can be found online at https://doi.org/10.1016/j.isci.2019.09.021. Document S1. Transparent Methods, Figures S1–S8, and Tables S1–S5

Large-scale shale gas exploitation greatly enriches ethane resources, making the oxidative dehydrogenation of ethane to ethylene quite fascinating, but the qualified catalyst with unique combination of enhanced activity/selectivity, enhanced heat transfer, and low pressure drop presents a grand challenge. Herein, a high-performance Nb2O5-NiO/Ni-foam catalyst engineered from nano- to macroscale for this reaction is tailored by finely tuning the performance-relevant Nb2O5-NiO interaction that is strongly dependent on NiO-precursor morphology. Three NiO-precursors of different morphologies (clump, rod, and nanosheet) were directly grown onto Ni-foam followed by Nb2O5 modification to obtain the catalyst products. Notably, the one from the NiO-precursor of nanosheet achieves the highest ethylene yield, in nature, because of markedly diminished unselective oxygen species due to enhanced interaction between Nb2O5 and NiO nanosheet. An advanced catalyst is developed by further thinning the NiO-precursor nanosheet, which achieves 60% conversion with 80% selectivity and is stable for at least 240 h.

An increase in the consumption of fossil-based resources has initiated the exploration of biomass conversion to produce high-value chemicals and fuels [1,2]. Furan-based chemicals, owing to the abundant biomass sources, have received considerable attention as valuable alternatives to chemicals obtained from fossil resources.[3–5]. 5-hydroxymethylfurfural (HMF) is considered a key molecule for sustainable development [6,7] because it is widely used as a platform chemical in biorefineries. HMF is hydrogenated to 2,5-dimethylfuran, 5-methylfurfural (MF), 2,5-bis(hydroxymethyl)furan (BHMF), and 5-methyl-2-furfuryl alcohol depending on the catalyst and reaction conditions [8,9].Among HMF hydrogenation products, BHMF is significant for the synthesis of several foams, polyethers, and crown ethers owing to the presence of a symmetrical diol functional group [10–14]. The key to selective hydrogenation of HMF to BHMF is to saturate the CO bond while avoiding cleavage of the CO bond. Various noble metal-based heterogeneous catalysts have shown high selectivity in BHMF formation [15–19]. Pt/C was the first catalyst for the synthesis of BHMF in 2012, achieving 82 % yield after 18 h [15]. Zhang et al. developed an Ir/TiO2 catalyst and achieved 94 % yield under harsh conditions and an H2 pressure of 6 MPa [16]. A layered double oxide Ru/ZnAlZr prepared by Gao et al. delivered nearly 94 % yield at 473 K [17]. Ohyama et al. reported an aluminum oxide-supported gold catalyst for BHMF synthesis with an 80 % yield after 2 h at an H2 pressure of 3.8 MPa without significant furan ring hydrogenation [18]. Recently, Nishimura et al. reported the selective hydroconversion of HMF over a Pd/Al2O3 catalyst under ambient conditions using sodium hypophosphite to form hydrogen atoms and tetrahydrofuran/water as the solvent with a low 60 % yield [19]. Although many significant results have been obtained using noble metal-based catalysts, their high cost, low abundance, and status as a strategic resource limits their applications. Consequently, the design of novel catalysts based on non-noble metals has attracted considerable attention.Transitionmetals have shown excellent performance as catalysts for the selective hydrogen reactions of various functional groups [20]. Co/C was used in the selective hydroconversion of HMF under a H2 pressure of 2 MPa for 6 h to furnish 93 % of the product[21]. Recently, Rao et al. synthesized a Cu/Al2O3 catalyst using solvent-free solid-state grinding to produce 92 % of BHMF under a hydrogen pressure of 3 MPa [22]. Elsayed et al. prepared CuO-Fe3O4/AC for the selective hydroconversion of HMF via catalytic transfer hydrogenation with 92 % yield at 413 K for 5 h [23]. Poor selectivity of Ni-based catalysts is a possible reason that they are rarely applied in BHMF synthesis; for example, the furan ring of BHMF was reduced over Raney Ni [24]. The recent development of nanoscience has made it feasible to regulate the catalyst function using special nanomaterials [25–27]. Carbon nanotubes (CNTs) have proven to be excellent supports for catalysts owing to their thermal conductivity, specific surface areas, and porous structures. This has led to increasing investigations on metal–carbon catalytic systems [28–30]. It is known that the surface of oxygen-functionalized Ni/CNTs bears free carboxyl groups (–COOH) that are mainly grafted onto the CNT surface and promote electron transfer from Ni atoms to the CNT support [31].In this study, a Ni/CNTs catalyst was synthesized and used for the selective hydrogenation of HMF to BHMF. Catalyst samples with different Ni/CNTs ratios were prepared using the impregnation synthesis method and characterized using various techniques. Factors affecting the hydrogenation process, such as the H2 pressure, reaction temperature, catalyst loading, and reaction time, were optimized to achieve a relatively high yield of BHMF.2 g of raw 10–20 nm multiwalled carbon nanotubes (CNTs) (purchased from XFNANO Co., ltd.) were oxidized with 200 mL of concentrated HNO3 (purchased from Sinopharm Chemical Reagent Co., ltd.) at 348 K for 24 h. Ni/CNTs samples were subsequently prepared by an impregnation synthesis method as reported by Lee et al [27] with slight modifications. Considering the target Ni:CNTs weight ratio of 3:17, 0.2 g of CNTs and 0.1765 g of nickel nitrate hexahydrate (purchased from Sinopharm Chemical Reagent Co., ltd.) were mixed in a beaker. Subsequently, 10 mL of ultrapure water was added with continuous stirring. The precursor was calcined in an H2/Ar atmosphere at 673 K for 2 h after drying at 393 K for 10 h. The applied heating rate from 323 to 673 K was 5 K/min. The prepared catalyst, denoted as 15 wt% Ni/CNTs, is a highly magnetic material. The catalysts with different Ni contents were prepared through the same procedure.X-ray diffraction (XRD) patterns of the prepared catalyst samples were recorded using a Ultima IV powder X-ray diffractometer (Rigaku, Japan) with a Cu K-α radiation source and a tube pressure of 40 kV for a diffraction angle (2θ) ranging from 10° to 90°. X-ray photoelectron spectroscopy (XPS) was performed using a K-α spectrometer (Thermo Scientific, USA) under vacuum conditions, and spectra were corrected based on the C1s line at 284.80 eV. Nitrogen adsorption measurements were performed on the samples using an ASAP 2460 sorption analyzer (Micromeritics, USA). The samples were outgassed at 473 K for 4 h before measurements. Transmission electron microscopy (TEM) images were obtained using a TF20 TEM (FEI, USA) equipped with a super X field emission gun.Typically, HMF (1 mmol, 126 mg), catalyst (50 mg), and tetrahydrofuran (10.0 mL) were added to a 50 mL autoclave which was sealed and purged with hydrogen four times. Hydrogenation of HMF was performed at a certain reaction temperature and hydrogen pressure with magnetic stirring. After the reaction was completed, the reactor was cooled using ice water. The solid product was removed, and the solution was analyzed using a gas chromatograph (GC; Nexis GC-2030, Shimadzu, Japan) equipped with a flame ionization detector (FID) and a capillary column (SH-Rtx-1701). Structural characteristics of the samples were analyzed by gas chromatography–mass spectrometry (GC–MS) (GC-2010, Shimadzu, Japan).The temperature procedure for GC was as follows:313 Kfor2 min, 313 to 373 K (at 20 K/min),3 min, 373 to 473 K (at 20 K/min), and 473 Kfor2 min. The equations for the HMF conversion, BHMF selectivity, and MF selectivity are shown in Eqs. (1)–(3). (1) H M F c o n v e r s i o n % = 1 - MoleofHMF InitialmoleofHMF × 100 % (2) B H M F Y i e l d % = MoleofBHMF InitalmoleofHMF × 100 % (3) M F Y i e l d % = MoleofMF InitalmoleofHMF × 100 % All the analyzed catalyst samples (5, 10, 15, 20, and 25 wt% Ni/CNTs) exhibited peaks characteristic of graphite-2H and the face-centered cubic crystal structure of metallic Ni as shown in Fig. 1 . The CNT structure was not destroyed during the synthesis process as demonstrated by peaks characterized at 2θ = 26.3° and 42.2° for all samples. Peaks at 2θ = 44.5°, 51.8°, and 76.4° correspond to (111), (200), and (220) crystal faces of the nickel face-centered cubic structure, respectively [27]. No peaks representing NiO are observed establishing the fact that no metal oxidation occurred during the synthesis. The impregnation method resulted in good loading of the metal onto the surface of CNTs as demonstrated by these results. Remarkably, the observed peak intensity of Ni increased as the Ni loading in the catalyst increased to 15 wt% in Ni/CNTs. Fig. 2 a shows the particle size distribution of 15 wt% Ni/CNTs. The distribution revealed an average particle size of 9.31 nm which is not visible in the raw TEM image. Fig. 2b shows the high-angle annular dark-field (HAADF)–TEM image indicating that Ni nanoparticles are strongly attached to the surface of the CNT support. Elemental maps of the catalysts are shown in Fig. 2c–f demonstrating that Ni is evenly distributed on the support surface with partial metal agglomeration. Functionalized carbon nanotube supports can effectively disperse elemental Ni, increase the number of active sites, and reduce the amount of metal. This result, in particular, encourages the investigation of low-loading CNTs as effective catalysts.The XPS full elemental survey shown in Fig. 2g confirms the presence of oxygen, carbon, and nickel in the catalyst. The presence of oxygen is attributed to the –COOH group on the surface of CNTs as well as to some adsorbed oxygen. In accordance with the published data [27,32], XPS spectra of the analyzed catalysts displayed peaks corresponding to Ni2+ (or NiO) and Ni0. Peaks at 853.1 and 856.1 eV correspond to Ni0 2p1/2 and Ni2+ 2p1/2, respectively. Binding energies at 871.1 and 874.5 eV correspond to Ni0 2p3/2 and Ni0 2p3/2, respectively. Two additional satellite peaks of Ni are detected at 861.6 and 880.1 eV. This is because the grafted carboxyl groups strengthen the Ni − CNT interaction resulting in the reduction of Ni2+ to Ni0 as observed in an amorphous form [31]. XPS peaks of the carboxyl group in Fig. 2i and XRD peaks of metallic Ni confirm this hypothesis. Table 1 shows the physical properties of the synthesized Ni/CNTs catalysts.The Brunauer–Emmett–Teller (BET) surface areas of 5–20 wt% Ni/CNTs) range from 132 to 140 m2g−1. For CNTs coated with Ni, the surface area of samples decreased gradually with increasing Ni loading. As these values are nearly equal to those of CNT carriers (143.41 m2g−1), it indicated that Ni nanoparticles were deposited on CNTs without collapsing the nanostructure. Fig. 3 indicates isotherms of all catalysts are typical type-IV isotherms with a hysteresis loop of type H1. These isotherms strongly prove the mesoporous structure of the prepared samples. Despite the mesoporous structures, the proportion of macropores gradually increased with the addition of Ni as observed from the pore size distribution curves. Fig. 3g indicates a large number of microporous in 25 wt% Ni/CNTs. The destruction of the mesoporous structure is the cause of catalytic activity degradation with increasing Ni content.Reaction conditions: 1 mmol HMF, 40 % catalyst/HMF, 10 mL tetrahydrofuran, 0.5 MPa hydrogen pressure, 3 h, 393 K.Ni/CNTs catalysts with different nickel loadings were used for selective hydrogenation of HMF at a temperature of 393 K and H2 pressure of 0.5 MPa (Table 2 ). MF and a few other substances identified by GC–MS were formed as reaction byproducts under these conditions. The BHMF yield increased with an increasing Ni content from 5 to 15 wt%. An HMF conversion of 52 % with 95.1 % of BHMF selectivity was achieved using 15 wt% Ni/CNTs. A further increase in nickel loading reduced the BHMF yield and the BET surface area of the catalysts to a small extent. This is because an excess Ni loading interferes with the surface structure of CNTs leading to a decreased catalytic activity. Thus, 15 wt% Ni/CNTs was considered the optimum loading to catalyze hydrogenation of HMF to a diol.15 wt% Ni/CNTs was selected for the next part of the study. The dependence of the product distribution on these parameters was explored by HMF hydrogenation at different reaction times ranging from 3 to 12 h. Fig. 4 shows that the BHMF yield is not affected by the reaction time (in the explored range). It was constant at 75 % after a reaction time of 6 h. BHMF was the main reaction product, and the selectivity of MF slightly increased with the reaction time. HMF conversion was not increased with time because of the substrate adsorption–desorption equilibrium on the catalyst. Thus, it was difficult to improve the BHMF yield.The conversion of HMF significantly improved with an increase in the catalyst ratio, whereas the selectivity of BHMF formation was maintained at approximately 95 % as shown in Fig. 5 . Notably, 99.8 % of HMF was converted to form 94.8 % of BHMF with the catalyst/HMF ratio of 100 %. Combined with the previous investigation of the dependence on the reaction time, this effect can be explained by the fact that more active sites are available for catalyzing HMF hydrogenation upon increasing the catalyst amount. This resulted in shifting the reaction equilibrium to form products.Hydrogen pressure had a significant effect on the conversion of HMF but a slight effect on the selectivity of BHMF as shown in Fig. 6 . The conversion was relatively low in a reactor in a low-pressure atmosphere (0.25 MPa). Increasing the hydrogen pressure resulted in a gradual increase in the conversion with a maximum conversion of 72.6 % at 1 MPa. A further increase in the hydrogen pressure to 1.25 MPa or higher did not significantly increase the HMF conversion. The yield of MF as a byproduct was very low (<1.9 %).The catalyst was used for the conversion of HMF to BHMF at different temperatures in the range of 353–453 K to investigate the effect of the reaction temperature. Hydrogenation of HMF rarely occurs at 353 K because the energy acquired at this temperature is insufficient to overcome the energy barrier. With an increase in the reaction temperature, BHMF selectivity decreased from approximately 95.0 % to 84.3 % at 453 K as shown in Fig. 7 . This behavior is attributed to the conversion of BHMF to MF and MFA at higher temperatures. The direct conversion of HMF to MF was excluded because it requires higher activation energy than that required for the conversion of BHMF. This result will be discussed in more detail in Section 3.3. In summary, our results suggest that a lower reaction temperature is beneficial for the selectivity of BHMF although it results in lower conversion.In previous reports, hydrogen under high pressure was necessary to produce BHMF from HMF (more than 3 MPa) [21,33,34]. In this work, the high dispersion of active sites due to the high specific surface area and defect sites formed due to the oxidation treatment of CNTs resulted in hydrogenation under the pressure of 0.5 MPa H2. In addition, Ni/CNTs is a strong magnetic material that is easier to recycle than nonmagnetic materials.Aldehyde and hydroxyl groups present on opposite sides of the HMF furan ring can react with a suitable hydrogen donor to saturate the CO bond and break CO bond to form different products. Therefore, MF was expected to be formed as a byproduct (Scheme 1 ). Hydroconversion of HMF was performed at three different temperatures to determine the kinetic parameters: 393, 413, and 433 K. Note that in all experiments, an excess amount of hydrogen as compared to the amount of the substrate (1 mmol HMF) was used under 0.5 MPa pressure for assuming a constant amount of hydrogen during the entire hydrogenation process. Under these conditions, kinetic rate constants and activation energies of the reactions were calculated using a Pseudo first order model. Equations (4)–(6) were used to determine the kinetic parameters for the hydrogenation of HMF. (4) d [ H M F ] / d t = - k 1 × [ H M F ] - k 1 × [ H M F ] (5) D [ B H M F ] / d t = k 1 × [ H M F ] (6) d [ M F ] / d t = k 2 × [ H M F ] where k1 and k2 are the Pseudo first order rate constants for formation of BHMF and MF, respectively, at a specific reaction temperature, and t is the reaction time (h).Equations (4)–(6) are integrated under the initial conditions corresponding to t = 0 and [ H M F ] = [ H M F ] 0 . The concentrations and reaction times are expressed using Eqs. (7)–(9). (7) H M F = [ H M F ] 0 × e x p ( - ( k 1 + k 2 ) × t ) (8) B H M F = [ H M F ] 0 × k 1 / ( k 1 + k 2 ) × ( 1 - exp - k 1 + k 2 × t ) (9) M F = [ H M F ] 0 × k 2 / ( k 1 + k 2 ) × ( 1 - exp - k 1 + k 2 × t ) All parameters were estimated using OriginPro Learning Edition software by performing a nonlinear curve fit to the selected data with the corresponding rate equation. Curve fits of the experimental data at the investigated reaction temperatures are shown in Fig. 8 . Table 3 lists the activation energies determined from the Arrhenius equation and reaction rate constants corresponding to the investigated temperatures. The similarity between experimental data and fitting curves confirms that the Pseudo first order reaction model accurately describes the hydrogenation process. As expected from the experimental data, the rate constant (k1) of BHMF formation was significantly larger than that of MF (k2). Similarly, the activation energy of BHMF formation (21.12 kJ/mol) was almost half as large as that of MF formation (51.46 kJ/mol). The results from the kinetic investigation explain the high selectivity of BHMF. Larger rate constants are always associated with higher temperatures which explain the higher MF yields observed at elevated temperatures.Various HMF derivatives can be produced by reducing the different functional groups present in HMF, thus making it a complex reaction. In this study, 15 wt% Ni/CNTs exhibited high selectivity for reducing CO among the other functional groups. The XPS results revealed that Ni0 and Ni2+ ions are present on the 15 wt% Ni/CNTs surface. These act as active sites for the conversion of HMF substrates in a hydrogen atmosphere. Additional experiments were performed to explain the role of both Ni0 and Ni2+ in the selective HMF hydrogenation. We observed a decrease in the Ni0/Ni2+ ratio on the 15 wt% Ni/CNTs surface with an increasing reaction temperature that was in agreement with previously reported data [27]. The smaller the size of Ni nanoparticles, a larger surface area, and larger NiO film area are expected.To understand the impact of nickel ions on the reaction, 15 wt% Ni/CNTs was prepared at different reduction temperatures, namely 573, 673, and 773 K. The catalyst was denoted as 15 wt% Ni/CNTsx, where × denotes the temperature. The samples were characterized by XPS (Fig. 9 and Table 4 ). The results clearly show that the lowest Ni0 content on the surface was associated with the slowest reaction rate. This relationship may be due to the slowing down of dissociative adsorption of hydrogen which requires Ni0. However, increasing the Ni0 content from 24.1 % to 37.2 % did not increase the catalytic activity. Based on this evidence, we believe that there is a synergistic effect between Ni0 and Ni2+ ions. Ni2+ adsorbs nucleophilic unsaturated functional groups and Ni0 provides activated hydrogen. Consequently, the presence of excessive Ni2+ on the surface adsorbs nucleophilic groups more efficiently and boosts the hydrolysis reaction to form MFA and DMF as products.The condensed local nucleophilicity index was obtained using the Multiwfn software [35] by calculating the molecular index of the optimized HMF molecule. According to Table 5 , 16(O) is more nucleophilic than 14(O) with a condensed local nucleophilicity index almost three times larger (Fig. 10 shows the numbering pattern). This evidence suggests that a Lewis acid, such as Ni2+, prefers to adsorb HMF through the aldehyde group instead of the hydroxyl group.Feng et al. proposed the coexistence of both metallic and electrophilic metal species as a prerequisite for selective hydrogenation of HMF [36]. We proposed a plausible reaction mechanism to explain the synergistic effect between Ni0 and Ni2+ as shown in Scheme 2 based on the above experimental results and literature reports [22,37–39]. Initially, the carbonyl group in HMF is adsorbed onto the electrophilic Ni2+ species on the catalyst surface and is activated. Simultaneously, hydrogen dissociates at the metallic nickel (Ni0) site. The electron lone pair of H− attacks the C atom of the activated carbonyl group, whereas the same on the CO bond is transferred to the O atom. HMF is converted to BHMF after the activated O atom and H+ form a CO bond.BHMF yields gradually decreased as the number of cycles increased as observed in Fig. 11 . The leaching test results shown in Fig. 12 indicate that the yield increases continuously for 3 h after which it became constant on removing 15 wt% Ni/CNTs from the reaction solution. This result demonstrates that active sites were not leached and 15 wt% Ni/CNTs is heterogeneous. To investigate the reason for the decreased catalytic activity, XRD tests of the fresh and spent catalysts were carried out as shown in Fig. 13 . XRD results excluded the presence of any additional peaks, demonstrating that the crystalline structure of the catalyst was not changed even after its use. However, the intensity of metallic nickel peaks slightly decreased along with XRD peaks at 2θ = 26.3° and 42.2°. It is possible that the substrate or other amorphous material was attached to the catalyst after the reaction resulting in a broad peak in the range of 15°–40°. The above result explains the decline in catalytic activity as shown in Fig. 11. Amorphous substances covering the active sites reduces the active sites with successive catalytic cycles. As previously reported, the spent catalyst partly recovers its catalytic activity after recalcination in a mixture of 20 % H2/Ar at 670 K [39]. Even so, repeated high-temperature treatments lead to metal agglomeration making the catalyst unable to recover completely.A highly efficient carbon nanotube-supported nickel catalyst (Ni/CNTs) was prepared using an impregnation method. The catalyst exhibits excellent activity and selectivity and is substantially less expensive. The high selectivity of the catalyst results from the optimal Ni0:Ni2+ ratio and the small size of nanoparticles. The reaction temperature and catalyst amount are crucial parameters for achieving a high BHMF yield. Under the optimal reaction conditions, a 93.1 % yield of BHMF was achieved. The kinetic study revealed that the conversion of HMF to BHMF is associated with the lowest activation energy (21.12 kJ/mol) which is half of that required to form MF (51.46 kJ/mol). The difference between the activation energies of BHMF and MF explains the high selectivity toward BHMF. These results provide a novel method for the selective hydrogenation of HMF to BHMF and promote research on biomass energy.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors gratefully acknowledge the financial support of the National Natural Science Foundation of China (Nos. 22278121 and 21975070), the China Postdoctoral Science Foundation (2019 M662787), and the Science and Technology Planning Project of Hunan Province (2021GK5083).

2,5-Bis(hydroxymethyl)furan (BHMF) is a high-value, bio-based, rigid diol that resembles aromatic monomers for the production of different polyesters. In this work, a carbonnanotubes (CNTs)-supported nickel catalyst (Ni/CNTs)was prepared and used for the selective hydrogenation of 5-hydroxymethylfurfural (HMF) to BHMF at low hydrogen pressure. The prepared catalyst was analyzed by nitrogen adsorption–desorption isotherms, X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS). According to kinetic studies, the rate constant for BHMF formation is significantly larger than that for the formation of the byproduct, 5-methyl furfural (MF). At optimal reaction conditions, conversion and selectivity rates of HMF and BHMF were 99.8 % and 95.0 %, respectively. The mechanistic study indicated the coexistence of Ni0 and Ni2+ species on the catalyst surface affects the catalytic performance. A possible mechanism was proposed to describe the synergetic effects of Ni0 and Ni2+. Furthermore, the catalyst can be easily separated from the reaction mixture for recycling.

The catalytic hydroconversion of n-paraffins is an important reaction to improve the quality of diesel and gasoline in the oil-refining industry (Zhou et al., 2022). Hydroisomerization of light alkane can produce high-octane fractions for gasoline blending with non-aromatic hydrocarbons to meet increasingly stringent environmental protection regulations (Zhan et al., 2022). Environmentally friendly solid superacid catalysts, especially for sulfated zirconia (SZ)-based samples, have been regarded as the most promising candidates for preparation of isomerization catalysts with high catalytic activity at low reaction temperature (Wang et al., 2020). However, the catalytic performance of pristine SZ is known to be unacceptable caused by the rapid deactivation in practical application. Therefore, the dopants of noble metals (Pt, Pd) and/or various transition metals (Fe, Ni, Mn, and Cu) have been adopted to modify the catalyst and results in much higher activity than that of raw SZ (Lyu et al., 2021; Song et al., 2015). Typically, the Ni belongs to the same family of Pd and Pt, more and more attentions have been paid to design low-cost Ni-modified SZ catalyst for replacement of noble metal. Though, the Ni-SZ catalyst exhibits acceptable isomerization activity at low temperature, the deactivation is also needed to be taken into consideration (Song et al., 2016a). Reasons for deactivation of SZ-based catalysts have been reported to be complex (Liu et al., 2020; Wang et al., 2016; Kim et al., 2000; Li and Stair, 1996), such as coke deposition (Li et al., 2006), leach of sulfate species (Ng and Horvát, 1995), change in surface acidity (González et al., 1997) and phase transition from tetragonal to monoclinic zirconia (Li and Stair, 1996). Therefore, it is of great of interest to design non-noble metal modified SZ catalysts with high catalytic activity and stability.Recently, many efforts have been made to alleviate the deactivation and the introduction of alumina into SZ are found to improve the catalytic activity and stability for isomerization performance (Hua et al., 2000; Gao et al., 1998). The modification of alumina contributed to the enhanced concentration of active sites and acid sites. Furthermore, the addition of alumina can retard the crystal phase transformation of ZrO2 from tetragonal to monoclinic phase (Zhou et al., 2021; Wang et al., 2022). As compared to intrinsic Pt–SO4/ZrO2 catalyst, the alumina-modified sample exhibited higher pore volume and specific surface area, more importantly, the alumina resulted in enhanced stability of tetragonal zirconia, which contributed to an excellent stability and activity for light naphtha isomerization (Zhou et al., 2022). Our previous research (Song et al., 2014) also found that the addition of an appropriate amount of Al (2.5 wt% of Al) can increase the amount of acid sites and the surface area, suppressing the phase transformation of tetragonal ZrO2 to monoclinic ZrO2. But an excessive amount of Al would decrease the number of surface tetragonal ZrO2 particles and led to a decrease in the formation of acid sites, which was generated by sulfate species adsorbed on the stepped edges of tetragonal ZrO2, and thus resulted in a significant decrease in activity. Therefore, the complete utilization of the skeleton structure and acid nature of γ-Al2O3 support is limited due to the low Al content of <5 wt%. Considering the disadvantages, the as-prepared catalyst often exhibited poor stability caused by the deactivation of SZ. (Song et al., 2016b).Generally, the pore structure and acidity of the support played a vital role on the coke deposition during n-alkane isomerization. For example, the pore with large size can facilitate the mass transfer, which reduces the residence time of hydrocarbon compounds on the catalyst surface and suppresses the deposition of carbon. Recently, core-shell structure materials have attracted worldwide attention due to their unique physical and chemical properties (Das et al., 2021). The core-shell nanoparticles exhibit many advantages, such as tunable surface modification, improved functionality, enhanced stability by protecting the active phase from contact with poisoning substances, lower consumption of precious materials, and so on (Gao et al., 2021). However, to the best of our knowledge, few studies have been reported concerning the use of γ-Al2O3 as a core support material for preparation of superacid SZ catalysts.Herein, a method for preparing highly active and highly stable non-noble-nickel-modified persulfated Al2O3@ZrO2 core-shell catalyst (Ni–S2O8 2−/Al2O3@ZrO2) was proposed to make full use of the respective advantages of Al2O3 and ZrO2. The Al2O3 core can impart the core-shell structure materials with high internal surface area and high mechanical strength for the support, which contributes to the formation of external shell with more and smaller tetragonal ZrO2 particles. As a result, the formation of superacid sites is accelerated due to the intimate contact between Zr and S species. Besides, the Al2O3 endows additional acid sites for the core-shell support and stabilizes the active tetragonal phase of ZrO2, which is also responsible for the improved catalytic performance. In the case of n-pentane isomerization, the core-shell Ni–S2O8 2−/Al2O3@ZrO2 catalyst showed a high isopentane yield (63%) with little or no deactivation within 5000 min. To the best of our knowledge, such a non-noble superacid catalyst with high isopentane yield and excellent stability at a low pressure (2.0 MPa) is extremely unusual.The Al2O3@ZrO2 (core@shell, A@Z) supports were synthesized by deposition of zirconia on the γ-Al2O3. In a typical procedure, a certain amount of γ-Al2O3 and butanol were mixed at room temperature, and deionized water was added dropwise to the suspension under vigorous stirring for further dispersion. Then, the calculated amount of zirconium (IV) butoxide was dissolved into the resulting suspension with different Zr/Al mass ratio, and stirred for another 30 min. Subsequently, the suspension was transferred into autoclaves for hydrothermal reaction at 443 K for 24 h. After cooling down to room temperature, the obtained products were separated by centrifugation, and then dried at 353 K for 24 h to obtain the A@Z-x samples, where x represented the percentage of Al content. Then the product was re-dispersed into a 0.75 M (NH4)2S2O8 solution and stirred for 15 min. After aging for 6 h, the samples were separated by centrifugation and dried at 353 K for 24 h to obtain the SA@Z-x samples.The supported core@shell nickel catalysts (Ni-SA@Z-x) were prepared by the incipient wetness impregnation method (Song et al., 2015). Typically, calculated amount of the SA@Z-x material and Ni(NO3)2·6H2O were added into 10 mL deionized water. Then the obtained samples were dried at 373 K for 12 h and calcined at 923 K for 3 h to obtain the Ni-SA@Z-x with Ni loading of 1.0 wt%.According to our previous study (Song et al., 2014), Pd-SZA catalyst made from Al content of 2.5 wt% exhibited the best performance. Therefore, for comparison the common SZA with Al2O3 content of 2.5 wt% was chosen to synthesize the supported Ni catalyst with Ni loading of 1.0 wt%. And the obtained catalyst was designed as Ni-SZA-2.5.X-ray powder diffraction (XRD) patterns were recorded on a D/max-2200PC X-ray diffractometer (40 kV, 40 mA) fitted with Cu Kα radiation (0.15404 nm). N2-adsorption was measured at 77 K using Micromeritics ASAP 2460 analyzer to obtain the microporous and mesoporous porosities, respectively. Transmission electron microscope (TEM) examinations were performed using the JEM-2010 instrument supplied by JEOL. Scanning electron microscope (SEM) with an acceleration voltage of 10 kV was conducted using Zeiss SIGMA equipment. Thermogravimetric analysis (TG) was performed on the samples (10 mg) after reaction using a Perkin-Elmer Diamond instrument under air with a flow rate of 100 mL min−1, from room temperature to 1123 K, and with a heating rate of 10 K min−1. Fourier transform infrared spectroscopy (FT-IR) measurements were carried out with a Bruker Tensor 27 FT-IR spectrometer. Fourier transform infrared spectroscopy of pyridine adsorption (Py-IR) was recorded on a Spectrum GX Fourier by adding 64 scans for the sample at a resolution of 4 cm−1. The metal loadings of the samples were determined by X-ray fluorescence (XRF) with a spectrometer XRF-1800. XPS were acquired with a PHI-1600 spectrometer equipped with a hemispherical electron analyzer and a Mg Kα (1253.6 eV) X-ray source.The isomerization reaction of n-pentane was chosen to evaluate the catalytic activity of the prepared catalysts. The reactions were performed in a fixed-bed flow reactor. Prior to reaction, 2 g of the catalyst was activated with flowing H2 stream (20 mL min−1) at 573 K for 3 h, and then cooled to the reaction temperature. The reaction conditions were set to a weight hourly space velocity (WHSV) of 1 h−1, an H2/n-pentane mole ratio of 4.0, a total pressure of 2.0 MPa and a temperature ranging from 433 to 533 K. The reaction products were analyzed by an online FL9790 gas chromatograph equipped with a FID detector.The XRD patterns of fresh and spent samples were shown in Fig. 1 . As depicted in Fig. 1a, all samples showed the diffraction peaks at 2θ = 30.3°, 35.3°, 50.4° and 60.4°, which were related to the (101), (110), (112) and (211) planes of tetragonal ZrO2, respectively (Reddy et al., 2018). The crystalline of tetragonal ZrO2 was affected by the dopant of Al, the diffraction peak of tetragonal ZrO2 was broadened with the increased Al content from 2.5 wt% to 50 wt%, indicating the decreased crystalline size of zirconia particles with incremental Al content. The absence of Al2O3 peaks in Ni-SA@Z-x with the high Al content of 30–50 wt% proved that Al2O3 core was totally coated by ZrO2 shell (Yang et al., 2013), suggesting the successful preparation of core-shell material. Besides, no crystalline phase of nickel oxide was detected due to the low content or high dispersion of nickel. Compared with the pattern of traditional Ni-SZA-2.5, the peak intensity of tetragonal ZrO2 decreased remarkably for all Ni-SA@Z-x catalysts, indicating that the core-shell structure can effectively suppress the growth of crystalline zirconia particles and result in much smaller particles size of zirconia even at the same Al content of 2.5 wt%.As shown in Fig. 1b, the diffraction peaks of monoclinic ZrO2 was detected for spent Ni-SZA-2.5 and Ni-SA@Z-2.5, which indicated the transformation of ZrO2 from the metastable tetragonal to the monoclinic phase during the isomerization reaction. Generally, the binary ZrO2/Al2O3 composite was often prepared by traditional sol-gel method and resulted in the uniform dispersion of Al and Zr species on the surface of binary nanocomposite, which led to the higher crystallizing temperature of tetragonal ZrO2 caused by the addition of Al2O3 (Zhao et al., 2007; Liu et al., 2012). Alternatively, the core-shell structure SA@Z showed advantages than ZrO2/Al2O3 composite, since the Al2O3 was totally covered by active tetragonal ZrO2 phase. What's more, the tetragonal ZrO2 phase has been reported to be necessary for isomerization performance (Liu et al., 2012). Besides, the monoclinic phase peaks of the spent Ni-SA@Z-x (x = 30–50) were very weak, suggesting the more stable tetragonal structure as compared to Ni-SA@Z-2.5 and higher catalytic stability in the case of n-pentane isomerization.The crystal sizes of tetragonal zirconia for all samples were calculated by the Debye-Scherrer equation and listed in Table 1 . Compared to traditional Ni-SZA-2.5 (9.7 nm), all Ni-SA@Z-x samples showed smaller crystallite size with increased Al addition. In detail, the tetragonal ZrO2 crystallite size of Ni-SA@Z-x decreased from 8.2 to 4.9 nm (decreased by 40.2%) with the increased Al content from 2.5 wt% to 50 wt%, indicating the positive effect of Al species on the formation of tetragonal ZrO2 crystallite with smaller size. Similar result has been reported by Zarubica et al. (2021), an increase in Al content promoted the stabilization of smaller tetragonal ZrO2 particles on the surface. As mentioned, the core-shell structure was beneficial to the formation of smaller tetragonal ZrO2 particles, which also accelerated the contact between Zr and S species to form Zr–S bonds and deduced the formation of a superacid structure and dispersion of active sites and acid sites. This will be further discussed in Sections 3.5 and 3.6. Besides, the ZrO2 crystallite size of Ni-SA@Z-x samples with 2.5 wt%∼50 wt% Al content increased about 1.6–0.7 nm after reaction, which was still much lower than that of Ni-SZA-2.5. This further confirmed that the core-shell structure could restrain sintering of the tetragonal ZrO2 phase and remain the integrity of its microscopic structure.The N2 adsorption-desorption isotherms and pore size distributions of the catalysts were shown in Fig. S1. Accordingly, all the isotherms showed a type IV characteristic feature of isotherm, indicating the presence of some mesopores (Thommes et al., 2015). Ni-SZA-2.5 showed a narrow pore size distribution centered at around 3 nm. The Ni-SA@Z-2.5 showed a broader pore size distribution from 5 to 10 nm, and the main peak was close to that of γ-Al2O3. With increasing Al content, the pore size increased remarkably owing to the abundant Al provided more mesopores, and some of the micropores gradually merged into mesopores.In comparison with Ni-SZA-2.5 (95.1 m2 g−1), the Ni-SA@Z-2.5 showed a slightly higher specific surface area (S BET) of 99.5 m2 g−1 (Table 1). In addition, the pore size (D p) increased remarkably from 3.7 to 5.6 nm, and the pore volume (V Total) increased from 0.089 to 0.103 cm3 g−1. This showed that the pore structures of these two catalysts were entirely different, even though the Al content and the compositions of the individual components are the same. The large D p and V Total of Ni-SA@Z-2.5 would enhance the diffusion rates of reactant and products. In particular, branched or large-sized products could pass through the pores more easily, suppressing carbon deposition on the surface of the catalyst. This would effectively arrest catalyst deactivation since carbon deposition is one of the main reasons for deactivation of catalysts of this kind (Song et al., 2016b). In addition, the large D p and V Total are also beneficial to the isomerization reaction. With increasing the Al content, the S BET, V Total and D p of Ni-SA@Z-x increased remarkably. It is worth noting that the D p and V Total of Ni-SA@Z-50 were 2.1 and 4.2 times higher than those of Ni-SA@Z-2.5, respectively.For Ni-SZA-2.5, S BET, V Total and D p was dramatically decreased after reaction. The narrowed D p indicated that carbon deposition mainly occurred inside the pores during the reaction. The deposited carbon would have coated the active metal sites and acid sites on the surface of the catalyst, leading to its deactivation. However, a slight decline in textural parameter of Ni-SA@Z-x catalysts was observed after reaction. Fig. 2 exhibited the TEM images of Ni-SZA-2.5, Ni-SA@Z-2.5 and Ni-SA@Z-50 catalysts. Mokari et al. (2005) proposed that Zr particles may be easily identified by their dark contrast in TEM, as a result of the electron density contrast between Al and Zr. Moreover, because of low content and high dispersion, Ni particles could not be observed (Nichele et al., 2012). It can be seen from Fig. 2a, c) that the ZrO2 particle size in the Ni-SZA-2.5 catalyst was approximately 8.4 nm with interplanar distances of 0.295 nm for ZrO2 (101) plane (Bang et al., 2020). In Fig. 2b, a light-color core surrounded by a dark shell can be clearly discerned. This indicated that a core-shell structure had been successfully synthesized. Besides, the core-shell structure of SA@Z-30 materials was also detected in SEM images (Fig. S2). The ZrO2 particle size in the core-shell Ni-SA@Z-2.5 catalyst was about 6.7 nm (Fig. 2d), smaller than that of Ni-SZA-2.5, implying that the core-shell structure was beneficial to the formation of smaller tetragonal ZrO2 particles. This observation was consistent with the XRD results (Table 1). For Ni-SA@Z-50, the Zr particle size decreased to about 3.9 nm with further increased Al content, indicating the positive effect of Al on dispersion of Zr species.The TG results of samples were shown in Fig. 3 . All samples display weight loss in the range of room temperature to 938 K, which is attributed to desorption of physically and chemically adsorbed water molecules and the dihydroxylation process on the surface of ZrO2 (Arkatova, 2010; Joo et al., 2013). Significant weight loss was clearly started at 938 K, which could be attributed to the decomposition of persulfate species with the evolution of sulfur dioxide, similar results have been reported elsewhere (Kim et al., 2006; Satam and Jayaram, 2008). Compared with Ni-SZA-2.5, the decomposition of persulfate species of Ni-SA@Z-x was shifted to higher temperatures of 963 K. These observations suggest that persulfate anions on the surface of Ni-SA@Z-x were bonded more strongly to dehydrated zirconia, leading to the increased thermal stability of superacid. This will be further discussed in Section 3.9.The FTIR spectrum of fresh catalysts was depicted in Fig. 4 . All the samples showed similar peaks, the band at 3422 cm−1 and 1630 cm−1 was assigned to the physically adsorbed water molecules and the bending mode (δ HOH) of coordinated molecular water associated with the persulfate group, respectively (Sarkar et al., 2007). The bands at 1156 cm−1 and 1077 cm−1 were assigned to the symmetric O–S–O stretching mode of bidentate persulfate ions coordinated to the metal ion, which was responsible for the Lewis acid sites in persulfated zirconia samples. The band at 1255 cm−1 corresponded to the antisymmetric OSO stretching frequency of persulfate ions bonded to ZrO2, which was responsible for the Brønsted acid sites in persulfated zirconia samples (Mishra et al., 2003). These three bands appearing at about 1077, 1156 and 1255 cm−1 were assigned to bidentate S ions coordinated to ZrO2 in C 2υ symmetry with a υ3 vibration, indicating the formation of a strongly superacid structure (Yadav and Murkute, 2004). The intensity and degree of splitting of the persulfate bands reflect the proportion of acid sites of the catalyst. The Ni-SA@Z-2.5 catalyst exhibited three vibration bands corresponding to SO and S–O bond, which showed higher intensity and degree of splitting than those of Ni-SZA-2.5. This indicated that the Ni-SA@Z-2.5 provided more acid sites and stronger acidity, as further confirmed by Py-IR results (see Section 3.6). With increasing Al content, the intensity and the degree of splitting of the vibrational bands corresponding to SO (1255 cm−1) and S–O (1077 cm−1 and 1156 cm−1) of Ni-SA@Z-x increased, and the Ni-SA@Z-30 possessed the highest intensity and the splitting degree among Ni-SA@Z-x.The Py-IR results of fresh and spent catalysts were listed in Tables S1 and S2. All of the catalysts possessed more Lewis acid sites than Brønsted acid sites, and both of them decreased with increasing desorption temperature. Compared with Ni-SZA-2.5, distinct increases in the amount of Brønsted and Lewis acid sites could be observed for Ni-SA@Z-2.5, which possessed smaller (Table 1) and more uniformly dispersed ZrO2 particles (Fig. 2 TEM) on the surface of mesoporous Al2O3 nanoparticles, facilitating interaction with S2O8 2− anions to generate acid sites. With increasing Al content, all the amounts of Lewis acid sites and Brønsted acid sites for Ni-SA@Z-x increased remarkably. These results indicated that the addition of Al improved the stability of the persulfate loaded on the surface to form stronger acid sites. Foo et al. (2015) proposed that the Brønsted acidity was associated with persulfuric acid clusters on zirconia. With increasing Al content, the persulfate anions were bonded more strongly to dehydrated zirconia (as shown in TG analysis) and thus formed more superacid sites.For all of the spent catalysts (Table S2), the amount of Brønsted acid sites and Lewis acid sites were both decreased as compared to the corresponding fresh one. However, the amount of Brønsted acid sites decreased more significantly than that of Lewis acid sites, implying that the former were the main active acid sites for isomerization (Yang and Weng, 2010). For Ni-SZA-2.5, the strong acid sites had completely disappeared after reaction. However, the spent Ni-SA@Z-2.5 still possessed the strong acid sites. In addition, the amounts of weak, moderate, and strong acid sites on Ni-SA@Z-x (x < 50) were still maintained at high levels after reaction. This can be attributed to a stabilizing effect of Al on S species on the catalyst surface and some suppression of the loss of acid sites (Hou et al., 2017). Analysis of the bulk sulfur content also confirmed it (Table2, Section 3.8).The surficial chemical composition of the catalysts was investigated by XPS analysis. As shown in Fig. 5 a, the full-scan XPS spectrum shows that the Ni-SZA-2.5 and Ni-SA@Z-x contains Ni, Zr, Al, S and O species, respectively. In Fig. 5b, the high-resolution S 2p spectrum consists of two contributions for all the samples. The peak centered at 169.1 eV can be assigned to S6+ species of peroxydisulfate (Shanthi et al., 2019). Sulfur with an oxidation state of +6 is known to be the most active and essential for the formation of solid superacid sites. While the peak appeared at 170.4–169.9 eV can be attributed to S–O–Zr bond. As compared to the Ni-SZA-2.5 and Ni-SA@Z-2.5, the binding energy of Ni-SA@Z-30 shifted to lower value, indicating that the electronic environment of S has changed when the Al content raised to 30 wt% (Wang et al., 2018). Table S3 showed the atomic contents obtained from XRF, XPS and carbon-sulfur analysis. The Ni contents of all of the catalysts were roughly equal to the stoichiometric content. In addition, compared to the fresh catalysts, no significant change was observed after reaction, showing that deactivation of the catalyst was not caused by the Ni leaching loss. Compared with Ni-SZA-2.5 (1.74 wt%), the sulfur content of the Ni-SA@Z-2.5 increased to a slightly higher value of 2.36 wt%, indicating that Ni-SA@Z-2.5 could stabilize more S species, as discussed in Section 3.7. With incremental Al content, the sulfur content further increased and the sulfur content of Ni-SA@Z-50 reached up to 3.41 wt%. After reaction, both Ni-SZA-2.5 and Ni-SA@Z-x catalysts underwent an overt sulfur loss. Many researchers (Yang and Weng, 2009; Saha and Sengupta, 2015) have found that the loss of loosely bound S species during reaction resulted in catalyst deactivation. Besides, for Ni-SZA-2.5, significant carbon deposition occurred during isomerization (0.45 wt%). The amount of carbon deposition on the spent Ni-SA@Z-x samples was improved as compared to that of the spent Ni-SZA-2.5, which supported the view that the larger mesopore volume of the core-shell catalysts effectively enhanced the diffusion rate and inhibited carbon deposition. Thus, it can be speculated that the Ni-SA@Z-x catalysts would exhibit excellent thermal stability (Kuznetsov et al., 2017) (see Scheme 1).The Ni-SZA-2.5 and Ni-SA@Z-x catalysts have been tested in the isomerization of n-pentane at a pressure of 2.0 MPa, an H2/n-pentane molar ratio of 4.0, and a WHSV of 1.0 h−1 and results were illustrated in Fig. 6 and Fig. S3. The catalytic activities of all of the catalysts first increased and reached a maximum, and then decreased with increasing temperature. The raw Ni-SZA-2.5 showed a maximum isopentane yield of 60.3% at optimized temperature (Fig. S3). For Ni-SA@Z-2.5, the isopentane yield of 65.6% was reached at 473 K. With increasing the Al content, the optimum temperature decreased and then increased. The Ni-SA@Z-30 possessed the lowest optimum temperature at 453 K with the high isopentane yield of 64.7%. Possible reasons to explain the high isopentane yield of Ni-SA@Z-30 catalyst at lower temperature may be as follows (Scheme 2 ). (i) More and stronger superacid sites are formed featured by the FTIR and Py-IR analysis (Table S1 and Fig. 5). In particular, the amount of strong Brønsted acid sites of Ni-SA@Z-30 was 6.0 and 54.3 times higher than that of Ni-SA@Z-2.5 and Ni-SZA-2.5, respectively. (ii) Better dispersion of active acid and metal sites was achieved due to the high surface area (Table 1). However, further increased Al content resulted in an adverse activity, which led to higher reaction temperature and downtrend in isopentane yield. According to Kamoun et al. (2015), addition of excessive Al to Ni/ZrO2–SO4 2− have the negative effect of the Al on isomerization activity at low temperature. Similar results have been reported in our previous study (Song et al., 2014), which showed that the isopentane yield over Pd-SZA-2.5 (Al content of 2.5%) was 64.3% at 511 K, however, when the amount of Al content increased to 5 wt%, the optimum temperature increased to 553 K with a sharp decline in isopentane yield. This can be attributed to the decrease in the tetragonal phase and its crystallinity at a higher Al content.(Reaction condition: p = 2.0 MPa, H2/n-pentane molar ratio = 4.0, WHSV = 1.0 h−1). Fig. 7 showed the stability results for Ni-SZA-2.5 and Ni-SA@Z-x over a period of 5000 min at their corresponding optimum reaction temperatures with other conditions maintained the same. The isopentane yield of the Ni-SZA-2.5 catalyst showed an obvious decline during isomerization, which decreased dramatically from 60.3% to 20.0% (decreased by 66.8%) after 1500 min. Compared to traditional Ni-SZA-2.5 catalysts, the Ni-SA@Z-2.5 exhibited much better stability, and the isopentane yield showed a slight decreased from 65.4% to 60.2% (decreased by 7.7%) after 1500 min and to 50.1% (decreased by 23.1%) after 5000 min. The Ni-SA@Z-30 catalyst exhibited the most promising catalytic performance and showed a high isopentane yield of approximately 63.1% with no or tiny deactivation after 5000 min. Possible reasons may be proposed to explain the great stability of the Ni-SA@Z-x catalyst for n-pentane isomerization (Scheme 2). (i) The pore sizes and volumes of the catalysts increased in the order: Ni-SZA-2.5 (3.7 nm, 0.089 cm3 g−1) < Ni-SA@Z-2.5 (5.6 nm, 0.103 cm3 g−1) < Ni-SA@Z-30 (6.9 nm, 0.214 cm3 g−1) (Table 1). The large pore size and pore volume enhanced the diffusion rates of the reactant and products and largely suppressed carbon deposition. This was confirmed by analysis of carbon deposition on the spent samples. The amounts of carbon deposited on the spent Ni-SA@Z-2.5 and Ni-SA@Z-30 were only 0.07 wt% and 0.05 wt%, respectively, much lower than that of spent Ni-SZA-2.5 (0.45 wt%, Table S3). The color changes of the catalysts after reaction also supported this (Fig. 7). (ii) The loss of sulfur entities can be suppressed for Ni-SA@Z-x. The ZrO2 shell, which consists of more and smaller tetragonal ZrO2 particles because of the large surface area of the Al2O3 core (Table 1), ensured intimate contact between Zr and S. Therefore, the superacid became more stable in thermally (See TG analysis). Elemental analysis showed that the sulfur content of Ni-SA@Z-2.5 and Ni-SA@Z-30 catalysts decreased by 16.1 and 18.1% after reaction for 5000 min on stream, respectively, whereas the Ni-SZA-2.5 underwent a higher sulfur loss of 27.6% after 1500 min (Table S3). Therefore, the deactivation of the catalysts caused by sulfur removal was somewhat suppressed for Ni-SA@Z-x. (iii) For the Ni-SZA-2.5 catalyst, the strong Brønsted acid sites, which played an important role in isomerization (Li et al., 2020), completely disappeared after reaction (Table S2). As contrast, the content of Brønsted acid for the spent Ni-SA@Z-2.5 and Ni-SA@Z-30 was 0.6 μmol g−1 and 6.5 μmol g−1, respectively. In addition, the contents of weak, moderate, and strong acid sites of Ni-SA@Z-30 still maintained to a great extent after reaction.(Reaction condition: p = 2.0 MPa, H2/n-pentane molar ratio = 4:1, WHSV = 1.0 h−1).This study paved a new path for the synthesis of highly active and highly stable non-noble Ni-SA@Z-x catalysts for n-pentane isomerization. The Ni-SA@Z-30 provided a sustained high isopentane yield (64.7%) with little or no deactivation within 5000 min at a low temperature of 453 K. The high isopentane yield of Ni-SA@Z-30 can be attributed to the formation of more and stronger superacid sites due to numerous small tetragonal ZrO2 particles derived from ZrO2 shell and better dispersion of active acid and metal sites. The excellent stability can be attributed to the following factors: (i) carbon deposition was greatly suppressed by the large pore size and huge pore volume; (ii) the loss of sulfur entities was suppressed due to the stronger interaction between small tetragonal ZrO2 particles and S species; (iii) the loss of strong Brønsted acid sites was improved during the isomerization reaction. To the best of our knowledge, such a non-noble superacid catalyst with high isopentane yield and excellent stability at a low pressure (2.0 MPa) is extremely unusual and being reported for the first time.The following is the Supplementary data to this article. Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.petsci.2023.02.027.

The non-noble metal modified sulfated zirconia was found easy to deactivate. Herein, highly active and highly stable non-noble core-shell Ni–S2O8 2−/Al2O3@ZrO2 catalysts (Ni-SA@Z-x, x = Al content in wt%) have been successfully prepared and investigated for n-pentane isomerization. The results showed that the core-shell Ni-SA@Z-30 provided a sustained high isopentane yield (63.1%) with little or no deactivation within 5000 min at a mild reaction pressure of 2.0 MPa, which can be attributed to the following factors: (i) carbon deposition was greatly suppressed by the large pore size and huge pore volume; (ii) the loss of sulfur entities was suppressed because the small and highly dispersed tetragonal ZrO2 particles can bond with the S species strongly; (iii) strong Brønsted acidity can be maintained well after the isomerization. The pore structures and acid nature of the core-shell Ni-SA@Z-x are entirely different from those of the normal structure Ni–S2O8 2−/ZrO2–Al2O3, even though the Al content and the compositions of the individual components are the same. The Al2O3 cores endow the catalysts with a high surface area, large pore size, huge pore volume, and high mechanical strength. Meanwhile, the ZrO2 shell, which consists of more and smaller tetragonal ZrO2 particles because of the large surface area of the Al2O3 core, promotes the formation of more stable sulfur species and stronger binding sites.

Data will be made available on request.The combination of green hydrogen with biogenic carbon dioxide feedstocks generates synthetic fuel with low carbon footprint [1]. So far, two main synthetic fuel routes have been extensively proposed: Power-to-Gas (Sabatier), which produces almost pure CH4; and Power-to-Liquid (Fischer-Tropsch), which aims to mimic the composition of current fossil liquid hydrocarbons (C5 +), as gasoline, kerosene, light and heavy diesel. In contrast, there is not a well-established low-carbon fuel route to produce light alkanes (C2-C4), which are now present in the fossil-based natural gas (1–10 %) and in liquefied petroleum gas (LPG) [2]. Indeed, C2-C4 hydrocarbons are vastly used (>300 MMT annually [3]) as fuel in heating appliances, cooking equipment and vehicle transport. Therefore, a novel catalytic route favouring CC coupling for the generation of a high-calorie synthetic gas (HC-SG) is of special interest for several applications and different locations.A mixture of CH4 and C2-C4 hydrocarbons composes the so-called HC-SG, which exhibits a higher heating value (∼57.72 MJ/Nm3) [4] than fossil natural gas (42–46 MJ/Nm3) and much higher than from the product of Sabatier synthesis (37.74 MJ/Nm3) [5]. As emerging fuel, there are no well-defined standards of HC-SG properties. As reference, a mixture exceeding 40 MJ/Nm3 can be considered a high-calorie gas, as it can be comparable with current natural gas specifications. To satisfy this standard, HC-SG should contain at least 5–15 vol% of C2–C4 paraffin hydrocarbons [6].In principle, HC-SG could be produced alternatively from CO and CO2 feedstocks. In this sense, the main reactions involved in the HC-SG synthesis depend on the carbon source. The direct pathway for the production of hydrocarbons is through the so-called modified Fischer-Tropsch reaction (m-FT, Eq. (1)). Nevertheless, an indirect pathway occurs when the CO2 molecule is converted to CO by means of the reverse Water Gas Shift reaction (rWGS, Eq. (2)), which generates the intermediate for the production C1-C4 hydrocarbons, similar to the Fischer-Tropsch reaction (FT, Eq. (3)). The relative extension of the abovementioned reactions (Eqs. (1) - (3)) depends on (i) the nature of the catalytic material and (ii) the reaction conditions, which need to be controlled to achieve the desired selectivity [7]. (1) nC O 2 + 3 n H 2 ⇋ C n H 2 n + 2 n H 2 O Δ H ° 298 K = - 128 KJ mol (2) C O 2 + H 2 ⇋ CO + H 2 O Δ ° 298 K = + 41 KJ mol (3) nCO + 2 n + 1 H 2 ⇋ C n H 2 n + 2 + n H 2 O Δ H ° 298 K = - 166 KJ mol Cobalt (Co) [8] and iron (Fe) [910] are suitable catalysts for the production of HC-SG. Cobalt-based catalysts have the best compromise between performance and cost for the synthesis of hydrocarbons from H2/CO mixtures [1112]. Qi et al. indicated that the synthesis of highly dispersed Co catalysts requires the initial formation of very small CoO or Co3O4 crystallites [13]. The formation of these small oxide clusters, in turn, requires strong interactions between the support and the Co precursor. Besides, Lee et al. reported that Co-based catalyst performance towards the production of C2-C4 hydrocarbons can be enhanced by the incorporation of a second metal. Recently, the effect of Mn and Ru on Co-based catalysts was evaluated [14]. They found that Mn is able to modify the surface acidity, and promote carbon-rich environment on the surface, which resulted in an increase of the C2-C4 yield. Concerning Ru, they claimed that this metal phase is able to increase the reducibility of catalysts, resulting in a high activity at a lower temperature. In other works, the combination of Co and Fe was also reported. Co-Fe/Al2O3 catalysts were more selective to light hydrocarbons (C2–C4), with respect to monometallic Co-based catalysts [15]. Furthermore, it was observed that the formation of FeCo alloy can destabilize the iron carbide phase and suppress the carbon chain growth [16].In addition to Co-Fe, other bimetallic catalysts have been proposed for the production of HC-SG from syngas, such as Fe-Ni [2], Fe-Zn [4], Fe-Cu [17] and Fe-Pd [18]. In the case of bimetallic Fe-Zn catalyst, Zn exhibited hydrogen spillover ability, which increases CO hydrogenation. Most recently, catalytic systems based on CeO2−Pt@mSiO2−Co[19], Ni3xCoxO4 [20], as well as tri-metallic Co-Fe-Ni catalysts [21] have been studied for the production of HC-SG. In this latter, Kim et al. concluded that the metal dispersion and reducibility were enhanced in the presence of nickel, leading to an improved catalytic activity.Those studies reported that the incorporation of a second metal increases the fraction of reduced metal and, consequently, its activity to HC-SG. Despite efforts to elucidate the effect of the second metal on the HC-SG reaction, some major key issues related to hydrocarbon C2-C4 promotion remained elusive. The adsorption trend of COx over cobalt catalyst should play a significant role in the product distribution in HC-SG and diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) studies can give valuable information about adsorption trends and formation of intermediates during the thermocatalytic reaction. In this direction, the beneficial role of lanthanide metal oxides on Al2O3 supports for CO2 methanation and Fischer-Tropsch reaction using nickel [22] and cobalt [23] catalysts was recently reported. In this aspect, lanthanide promotion on bimetallic cobalt catalyst would be able to incorporate moderate basic sites, and thus, facilitate COx adsorption, which arises as an interesting strategy to increase the production of HC-SG.The main goal of this work is to propose a La2O3 promoted bimetallic catalytic system able to increase HC-SG production from CO and CO2 carbon sources under moderate pressure. As well, to identify the most favourable reaction temperature conditions for each catalyst and carbon source. To the best of our knowledge, this catalytic system has not reported for HC-SG synthesis in the open literature. With this aim, we developed a series of catalysts based on Co-X/La2O3-Al2O3, using Ni, Pt and Fe as promising second active metal phases (X). The promoted support, monometallic and bimetallic catalysts were evaluated at 200–300 °C, 10 bar·g and relatively high gas flowrates. In-situ DRIFTS experiments were used to elucidate the role of the second metal when exposed to the different carbon sources. The heating values of the obtained gas mixtures and potential reaction engineering design for HC-SG production is hereby discussed.A series of micro-catalysts with particle sizes between 200 and 300 μm, were prepared by a melting infiltration method previously proposed by our group [24]. Catalyst samples were composed by 80 wt% of the promoted support (65 wt% of γ-Al2O3 and 15 wt% of La2O3) and 20 wt% of metal active phase (10 wt% Co + 10 wt% second metal Ni, Pt or Fe for bimetallic and Co for monometallic). The bimetallic catalysts were denoted as Co-Ni, Co-Pt and Co-Fe. The content of the promoter phase (15 wt%) was selected according to a previous work [22].For the impregnation of a 5 g-batch, the salt precursors (cobalt + second metal + promoter) were added to the alumina support, mixed and dissolved on a rotary evaporator at 120 °C for 1 h. In the case of Co-Pt catalyst, 3.9 mL of water were added to guarantee the dissolution of the PtCl4 metal precursor. Then, the temperature was reduced to 90 °C and vacuum was applied until complete evaporation, 4 h approximately. The impregnated material was kept at 110 °C in an atmospheric oven overnight. Subsequently, the catalysts were calcined at 450 °C for 30 min, with a heating ramp of 1 °C/min.Chemicals used for catalyst synthesis were γ-Al2O3 in shape of microspheres with particle diameters dp = 200–300 µm (Puralox) as support, salt precursor of lanthanum (III) nitrate hexahydrate [La(NO3)3·6H2O] (99.99 % purity, Aldrich) as promoter, and salt precursors of cobalt (II) nitrate hexahydrate [Co(NO3)2·6H2O] (100 % purity, Emsure), nickel (II) nitrate hexahydrate [Ni(NO3)2·6H2O] (98 % purity, Alfa Aesar), tetra platinum (IV) chloride [PtCl4] (99.99 % purity, Alfa Aesar), iron (III) nitrate nonahydrate [Fe(NO3)3·9H2O] (98 % purity, Sigma-Aldrich) as active phases.The microstructure morphology and elemental composition analysis of the catalysts were studied using scanning electron microscopy (Zeiss Auriga 60) equipped with an energy dispersive X-rays spectroscopy detector (EDX, Oxford Instruments), respectively. SEM images were recorded using the SE2 detector at a power beam range of 3 kV, working distance (WD) of 5.2 mm and a magnification of 100 X. In the case of SEM-EDX analysis, these were conducted at 20 kV using a copper standard for the system calibration. The chemical composition analysis was restricted to Co, Ni, Pt, Fe, Al, La and O, and it was calculated as the average over five measurements (standard deviation σ ± 1) on different regions for each sample.N2-physisorption (adsorption/desorption) measurements were determined at liquid nitrogen temperature using an automated TriStar II 3020-Micromeritics analyzer. Samples were degassed at 90 °C for 1 h, and then at 250 °C for 4 h in a FlowPrep 060-Micromeritics. Brunauer-Emmett-Teller (BET) method was used to calculate the BET surface area for a relative pressure (P/Po) range of 0.05–0.30. Barrett-Joyner-Halenda (BJH) method was applied to desorption branch of the isotherms to determine the average pore size and the total pore volume, which was calculated from the maximum adsorption value at P/Po = 0.999.The true densities of catalysts were studied using a helium pycnometer (Ultrapyc pycnometer 1200e, Quantachrome Instruments). Experiments were carried out on a large sample cell that was filled only the 75 % of its volume to ensure accuracy (±0.02). Prior to measurements, the cell loaded with catalyst was transferred to the sample chamber. True density values were estimated by the average of collected data points from three runs measured at 20 psi.Micrometrics Autochem II equipment was used to study the reducibility of the catalysts in the programmed temperature range from 25 to 800 °C. For the analysis, 0.1 g of each sample was placed in a U-shaped quartz reactor and supported on quartz wool. A mixture of 10 vol% H2/Argon (50NmL/min) was used as a reducing gas in the tests, while the temperature was raised from 25 °C to 800 °C with a ramp of 10 °C/min. The signal of H2 consumption was detected by a thermal conductivity detector (TCD). The amount of reduced metal oxides to metal species was calculated by integrating the reduction peaks in the H2-TPR profiles and expressed as a percentage of consumption to reduce the metal species in the catalysts.XRD patterns were collected within the 2ϴ range 20-80° in a Bruker type XRD D8 Advance A25 diffractometer using a Cu Kα radiation (λ = 1.5406 Å), a voltage of 40 kV, a current of 40 mA and a step size of 0.05° (with 3 s duration at each step). Dataset was normalized to guarantee a proper interpretation of the results. For calcined sample, the average crystallite size of Co3O4 was estimated using the Scherrer’s equation at 2Θ = 36.9°. D=(Kλ/βCosϴ), where λ is the X-ray wavelength, β is the full width of the diffraction line at half maximum (FWHM), and ϴ is the Bragg angle. On the other hand, the average crystallite sizes of the metallic Co (Co0) and alloys (CoX0) were estimated at 2Θ = 44.21° for Co [111], 44.50° for CoNi [111], 41.55° for CoPt [111], and 44.83° for CoFe [110].The metal dispersion (D) was calculated from the average metal crystallite size (M = Co0 and CoX0), by using Eq. (4). It is important to mentioned that the applicability of this equation is viable only if we assumed that the promoter phase and/or second metal phase is not present in the catalytic composition. In other words, all the catalysts should be considered monometallic Co-based catalysts with spherical uniform metal crystallite with a site density of 14.6 atoms/nm2. (4) D ( % ) = 96 / d ( M 0 ) The reactions for study of catalytic activity were conducted on a laboratory fixed-bed rector with a diameter of 13 mm and a length of 305 mm (Microactivity Reference, PID Eng&Tech). The tubular stainless-steel reactor was placed inside a ceramic chamber, which was heated by an electrical resistance. The reaction temperature was monitored using a K-type thermocouple placed in the middle of the catalytic bed. Experiments were carried out using 300 mg of catalyst, which was diluted with 3 g of silicon carbide of similar particle size (355 μm) to guarantee an isothermal catalytic bed. The mixture reactants (H2 (99.999 %, Linde), CO2 (99.999 %, Linde) and CO (99.999 %, Linde)) were supplied by mass flow controllers (MFC, Bronkshorst) at 200 N mL/min. H2:CO2 = 3 and H2:CO = 3 molar ratio was set. Thus, experiments were carried at 40.000 N mL/gcat·h of gas hourly space velocity. Pressure was set at the reactor outlet by an automatic valve at 10 bar·g.After reaction, the products passed through a cold liquid–gas separator (5 °C), where water was trapped, and then the dry flow was measured by a mass flow meter (MF, Bronkshorst). The composition of the dry gas was analysed by a micro-chromatograph Aglient Technologies 490 Micro GC Biogas Analyzer model. It was equipped with three channels, the first channel (CP-Sil 5 CB) analysed C3H6, C3H8, C4H10 and C5+; the second channel (CP-PoralPLOT U) analysed CO2, C2H4 and C2H6; and the third channel (CP-Molsieve 5A) analysed H2, CH4 and CO.Prior to reaction, catalysts were in-situ reduced under H2 flow (100 N mL/min) at 500 °C for 3 h using a heating ramp of 1 °C/min, and then cooled to 50 °C with the same ramp rate. The catalytic activity was evaluated in a range of temperature from 200 to 300 °C, with an interval of 50 °C.The conversion of COx and C1-balance selectivity toward the hydrocarbon products were calculated using Eqs. (5) - (7): (5) Conversion o f C O x % = 1 - C O x , o u t C O x , i n · 100 where CoX (x = 1 for CO and 2 for CO2) represents the molar flow rate of the species in the inlet and outlet gas. (6) Selectivity C n H m ( % ) = n · C n H m ∑ ( n · C n H m ) out + C O x , o u t · 100 where CnHm is the hydrocarbon of carbon n and hydrogen m (m = 2n + 2 for paraffins y m = 2n for olefins). (7) Selectivity t o C O x ( % ) = C O x , o u t ∑ ( n · C n H m ) out + C O x , o u t · 100 At the outlet, products such as CO, CO2 and H2 species were not considered as part of the HC-SG. Therefore, the heating values (Eq. (8)) for HC-SG were only calculated based on the NIST Chemistry WebBook data for heat of combustion of the methane and C2-C4 hydrocarbons (CH4:891 MJ/mol; C2H4:1411 MJ/mol; C2H6:1561 MJ/mol; C3H6:2058 MJ/mol; C3H8: 2220 MJ/mol; C4H10: 2878 MJ/mol) [5]. (8) Heating V a l u e MJ N m 3 = ∑ n = 1 4 Volume f r a c t i o n · H e a t o f c o m b u s t i o n MJ mol Specific v o l u m e N m 3 mol o f t h e h y d r o c a r b o n o f c a r b o n n u m b e r n DRIFTS measurements were performed on a Bruker-Vertex70 spectrophotometer equipped with a MCT detector and a high temperature reaction cell (Harrick Praying Mantis) with two ZnSe windows. Prior to the experiments, the samples were reduced at 500 °C in the reaction cell under an Ar/H2 flux. A flux of 20 mL/min with an Ar:H2:CO2 ratio of 12:3:1 was applied for the reaction with CO2 and a flux of 40 mL/min was applied for the reaction with CO with an Ar:H2:CO ratio of 12:3:1. The reactions were studied in the temperature range of 50–300 °C, at intervals of 50 °C. Background spectra were recorded under Ar at each temperature.A series of catalysts based on Co-X/La2O3-Al2O3 were prepared, characterized and evaluated. The physicochemical properties of the γ-Al2O3 support, the promoted La2O3 support, the monometallic Co catalyst and the bimetallic Co-Ni, Co-Pt and Co-Fe catalysts are described as follows.SEM-EDX analysis of Co-X/La2O3-Al2O3 revealed the presence and distribution of the metals related to the active phases (Co-X; X = Ni, Pt, Fe) and promoter phase (La), thus confirming metal impregnation. As a representative example, SEM-EDX mapping of bimetallic Co-Ni catalyst is shown in Fig. 1 . As it can be observed, Ni, Co and La elements were distributed uniformly over the support. Furthermore, the SEM image of the Ni-Co indicated that the topological characteristics (size and shape) of the bimetallic catalyst were analogous to those of the γ-Al2O3 support, i.e. micro-spherical catalysts with particle diameters between 200 and 300 μm. An approximatio of the elemental composition of the series of catalysts is presented in Table 1 . The percentage in weight of the bimetallic active phase (Co-X = 17–21 wt%) and the metal oxide promoter (La2O3 = 12–16 wt%) phase were fairly close to the nominal ones. Therefore, EDX data suggest a good and consistent impregnation of whole series of catalysts.The nitrogen adsorption/desorption isotherms of the catalysts were type IV classification (see Figure SI1) [25]. As expected, the γ-Al2O3 support presented larger BET surface area, pore volume and pore diameter than the promoted support and the rest of catalysts. The addition of La2O3 promoter together with the active phases Co or Co-Ni, Co-Pt and Co-Fe in γ-Al2O3 support resulted in a generalized reduction in their textural properties of the catalysts caused by the incorporation of non-porous metal-oxides on a porous support. Fig. 2 suggests that between Co and Co-X samples a narrow distribution in the mesoporous range was achieved, peaking higher than 6.71 nm. The true density of the catalysts was always increased after metal loading to the support.The H2-TPR profiles are displayed in Fig. 3 . At the studied reduction temperature range (T = 25–800 °C), the promoted support composed by La2O3-γ-Al2O3 showed one characteristic peak at around 450 °C, which was related to the reduction of the La2O3 promoter, whereas the reduction of γ-Al2O3 support was not identified at this temperature range. The main broad peaks for monometallic Co catalyst were categorized in two zones: a low temperature zone (250–375 °C) related to the reduction of Co3O4 to CoO and a high temperature zone (420–600 °C) related to the final reduction of CoO to Co0 [26].Bimetallic catalystsexhibited a different reduction behaviour than monometallic Co. H2-TPR profiles of bimetallic catalysts presented a deviation to lower temperatures and new reduction peaks appeared. As for Co-Ni, a shoulder located around 200–275 °C was detected and assigned to the reduction of NiO to Ni0 [27]. Compared to other bimetallic catalysts, the reduction of the PtOx species over Co-Pt catalyst was identified at much lower temperature, <200 °C [28]. Regarding to Fe-Co catalyst, the reduction peaks located around 280 °C and 400 °C were assigned to the reduction of FexOy species [29]. According to these results, it can be inferred that cobalt oxide particles have a different interaction degree with the promoted La2O3-Al2O3 support and strong Co-X bonds benefit Co reduction. It was well reported that La2O3 on Co/Al2O3 increased catalyst reducibility [30]. On the other hand, the total percentage of catalyst reduction is presented in Table 2 . At the selected reduction temperature of 500 °C, all bimetallic catalysts, except Co-Fe, showed a high reducibility (≥74 %) compared to the monometallic Co analogue (≈69 %). Furthermore, as the total reduction was only achieved for Co-Pt, it was inferred that the reduced catalyst structure of Co, Co-Ni and Co-Pt were composed by a mixture of metallic oxide particles (CoO, La2O3-Al2O3) and active metal sites in a single form (Co) for the monometallic Co and alloy form (CoNi, CoFe, CoPt) for bimetallic catalysts.The X-ray diffraction patterns of the series of catalysts in their calcined states, are reported in Figure SI2. The addition of La2O3 did not give rise to crystalline phases and only contributed to the reduction in the intensity of the γ-Al2O3 reflections. The [220], [311], [222], [400], [511] and [440] crystal planes corresponding to γ-Al2O3 phase (JCPDS:00–010-0425) were identified at 2θ = 32.35, 37.90, 39.11, 46.15, 61.25 and 67.25°, respectively. In addition to γ-Al2O3, Co3O4 phase was detected in all the Co-based catalysts. The reflections of the Co3O4 phase (JCPDS:00–043-1003) were recognized at 2θ = 31.24, 36.96, 44.83, 59.17 and 65.18°, corresponding to the [220], [311], [400], [400] and [440] crystal planes. In the bimetallic catalysts, the Co3O4 phase was shifted to the left (see Figure SI3), indicating a change in the lattice parameter of this phase. The lattice CO3O4 deviation can be caused by its interaction with the NiOx, PtOx, and FexOy atoms of the second metal phase. Furthermore, no well-defined reflections linked to the oxide phase of the second metal were detected in the bimetallic catalysts. The absence of these reflections indicates that the metal oxide species could be present in an amorphous phase, in a highly dispersed crystalline phase or in the formation of a mixed oxide. Therefore, in order to confirm the reduction of metal oxides and the formation of CoX alloys, the structural properties of all catalysts in their reduced state were also evaluated.XRD patterns of the reduced catalysts are shown in Fig. 4 . After the reduction of the samples, Al2O3 and La2O3 phases related to the support and promoter were respectively detected. The new reflection of La2O3 phase (JCPDS:00–050-0602) was located at 2θ = 28.59°. Besides Al2O3 and La2O3 phases, it was expected the presence of CoO, as most of the catalysts were not totally reduced at 500 °C, according to TPR results. However, this metal oxide phase cannot be identified over the reduced samples. The absence of this reflection was attributed to its highly dispersed crystalline phase. In contrast, metallic Co was identified in both mono and bimetallic catalysts. The [111] and [200] crystal planes of the Co phase (JCPDS:00–015-0806) were detected at 2θ = 44.21 and 51.52°, respectively. Interestingly, in the reduced bimetallic catalysts, new reflections attributed to the formation of CoX alloys were identified. The main characteristic reflections appearing at 44.50°, 41.66° and 44.83° correspond to CoNi [111] (JCPDS:00–010-8308), CoPt [101] (JCPDS:00–043-1358) and CoFe [110] (JCPDS:00–044-1483), respectively. In particular, in the reduced Co-Pt, three reflections were additionally located at 25.81, 30.59 and 34.06° and assigned to PtCl4 [131], [240] and [241] crystal planes (JCPDS:00–030-0886); indicating that the chemical precursor was still present in Co-Pt catalyst.The metallic crystallite sizes of Co and alloys (CoNi, CoPt and CoFe) were calculated from XRD patterns using the Scherrer’s equation. A crystallite size of 9.05 nm was estimated for the reduced Co catalyst. For the reduced bimetallic catalysts, the interaction of Co and the second metal (X: Ni, Pt and Fe) over promoted La2O3-Al2O3 support led to the formation of CoX crystallites with sizes higher than 10 nm, suggesting that the structure of the bimetallic Co-X phases were preferentially conformed by CoX alloys. As it is shown in Table 2, CoPt crystallite size (16.42 nm) was much higher than that estimated for CoNi (14.12 nm) and CoFe (10.43 nm), causing an inferior metal dispersion over the bimetallic catalysts. In particular, the low Pt dispersion identified over Co-Pt can be also influenced by the presence of PtCl4. This compose has measurable vapor pressure and is mobile, and therefore susceptible to segregation [31]. On the other hand, the active metal content (>8·10-6 mol/g) estimated from SEM-EDX, XRD, and TPR data, suggested that the percentage of reduction of the catalysts is a key point for their performances.The catalytic performance of the different catalyst formulations, the support and the promoted support was evaluated on the synthesis of HC-SG from both CO2 and CO as carbon sources at different reaction temperatures.All the catalysts, Co, Co-Ni, Co-Pt and Co-Fe, were active at the selected conditions and CO2 conversions always increased with temperature (see Fig. 5 ). Overall, the catalytic activity followed this order: Co-Ni > Co > Co-Fe > Co-Pt ≫ promoted support. Co-Ni was the most active, achieving a maximum CO2 conversion of 49.31 %. Therefore, the strategy of adding a second active phase only seems to be beneficial in the case of Ni, in the view of the CO2 conversion results.The main product species measured were CO, CH4, and C2-C4, whereas large C5+ hydrocarbons were not detected from CO2 hydrogenation[323232]. In contrast, the La2O3 promoted support was not able to form hydrocarbons. Fig. 6 shows the product distribution of the different catalysts and temperature conditions, and it reveals that low temperatures were preferred to produce C2-C4 hydrocarbons. It can be observed that the monometallic Co catalyst was the less selective towards C2-C4 hydrocarbons. Therefore, this catalytic behaviour revealed that the incorporation of the second metal was a positive strategy in terms of selectivity to C2-C4 hydrocarbons. At the other end, a very different mixture, which was composed by CO and C2-C4 hydrocarbons species were formed over Co-Pt catalyst. In the case of Co-Ni, the most active catalyst, it was preferentially selective to form CH4. A similar behaviour was identified for bimetallic Co-Fe, which displayed a drop in C2-C4 hydrocarbon selectivity as temperature increased. Fig. 7 shows the DRIFTS spectra recorded over the Co-based catalysts at a temperature of 250 °C using CO2 as carbon source. At this temperature, methane is the main product of the hydrogenation reaction, confirmed by its characteristic peaks at 3015 and 1314 cm−1 present in all the spectra, which is well aligned with the results obtained in the catalytic experiments. There are, however, different species adsorbed at the surface of the catalysts at every temperature that account for the different reactivity observed in the catalytic experiments. Over monometallic Co (see Figure SI4), besides generation of methane above 200 °C, carbonate species (1700–1340 cm−1) and accumulation of physisorbed water (3240 cm−1) were also observed on the surface of the catalysts. At 250 °C, a new peak was identified at 1340 cm−1 and assigned to monodentate carbonate species. This characteristic peak was also observed, although less intense, over bimetallic catalysts. It should be noted that in Co-Ni, release of methane is observed from 150 °C (see Figure SI5), proving the high activity of this catalyst towards the methanation of CO2. At the same time, the ill-defined bands between 1700 and 1400 cm−1 are attributed to the presence of carbonate and carboxylates species adsorbed on the support [33]. A comparable behaviour is observed for the Fe-Co catalyst, which displays similar and less intense peaks (see Figure SI6). On the other hand, when using the Co-Pt catalyst, coordination of CO on Pt sites is indicated by the presence of a peak at 2070 and a shoulder at 1990 cm−1 [34] (see Figure SI7). This observation is in line with the catalytic experiments, as the Co-Pt catalyst is the only one that significantly yielded CO as product at all the temperatures studied. It can be therefore inferred that the Co-Pt bimetallic catalyst facilitates the rWGS reaction [35], which explains the lower production of methane of this catalyst. Two broad bands centred at 1560 and 1375 cm−1, that decrease in intensity at higher temperatures, are attributed to the adsorption of formate species on the promoted support [36].Catalytic performance using CO as carbon source is displayed in Fig. 8 . In comparison to CO2, the use of CO as a carbon source was very advantageous in terms of gas reactivity. The achieved CO conversion was very dependent on the temperature and ranged between 0.89 and 90.65 %, much higher values with respect to CO2 conversion (<50 %). In the present reaction system, monometallic Co was more active than the bimetallic Co-X catalysts, implying that Ni, Pt and Fe are less active when CO is used as carbon source. These results are in correlation with the literature since cobalt-based catalysts are usually found as an active catalyst for mixtures H2/CO in the FTS process [37]. CO conversion on the studied catalysts at all the used temperatures complies with the following order: Co > Co-Ni > Co-Fe > Co-Pt ≫ promoted support.Selectivity from COhydrogenation is presented in Fig. 9 . Besides conversion, selectivity to C2-C4 hydrocarbons was also enhanced by the utilization of CO as a carbon source. Species such as CO2, CH4, C2-C4 and even C5 were detected in the evaluated temperature range of 200–300 °C. In this case, the promoted support was preferentially selective to form small amounts of CO2. The best results of selectivity to C2-C4 hydrocarbons were achieved over bimetallic catalysts at 250 °C, being the Co-Ni the most promising compared to Co-Pt and even more than Co-Fe. However, its important to note that in terms of hydrocarbon selectivity, the Co-Fe shows competitive values at the higher tested temperature of 300 °C, implying that Fe was beneficial to form C2+ hydrocarbons and Ni was also favourable to form CH4. Therefore, the addition of a second metal as a catalyst design strategy was proved to improve the selectivity towards the formation of C2-C4 hydrocarbons. Similar to CO2 hydrogenation, low temperatures are preferred to favour C2-C4 hydrocarbon production. Fig. 10 shows the DRIFTS spectra collected for the Co-based catalysts at a reaction temperature of 250 °C. In the hydrogenation of CO, peaks related to hydroxyl groups (400–3500 cm−1) and CO species adsorbed on Lewis acid sites (1606 and 1573 cm−1) and Brønsted acid sites (1651 cm−1) of the La2O3-Al2O3 [3839] support were identified. Furthermore, release of hydrocarbons is observed by the characteristic ν(CH) modes of methyl (CH3) and methylene (CH2) groups at 2958, 2924 and 2850 cm−1 [40 41], respectively, as well as methane at 3015 and 1305 cm−1 (see Figure SI8). Compared to monometallic Co, new peaks attributed to CO species adsorbed on Lewis acid sites (1629, 1620, 1610, 1492 and 1450) and strong Brønsted acid sites (1639 cm−1) were identified over Co-Ni [3839]. The production of CH4 and hydrocarbons (2990 and 2968 cm−1 (methyl), 2896, 2873 and 2862 cm−1 (methylene)) was mainly visible at temperatures above 200 °C [42] (see Figure SI9). Formate species detected 1585 cm−1 were related to the formation of methane as it exhibited an analogous behaviour to the methane band. In addition, the signal at 2360 cm−1 detected at all the temperatures studied is attributed to formation of gaseous CO2, which indicates that the water gas shift (WGS) reaction takes place from very low temperatures. At 300 °C, the signal of CO2 significantly increases in intensity while that of methane was maintained, and those of methyl and methylene groups even slightly decrease, which could suggest a strong competition between the WGS reaction and the FT reaction at this temperature. When the Co-Fe catalyst is used, formation of CO2 and water is observed by the broad band centred at 3250 cm−1. The series of multiple peaks between 1700 and 1200 cm−1 can be assigned to carbonate species adsorbed on the surface of the promoted support (see Figure SI10). For the Co-Pt catalyst, adsorption of linear CO species on Pt sites of different natures is detected by the presence of a peak at 2080 cm−1 and a shoulder at 2057 cm−1 [43], appearing at higher temperatures (see Figure SI11). Generation of methane and longer hydrocarbons is observed at temperatures above 200 °C by the appearance of peaks at 3015, 2960, 2930 and 2870 cm−1, along with a broad band centred at 3240 cm−1 and attributed to water, which is product of the C2-C4 formation reactions. Again, gaseous CO2 is observed by the peaks at 2354 and 2320 cm−1, as a result of the water gas shift reaction.According to these results, the promising production of CH4 and C2-C4 hydrocarbons over bimetallic Co-X can be attributed to CO adsorbed on Lewis and Brønsted active sites. For the promoted support (See Figure SI12), the peaks associated with the CO adsorption on La2O3-AL2O3 surface around 1700–1400 cm−1 was enhanced as temperature increased from 200 to 300 °C. However, with the addition of the second metal phase, the peaks of Lewis and Brønsted were different, indicating that the acid strengths differed between monometallic and bimetallic samples. Between Co and Co-Ni, the presence of new peaks and the difference in intensities indicates a difference in the amount of acid sites between, and thus in the formation of C2-C4 hydrocarbons (see Figure SI13).In summary, the main highlightsof thecatalytic resultsobtained at the selected conditions are the following: i) In both cases (CO2 or CO), competitive C2-C4 hydrocarbons selectivities were achieved using as low as possible temperatures at the expense of the conversion. ii) CO as carbon source was beneficial in terms of activity and C2-C4 hydrocarbons selectivity. iii) Co-Ni was identified as the most promising catalyst as led to an enhanced production of CH4 and C2-C4 hydrocarbon species, compared to the rest of Co-X bimetallic and monometallic Co catalyst. iv) DRIFTS experiments revealed that the chemical properties of promoted support have close relationships with the COx activation as different carbon species adsorbed on Lewis and Brønsted active sites can be identified over the La2O3 promoted Co-X based catalysts. Furthermore, it was confirmed by DRIFTS that the addition of the second metal promoted the formation of species CH4 and C2-C4 hydrocarbons. The most promising HC-SG production detected over bimetallic Ni-Co can be attributed to the long-chain hydrocarbons typically formed on Co and effectively hydrocracked by Ni, which is known to be active in C  C bond cleavage [44]. In both cases (CO2 or CO), competitive C2-C4 hydrocarbons selectivities were achieved using as low as possible temperatures at the expense of the conversion.CO as carbon source was beneficial in terms of activity and C2-C4 hydrocarbons selectivity.Co-Ni was identified as the most promising catalyst as led to an enhanced production of CH4 and C2-C4 hydrocarbon species, compared to the rest of Co-X bimetallic and monometallic Co catalyst.DRIFTS experiments revealed that the chemical properties of promoted support have close relationships with the COx activation as different carbon species adsorbed on Lewis and Brønsted active sites can be identified over the La2O3 promoted Co-X based catalysts. Furthermore, it was confirmed by DRIFTS that the addition of the second metal promoted the formation of species CH4 and C2-C4 hydrocarbons. The most promising HC-SG production detected over bimetallic Ni-Co can be attributed to the long-chain hydrocarbons typically formed on Co and effectively hydrocracked by Ni, which is known to be active in C  C bond cleavage [44].Therefore, the characterization of the materials indicated that the addition of Ni and Pt on Co-based catalyst improved its reducibility, while the addition of Fe was noticed to enhance its metal dispersion. Furthermore, the modification of the Al2O3 support with La2O3 promotes the formation of CoX alloys, favouring the hydrogenation reaction at low temperatures and controlling methane and hydrocarbon selectivity production. Finally, it can be claimed that the high catalytic activity and preferential selectivity to C4 and C2-C4 hydrocarbons of the Co-Ni was due to the formation of CoX alloy, high reducibility (73.82 %) and suitable active metal content (9.65x10-6mmol/g).A summary of the product distribution over the series of catalyst at the most promising reaction temperature is presented in Table 3 . Product distribution at the rest of temperatures can be found in supporting information, Table SI1. Competitive HC-SG mixtures were successfully achieved during the hydrogenation of CO. In particular, a gas product with a HHV of 57.90 MJ/Nm3 was achieved under CO hydrogenation and using the bimetallic Ni-Co as catalytic material at 250 °C. The HHV of the generated HC-SG was in the range of the reported ones (<57.72 MJ/Nm3), which operated at very low gas hourly space velocities (GHSV = 6,000 NmL/gcat·h) and using non-promoted bimetallic Fe-Zn/Al2O3 [4] and tri-metallic Ni-Co-Fe/Al2O3 systems [21]. Therefore, the addition of La2O3 to the traditional bimetallic system based on Co-X/Al2O3 was found to be positive, since higher GHSVs can be used during the hydrogenation of CO [23].In contrast, the use of CO2 as carbon source seems more challenging. In this case, the maximum HHV was also achieved over bimetallic Ni-Co (39.73 MJ/Nm3) at the lowest temperature, 200 °C, being significantly lower than the use of CO as carbon source. These results also reveal the lower CO2 conversion values compared to CO as carbon source. As previously described, Co-Pt catalyst favors CO and C2-C4 formation. However, experiments on CO indicated that part of the generated CO would be converted back to CO2, therefore reducing the global CO2 conversion. According to these results, it seems that the utilization of a single catalyst seems not feasible for CO2 hydrogenation as the selectivity to C2-C4 is limited or rWGS reaction to CO is favoured, restricting the HHV obtained.The implementation of a dual catalytic bed configured by two different Co-X catalysts can be an interesting strategy when using CO2 as carbon source, as reported by Gao et al. [45]. In the present HC-SG reactor engineering concept, the catalytic bed would be composed by two zones, which will work to different reaction conditions. A schematic representation is shown in Figure SI14. The first one is denoted as the CO2 decomposition zone and designed to favour the conversion of CO2 to CO and CHx (x = 1,2,3) species. In this zone, the bimetallic catalyst based on Co-Pt can be used to guarantee the reactive mixture composition. As the temperature is a key reaction condition to achieve high CO selectivities, the temperature of the catalytic bed in this zone can be fixed at 200 °C. After the first zone, in the same catalytic bed, a second zone denoted as the HC-SG formation zone is designed to favour the conversion of CO to HC-SG (CH4 and C2-C4 hydrocarbons). As Ni-Co exhibited the most promising HC-SG production, this can be the bimetallic catalyst implemented in the second zone. Compared to the previous one, the hydrogenation of CO to CH4 and C2-C4 hydrocarbons over Co-Ni should be performed at a higher temperature 250 °C.Unfortunately, the inefficient catalytic performance under the H2/CO2 mixture has been also identified by other reported HC-SG catalysts, see Table SI2. Literature suggested that reaction temperatures higher than 250 °C and pressures of 30 bar·g should be used to achieve relatively high conversions (<42 %). A comparison of HC-SNG productivity of the estate-of-the-art of catalysts is shown in Fig. 11 . For both cases, CO2 or CO as a carbon source, the relationship between GHSV and HC-SG selectivity positioned the Co-Ni as a rentable material since significant productivity can be achieved by the implementation of technically feasible reaction conditions, leading to an HC-SG process economically profitable to be scaled-up at industrial levels. The productivity of the Co-Ni was around 8.08x102 mL/gcat·h using CO2 and 5.15x103 mL/gcat·h using CO.In this work, HC-SG synthesis was performed over a series of bimetallic Co-X (X = Ni, Pt and Fe) catalysts for the selective production of CH4 and C2 − C4 hydrocarbons from CO2 and CO as carbon sources. Catalytic results indicated that the utilization of CO as carbon source is very positive in both conversion and C2-C4 hydrocarbon selectivities. Among catalysts, Co-Ni was the most promising catalyst for production of HC-SG. Therefore, the strategy of adding a second metal proved to the positive. At H2/CO = 3, T = 250 °C, and P = 10 bar·g, very interesting selectivities to CH4 (40.01 %) and C2–C4 hydrocarbons (50.04 %) were obtained, with a reduced selectivity to CO2 (5.05 %) and C5+ (4.89 %) formation. In this direction, a competitive HC-SG with a heating value of 57.90 MJ/Nm3 was achieved using Co-Ni bimetallic catalysts.The successful catalytic performance was attributed to the acid-basic sites formed on the catalyst surface by the synergic effects caused by the presence of La2O3 and CoNi alloy phases, which favours in the production of CH4 and C2-C4 hydrocarbons under lower temperatures. Besides, the bimetallic Ni-Co catalyst showed higher reducibility (73.82 %) and active metal content (9.65x10-6mmol/g). The findings from this study contribute to our understanding of the low temperature CO2 and CO hydrogenation activities ofLa2O3 promoted Co-X/Al2O3 based catalysts and provide insights for the design of materials for HC-SG production.Bimetallic Co-Ni catalyst can be used as a benchmark to optimize or design novel reactor approaches for the HC-SG process intensification. In the case of using CO2 as carbon source, an adapted HC-SG reactor concept, configured by two Co-X catalytic zones is proposed to promote the use of CO2 as carbon source. A first Co-Pt catalytic zone operated at 200 °C to favour the conversion CO2 to CO and CHx (x = 1,2,3) species, followed by a second Co-Ni catalytic zone operated at 250 °C to achieve the conversion CO to HC-SG (CH4 and C2-C4 hydrocarbons). However, further studies should be carried out for the validation of this reactor engineering concept in a full-scale reactor. In any case, the use of bimetallic catalysts is interesting to divert selectivity towards the most desired products on each occasion. Andreina Alarcón: Writing – original draft, Investigation, Formal analysis, Visualization. Olatz Palma: Investigation, Validation. Elena Martín Morales: Investigation, Formal analysis, Writing – review & editing. Martí Biset-Peiró: Methodology, Resources. Teresa Andreu: Conceptualization, Validation, Supervision, Funding acquisition, Writing – review & editing. Jordi Guilera: Writing – original draft, Conceptualization, Methodology, Resources, Supervision, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by project TED2021-132365B-I00, funded by MCIN/AEI/10.13039/ 501100011033 and by the European Union “NextGenerationEU”/PRTR and PID2019-108136RB-C33 (MCIN/AEI/10.13039/501100011033). Andreina is grateful for support by the Margarita Sala Grant funded by the University of Barcelona (UNI/551/2021). The authors thank SASOL for kindly providing alumina support material (Puralox). Authors kindly thank Dr. Albert Llorente for assistance withthe characterization of materials.Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuel.2023.127726.The following are the Supplementary data to this article: Supplementary Data 1

A new catalytic route for the production of a high-calorie synthetic gas (40–60 MJ/Nm3), composed by C1-C4 hydrocarbons, has industrial interest for gas applications and locations with high heating requirements. In this work, a series of bimetallic Co-X (X = Ni, Pt and Fe) catalysts supported on La2O3 promoted Al2O3 micro-spheres were evaluated using both CO2 and CO carbon sources under mild temperature (T = 200–300 °C), moderate pressure (P = 10 bar·g) and relatively high gas hourly space velocity (40,000 N mL/gcat·h). Experimental results proved that the incorporation of nickel as a second metal is beneficial for high-calorie gas application. Besides, catalytic results showed that the utilization of CO as carbon source is beneficial in both conversion and C1-C4 hydrocarbon selectivities. Co-Ni presented the most interesting results, leading to a heating value of 57.9 MJ/Nm3 (40.01 % CH4 and 50.04 % C2-C4 hydrocarbon) at 250 °C through CO hydrogenation. The enhanced catalytic performance achieved over bimetallic Co-Ni was attributed to CoNi alloy catalytic activity, high reducibility (73.82 %), active metal content (9.65x10-4 mmol/g) and appropriate acid-basic sites for COx activation. In contrast, the conversion of CO2 to high-calorie gas was found to be more challenging and lower gas heating values were achieved (39.73 MJ/Nm3). In this case, an adapted reactor concept using a dual bimetallic catalyst and different reaction conditions is hereby proposed to shift selectivity towards the targeted products. This findings represent a step forwards in catalytic engineering for the development of high-calorie synthetic gas reactors.

Conversion of low-cost biomass-derived oxygenates, such as glycerol (C3H6O8), to hydrogen (H2) is a promising route for making eco-friendly renewable H2 fuels, being able to increase the share and availability of clean energy while lowering the greenhouse gas (GHG) emissions [1]. Glycerol is commonly produced as an organic waste from biodiesel production processes over homogeneous alkaline catalysts (via transesterification of lipids such as plants oils and/or animal fats with alcohols), and hence conversion of glycerol to H2 can undoubtedly improve the economics of biodiesel production [2,3]. Steam reforming of glycerol (SRG) is a promising and industrially important reaction, in which renewable H2 or synthesis gas can be produced to valorise glycerol [4]. Importantly, H2 produced from SRG could not contribute to global warming based on the assumption of using crude glycerol generated as a biowaste of biodiesel mass-production. According to the overall reaction of SRG (Eq. 1), the relative proportion of hydrogen in crude glycerol (theoretically, 7 mol of hydrogen which can be produced from every 1 mol of crude glycerol) makes it not only an advantageous option for producing renewable H2, but also more economically and environmentally competitive compared to fossil fuels, e.g., methane [5]. Additionally, steam reforming is a well-established technology, indicating that the feasibility of shifting the current fossil feedstock to glycerol without significant modification of the current infrastructure. (1) C 3 H 8 O 3 + 3 H 2 O → 3 CO 2 + 7 H 2 Δ H 298 = 127.7 kJ / mol SRG is endothermic, which is favoured at high reaction temperatures (>600 °C [4,6]), and hence is inevitably associated with catalyst sintering and coking issues. Also in comparison with fossil hydrocarbons the reaction network of SRG is rather complex (as shown in Table S1 in the Supporting Information, SI [4,5]), with a higher susceptibility to undesired products formation through many side reactions, which could affects the overall selectivity and yield of H2 [7].Ni-based catalysts are common reforming catalysts because of their good ability to cleave CC, OH and CH bonds in chemical reactions and low cost. However, the performances of Ni-based catalysts frequently depletes from rapid deactivation caused by Ni aggregation (due to sintering) and carbon deposition (induced by undesired reactions such as thermal decomposition of methane and CO reduction, Table S1) in reforming catalysis at high temperatures (typically >800 °C) [8,9] such as SRG and methane dry reforming. Therefore, various strategies have been exploited to develop anti-coking and anti-sintering reforming catalysts. Stabilisation of Ni nanoparticles (NPs) and regulation of the particle sizes can be achieved via (i) appropriate selection of promoters [10–12], (ii) use of porous supports [13], (iii) doping with the second metallic phase (i.e., bimetallic catalysts) [14–16] and (iv) design of unique catalysts structures such as core-shell to improve Ni dispersion and reduce coke deposition on the Ni-based catalysts in various reforming reactions [17–21].In SRG, modification of the Ni@γ-Al2O3 catalyst was achieved by exploring the benefits of different preparation procedures and employing promoters to improve the catalyst surface and prevent carbon formation [22,23]. For example, CeO2 was used to dope the conventional Ni@Al2O3 (i.e. Ni@12Ce-Al2O3) catalyst, which enhanced Ni dispersion on γ-Al2O3 support with relatively smaller Ni NPs and showed the improved performance in SRG, and high ability to resist coking compared to the benchmark Ni@γ-Al2O3 [23]. Recently, encapsulation strategies based on porous materials are shown to be promising to prepare highly dispersed yet segregated metal particles, which can prevent metal particles sintering and carbon deposition in high-temperature catalytic reactions effectively including reforming reactions [24], which were exemplified by silicalite-1 zeolite encapsulated Ni catalyst for dry reforming of methane with CO2 [25].Zeolites as a class of porous materials are suitable supports to confine metal NPs, mainly owing to their high porosities and large surface areas [26,27]. However, due to the intrinsic microporosity of zeolites and the associated accessibility/diffusion issues, metal-supported zeolite catalysts prepared using the conventional impregnation procedures often have distribution of large metal-particles on the external surface of the zeolite with poor dispersion of metal particles, which are not ideal for catalysis. Conversely, metal precursors inclusion can also be integrated with the zeolite synthesis cleverly to ensure the encapsulation of metal NPs after reduction. In such encapsulated metal NPs catalysts, segregation of metal NPs using the inorganic crystalline framework can mitigate sintering and improve stability effectively during catalysis. Additionally, if the dimension of the encapsulated metal NPs could be managed, the activity and selectivity of the catalysis can be tuned as well. Although there are many benefits, most of the metal NPs are confined in space with the protective and microporous zeolitic shell, and hence diffusion limitation in these zeolite encapsulated metal catalysts can potentially jeopardise their performance in catalysis [28].Diffusion issues in such zeolite encapsulated catalysts can be addressed by introducing mesoporous structures in them, which can reduce the average diffusion length and enhance the accessibility to the encapsulated metal NPs, being beneficial for catalytic performance and reaction kinetics [29]. Shiwen Li et al. [30] prepared Ni, Co and Cu encapsulated in mesoporous MFI zeolites with hollow structures for catalytic reduction of hydrocarbons particularly toluene and mesitylene with kinetic diameters of 0.58 and 0.87 nm, respectively. The findings confirmed that the catalytic performance is strongly linked to the characteristic diffusion behaviour of molecules over the MFI zeolite layer, with the silicalite-1 encapsulated Ni NPs being a highly active phase for toluene reduction. In catalytic SRG, the reactant of glycerol has a kinetic diameter of ~0.60 nm, therefore, diffusion resistance through a zeolite-based catalyst, including the encapsulated ones, can be expected, and introduction of mesoporous structure into the catalysts can be beneficial to the catalysis.Herein, a strategy of developing Ni NPs encapsulated in mesoporous hollow silicalite-1 zeolite catalyst (i.e., Ni@HolSi-1) was explored. In detail, the conventional encapsulated Ni catalyst (i.e., Ni@Si-1) was prepared via one-pot hydrothermal synthesis of silicalite-1 in presence of the Ni precursor, and the mesoporous hollow structure was achieved by treating the prepared Ni@Si-1 catalyst with tetrapropylammonium hydroxide (TPAOH) solution. The encapsulation strategy enabled the formation of highly dispersed ultra-small Ni NPs, and the post-treatment rendered the formation of mesoporous hollow structure, which were highly beneficial to catalytic SRG. The physiochemical features of the investigated catalysts were determined comprehensively by employing various methods and techniques. Comparative and systematic catalytic SRG was performed over the developed catalysts to assess their performance regarding conversion of glycerol, H2 yield/selectivity, and distribution of CO2, CO and CH4, respectively. The findings show that the developed Ni@HolSi-1 catalyst was catalytically active and stable in SRG even after 100 h on stream, as well as being coke resistant.Pristine Si-1 zeolite was prepared via crystallisation of the synthesis solution, consisting of tetraethyl orthosilicate (TEOS, Sigma-Aldrich, ≥99.0%), tetrapropylammonium hydroxide solution (TPAOH, Sigma-Aldrich, 25 wt% in H2O) and deionised water, according to a previously reported method [31]. Typically, a mixture, containing TPAOH (~13 g) and deionised water was prepared under continuous stirring at room temperature (RT). Then 8.32 g of TEOS was gradually introduced into the aforementioned mixture under continuous stirring in a Teflon beaker (to hydrolyse TEOS fully). The mixture was then transferred and heated in a stainless-steel autoclave lined with Teflon, under the constant hydrothermal conditions (at 170 °C and 96 h). After crystallisation, the final product was recovered through centrifugation, and rinsed with distilled water and ethanol several times before being dried overnight and air-calcined for 8 h at 550 °C.A control catalyst of Ni/Si-1 was synthesised using the conventional method of incipient wetness impregnation (IWI). To impregnate 2 g of calcined Si-1 zeolite containing 5 wt% of theoretical Ni loading, the prepared aqueous solution of Ni precursor salt (i.e. nickel nitrate hexahydrate (Ni[NO3]2·6H2O ≥ 99%, Sigma-Aldrich) was used to develop a solid product, which was then oven-dried overnight and calcined in air for 8 h at 550 °C.The encapsulated Ni catalyst was synthesised in a single pot by hydrothermal procedure under the same condition using the same protocol for preparing Si-1 zeolite except the addition of the pre-prepared [Ni(NH2CH2CH2NH2)3](NO3)2 solution which was employed as the precursor to achieve the encapsulation of 5 wt% theoretical Ni loading during the synthesis of Si-1 crystals [25,31,32]. Specifically, 0.95 g of Ni precursor salt (Ni(NO3)2·6H2O, ≥99%) was dissolved into 10 mL of aqueous solution containing 2 ml of ethylenediamine (NH2CH2CH2NH2) until it was fully completed under continuous agitation at RT. Finally, the prepared gel was obtained with a molar proportion of 1 SiO2: 0.4 TPAOH: 35 H2O: 0.045 [Ni(NH2CH2CH2NH2)3]2+).Ni@HolSi-1 was prepared using a post-synthetic treatment method with Ni@Si-1 as the parent material and TPAOH solution [33]. In detail, 20 mL of 0.3 M TPAOH aqueous solution was used to treat the as-synthesised Ni@Si-l inside an autoclave at 170 °C for 24 h. After the treatment, the obtained solid product was separated from the solution at RT by centrifugation, rinsed multiple times with deionised water and ethanol, and oven-dried overnight at 80 °C. Then, the Ni@HolSi-1 catalyst was achieved following similar drying and calcination conditions as described previously.The crystal patterns of the calcined and reduced catalysts were measured by Powder X-ray diffraction (XRD) on a Philips X'Pert X-ray diffractometer by employing a CuKα1 X-ray source radiation. The analysis of all XRD patterns of the catalysts were matched and compared with the known materials data available in the database (ICDD, JADE 6 software, Materials Data Inc., Livermore, CA). Quantachrome Quadrasorb instrument was employed to detect the N2 adsorption-desorption of the catalysts at −196.15 °C, and their textural properties covering specific surface areas (S BET) and pore information of the catalysts. Before N2 physisorption, all the catalysts were pre-treated in a vacuum at 350 °C for 24 h. The Brunaur-Emmett-Teller (BET) was used to calculate the S BET, and total specific pore volume (V total) was determined using the adsorbed quantity of N2 at the relative pressure of p/p 0  = 0.99. Similarly, t-plot method was used to calculate the specific micropore volume (V micro), whereas the V micro is taken off the V total to determine the specific mesoporous volume (V meso). Hydrogen temperature programmed reduction (H2-TPR) profiles of the calcined catalysts were examined on a ChemBET Pulsar TPR/TPD equipment (ChemBET-3000), and all the catalysts were initially degassed as reported elsewhere [23]. Comparative characterisation of the surface acidity of fresh catalysts was carried out using NH3 temperature-programmed desorption (NH3-TPD) experiments in the same instrument employed for H2-TPR measurements, as previously described elsewhere [23]. For each catalyst under investigation, the surface areas and dispersions of Ni were determined using the same H2 pulse chemisorption methodology, as previously reported elsewhere [23]. Inductively coupled plasma optical emission spectrometry (ICP-OES, Plasma Quant PQ 9000) was used to measure the actual quantity of Ni available in the calcined catalysts. Typically, the catalysts were microwave-digested in a mixture of acidic solution prior to ICP analysis. Bruker Vertex 7.0 fourier transform infrared (FT-IR) spectrometer was employed to collect the spectroscopies of the calcined catalysts, with a scanning wavenumber (ranging from 400 through 4000 cm−1) and a spectral resolution within the interval of 4 cm−1. Electron microscopes such as scanning electron microscope (SEM, Tescan Mira FEG operating at 5 kV accelerating voltage) and high-resolution transmission electron microscope (HRTEM, FEI Tecnai G2 F20 electron microscope at 300 kV) were used to examine the morphologies of the catalysts. Prior to SEM measurement, all catalysts were coated using platinum (Pt) metal and the corresponding EDX spectrum (energy-dispersive X-ray spectroscopy) were detected with an Oxford Ultim® Max system. ESCALAB™ 250Xi electron spectrometer (Thermo Scientific) was used to analysed the X-ray photoelectron spectroscopy (XPS) spectra of the catalysts using Al-Kα (hv = 1486.6 eV) as the source of X-ray radiation during the measurements. C1s peak was used as the XPS calibration standard with a fixed binding energy (BE, at ~284.6 eV). Thermogravimetric analysis (TGA) experiments were conducted on a thermal system (TGA 550) in air atmosphere (at 40 mL min−1) with a heating rate of 10 °C min−1 from RT to 900 °C. Before the analysis, all species including adsorbed water and other physisorbed molecules were removed from the spent catalysts, and the amount of carbon deposition was determined based on the derivative weight reduction at temperatures above 500 °C.Catalytic SRG over different catalysts was carried out in a continuous-flow quartz tubular-reactor (12 mm I.D. × 450 mm length) and atmospheric pressure, as described previously elsewhere [23]. Specifically, the temperature of the packed bed was observed using a K-type thermocouple (OMEGA®), and the bed was supported by quartz wool in a tubular furnace (Carbolite, EVT-12). Typically, ~200 mg of each catalyst was crushed and packed in form of pellets (~250–425 μm), and then treated in situ from RT to 800 °C under H2 flow (at 100 mL min−1 STP). Then the carrier gas (N2 at 50 mL (STP) min−1) was employed to sweep the H2 gas until the temperature of the bed was completely set to the required reaction temperature. Following that, a mixture of steam and glycerol (SGFR = 10:1) was injected into a vaporiser that was held at 320 °C at a continuous feed flow rate of 6 mL h−1, using a syringe pump (Harvard Apparatus, PHD ULTRA). During SRG, the flowrates of gases (with a total gas hourly space velocity (GHSV) of 3120 h−1 (STP)) were maintained using mass flow controllers (MKS instruments). The outlet stream of the reaction was analysed one hour after the reaction condition was changed, in the temperature interval 500–750 °C, using 50 °C increments) under steady-state conditions. Gas chromatography (PerkinElmer Clarus® 580) equipped with HayeSep DB 100/120 mesh and Shin Carbon ST 100/120 mesh columns, thermal conductivity (TCD) and flame ionisation (FID) detectors, was used to examine the gas composition of the outlet stream. The condensable products from SRG (for example, unreacted glycerol) were continuously collected at the reactor's outlet stream by a water trap cooled by a circulating bath. Subsequently, the condensed unreacted glycerol was analysed off-line by GC (Agilent 7820A, in a 30 m long, 0.32 mm I.D. Stabil-wax column, operated with the program: 5 min at 100 °C, heating to 10 °C min−1 up to 180 °C, keeping this temperature for 15 min, using FID). The performance of catalytic SRG was evaluated by measuring the dry gas flowrate with a bubble flowmeter, and the glycerol conversion (X G ), selectivity (S) and yield (Y) of the gaseous products were specified in Table 1 . Kinetic experiments were performed to establish the specific reaction rate and activation energy for SRG systems employing different catalysts over temperatures ranging from 300 to 450 °C. However, in order to avoid the diffusion limitation during SRG reaction and obtain relevant information close to intrinsic kinetic, a tubular reactor containing a small amount of the catalysts under investigation (i.e., 12 mm I.D. × 12.5 mm length, with 100 mg of the pelletized Ni/Si-1, Ni@Si-1 and Ni@HolSi-1 catalysts and bed volume of about 1.4 cm3) was used for kinetic studies (with a flow rate = 12 ml (STP) h−1, GHSV = 10,360 (STP) h−1) at low conversions below 20% to preclude the effect of mass and heat transfer limitations.XRD technique was used to characterise the crystal structures of the Si-1, the calcined and reduced catalysts, and their corresponding diffraction patterns are illustrated in Fig. 1 . In the case of all calcined samples, the diffraction phases associated with the MFI-type zeolite framework was well-recognised at 2θ < 40° (JCPDS 44–0696), proving that the retained silicalite-1 phase after the thermal treatments (Fig. 1a). The diffraction peak associated with NiO species (at 2θ = ~43.4°) was detected clearly in the structures of the calcined Ni/Si-1 catalyst developed by the impregnation method (shaded by the orange rectangle in Fig. 1b), whilst it was not found in Ni@Si-1 and Ni@HolSi-1, which were prepared via encapsulation. This suggests the presence of homogeneously distributed NiO species within the Si-1 framework of the two catalysts. After reduction, the crystalline structure of Si-1 support was maintained in the catalysts, as shown in Fig. 1c. Same findings regarding the structural integrity of Si-1 were also obtained by FT-IR analysis (Fig. S1). Fig. 1d shows that the corresponding diffractions phases of metallic Ni (at 2θ = ~44.8°, shaded by the orange rectangle) was measured in the reduced Ni/Si-1 rather than the encapsulated Ni@Si-1 and Ni@HolSi-1 catalysts, showing the possible formation of widely dispersed small Ni NPs in Si-1 zeolite. Fig. 2 and Fig. S2 shows the morphologies and the associated EDX mapping assessment of the catalysts under investigation. The morphology of Si-1 zeolite was very spherical in shape with an average crystal sizes of ~0.28 μm (Fig. S2). After Ni impregnation, the impregnated Ni/Si-1 catalyst has lower particle sizes of about ~0.18 μm to that of Si-1 (Fig. 2a), and a large amount of Ni species was found on Ni/Si-1, as observed in the EDX mapping (Fig. 2b). The encapsulated Ni@Si-1 particles via the one-pot synthesis have larger sizes (~1.9 μm, Fig. 2c) in comparison to the particle sizes in Ni/Si-1. Comparatively, the particle size of the Ni@HolSi-1 catalyst after the post-synthetic treatment (of Ni@Si-1) was reduced at ~0.30 μm, as shown in Fig. 2e. EDX analysis of Ni@Si-1 and Ni@HolSi-1 (Fig. 2d and Fig. 2f) show that the surface Ni species in the two catalysts are less dense than the one of Ni/Si-1, showing that proportion of Ni species were likely encapsulated within the Si-1 framework which cannot be detected by surface EDX elemental mapping. Fig. 3a presented the N2 adsorption-desorption isotherms, and the associated textural and structural characteristics of the materials are summarised in Table 2 . Physisorption isotherms of Si-1, Ni/Si-1 and Ni@Si-1 exhibit the shape close to Type-1 isotherm for microporous materials. Conversely, a H2 hysteresis curve (closes at p/p 0  = 0.45 in the desorption branch) was found in the isotherm of Ni@HolSi-1 (type IV), which proves the presence of mesoporous structure in the catalyst. The Ni/Si-1 with values of the S BET and V total (S BET = 522 m2 g−1 and V total = 0.56 cm3 g−1) are slightly lower than that of Si-1 (S BET = 698 m2 g−1 and V total = 0.61 cm3 g−1, Table 2) which could be due to pore clogging caused (by Ni deposition, as shown by the reduced V micro, values from 0.23 to 0.16 cm3 g−1). Synthesis of Si-1 in the presence of Ni precursor caused the decrease in the BET specific surface area, i.e., S BET = 347 m2 g−1 for Ni@Si-1, as well as the reduced micropore volume (V micro = 0.10 cm3 g−1) in comparison with that of Si-1 (Table 2). Compared to Si-1, the micropore volume in Ni@Si-1 was decreased by ~57%, indicating that after the one-pot hydrothermal synthesis, the encapsulated Ni clusters could preoccupy some spaces inside the Si-1 zeolite (e.g., species of metal confined within zeolite crystals and/or encapsulated within hollow cavities in zeolite crystals [34]), and hence affecting the porous structure. After the post-treatment (of Ni@Si-1) using TPAOH, the resulting Ni@HolSi-1 showed significant mesoporous features, as evidenced by the increased mesopore volume (of 0.5 cm3 g−1), being much more significant that the parent Ni@Si-1 (V meso = 0.24 cm3 g−1). Also, the well-developed mesopores features in Ni@HolSi-1 was also reflected based on the comparison of pore size distributions (PSD) of the materials under study (as illustrated in Fig. 3b), in which Ni@HolSi-1 shows the PSD of mesopores centred at about 5 nm.TEM measurement was undertaken to explore the microscopic feature of Ni NPs in the developed catalysts. The TEM images of the Ni/Si-1 catalyst (Fig. 4a and b) show the presence of Ni particles with an average size of 2.9 ± 0.9 nm, suggesting the possible Ni NPs location on the surface of the Si-1 crystals (Figs. 4b). When compared with Ni/Si-1, the existence of Ni NPs in the encapsulated Ni@Si-1 catalyst (Fig. 4c and d) and Ni@HolSi-1 (Fig. 4e and f) catalyst could not be clearly identified by the current TEM analyses. This might be due to the dispersion of Ni NPs that are encapsulated within the Si-1 crystals. Based on the TEM micrograph of Ni@HolSi-1, clearly, after the post-treatment of the encapsulated Ni-based catalyst (i.e., Ni@Si-1) with TPAOH solution, large void structures were formed in the interior crystals of Si-1 zeolite due to silicon extraction under alkaline conditions and subsequent recrystallisation in presence of TPA+. The TEM results correspond well to the findings by N2 physisorption analysis discussed above.All the catalysts in this work were prepared with the 5 wt% theoretical Ni loading, and the quantified Ni content of the catalysts (by ICP-OE) are lower than the value as shown in Table 3 . H2 pulse chemisorption analysis was used to determine the Ni dispersion and metallic surface area of the catalysts, and the corresponding findings are summarised in Table 3. Regarding the measured metallic Ni surface area and Ni dispersion, the impregnated Ni/Si-1 and encapsulated Ni@Si-1 catalysts are rather comparable. Conversely, the Ni@HolSi-1 catalyst with the encapsulated Ni and mesoporous hollow structures demonstrated the highest Ni dispersion and metallic surface area at 1.4% and 9.3 m2 gNi −1, respectively, as presented in Table 3. Specially, in comparison with the parent Ni@Si-1 1 (with 0.30% Ni dispersion and 1.7 m2 gNi −1 metallic Ni surface area), the increase in the metallic Ni dispersion and surface area of Ni@HolSi-1 suggests that the post-synthetic TPAOH treatment can improve the exposure of Ni phases significantly, which can potentially benefit catalysis.The acidic characteristic of the materials was investigated by NH3-TPD analysis, as presented in Fig. 5a, Fig. S3 and Table S2. According to results indicated in Fig. 5a and Fig. S3, all materials show the presence of weak suface acidity, which can bind NH3 and leads to the assocaited NH3 desoption at 100–350 °C during NH3-TPD analyses. Comapratively, the measured total surface acidity of the Si-1 support (~3401 μmol gcat −1), the impregnated Ni/Si-1 (~5623 μmol gcat −1), and the encapsulated Ni@Si-1 (~4112 μmol gcat −1) are rather similar, whilst that of Ni@HolSi-1 is much lower at ~1491 μmol gcat −1. The reduced surface acdity of Ni@HolSi-1 could be due to the post-treatment, which can potentially benefit reforming reactions since the presence of acidity in the catalysts tends to encourage coking. The reducing behaviour of Ni species in the synthesised catalysts was probed by H2-TPR analysis, and the relevant peaks are shown in Fig. 5b. For the impregnated Ni/Si-1 catalyst, the major reduction peak of the Ni species in it was at about 465 °C, and it can be attributed to the reduction of bulk NiO crystallites situated on the external surface of the Si-1 zeolite support with weak interaction [35]. For the encapsulated Ni@Si-1 and Ni@HolSi-1 catalysts, the major reduction behaviour of the Ni species in them occurred at about 790 °C, which are much higher than that of Ni/Si-1, suggesting an improved interaction between the encapsulated Ni species and Si-1 framework [36,37]. Based on the findings above, the NiO phases in the encapsulated catalysts are more resistant to thermal reduction (under H2) in comparison with the large NiO crystallites deposited on the outer surface of Si-1 material, due to the stabilisation of the encapsulated NiO species within the Si-1 framework [16,38]. Also, based on the findings from the H2-TPR analysis, a temperature of 800 °C was set to reduce the Ni catalysts under investigation for catalytic SRG.The chemical surface states and the position of Ni phases in the calcined catalysts were examined using XPS, and the XPS survey scans of the relevant catalysts are illustrated in Fig. 6 . As depicted in Fig. 6a, the band associated with Ni in the impregnated Ni/Si-1 was much stronger than that of the encapsulated Ni@Si-1 and Ni@HolSi-1, respectively. The finding confirms that the Ni phases in Ni/Si-1 are mostly on the outer layer of the Si-1 material since XPS is the technique for probing relevant surface properties in the outermost 2–10 nm of a solid surface. Similarly, high-resolution XPS spectra of the Ni phases (as presented in Fig. 6b), show that all the developed catalysts exhibit a Ni 2p3/2 as the main peak (at the binding energy, B.E., of about 852–859 eV) along with the associated shake-up satellite peak at the B.E. of 859–871 eV and a Ni 2p1/2 peak (at the B.E. of about 871–876 eV) along with the satellite peak at the B.E. of 876–888 eV, respectively. Meanwhile, in all the developed catalysts, the characteristics peaks at a lower B.E. (i.e., ~855–857 eV and ~ 872–874 eV) representing the proportion of NiO species. To compare the developed catalysts in this work with the literature data, the B.E. values (Ni2+ 2P3/2) of the developed catalysts are in between that corresponding to the B.E. of the pure NiO phase (i.e., Ni2+ 2p3/2, at about ~854.4 eV) and that of pure NiAl2O4 phase (i.e., Ni2+ 2p3/2, at about ~857.3 eV) [39]. Furthermore, the Ni2+ B.E. of Ni@HolSi-1 (at about 55.6 eV) was rather comparable to that of the reference NiO phase, suggesting a higher proportion of Ni2+ in the form of NiO oxide than on the surface of the catalyst in spinel structure. The developed Ni/Si-1 catalyst prepared by impregnation exhibits relatively high B.E. values which are close to that of the spinel NiAl2O4 (i.e., 856.7 eV versus 857.3 eV). Comparatively, the intensity of the Ni 2p regions observed in Ni/Si-1 is much higher than that of the encapsulated Ni@Si-1 and Ni@HolSi-1 catalysts, which confirms that most of the Ni species in Ni/Si-1 are positioned on the outer surface of the Si-1 support. Considering Ni@Si-1, the Ni species are likely dispersed uniformly within Si-1 framework including the top layer of the catalyst (within 10 nm), which is proved by the relatively low signal intensity (note that all three catalysts have comparable actual Ni loadings, Table 3). Regarding Ni@HolSi-1, the intensity of its Ni2p regions is the lowest, indicating that bulk of the Ni species are situated within the hollow Si-1 structures (Fig. 4f) due to the dissolution-recrystallization mechanism of the post-treatment of Si-1 zeolite using TPAOH aqueous solutions.The performance of catalytic SRG over the catalysts under investigation was comparatively evaluated with respect to glycerol conversion, product selectivity and yield towards the desired H2 versus CH4 and CO2 versus CO, under steady-state conditions at 500–750 °C and atmospheric pressure. The gaseous products of H2, CO2, CO and CH4 were detected at the end of each run during the catalyst measurement. Based on these observations, the conversion of glycerol could be attributed to the transformation of the glycerol molecule into H2 formation via SRG (Eq. 1). The promotion of H2 generation and suppression of the formation of CH4 and CO was accompanied by several side reactions (as shown in Table S1) such as CH4 reforming into (Eq. S2), production of CO and H2 through thermal decomposition of glycerol (Eq. S3), water-gas shift reaction (WGSR) of converting CO into CO2 along with H2 (Eq. S4) and CO and CO2 methanation reactions as shown in Eqns. S5(a) and 5(b), respectively. Fig. 7 showed the glycerol conversions and selectivites/yields of H2, CO2, CO and CH4 as a function of the reaction temperature in the stream. Noticeably, glycerol conversion increases with reaction temperature in all the developed catalysts displayed in Fig. 7, which is in accordance with the temperature dependence of the reforming reactions. However, as shown in Fig. 7a and b, the pristine Si-1 showed insignificant glycerol conversion of <10% and hydrogen yield of <2%, respectively, which could be attributed to the non-catalytic gas-phase reactions under thermal conditions. The product distribution of gaseous products (under consideration in this work) were determined for the non-catalytic thermal system, as depicted in Fig. 7c–7f, which shows that the selectivity to CO was the highest at ~60% over the temperature range due to thermal decomposition of glycerol (Eq. S3). Production of H2 was enabled at T > 500 °C with the selectivity of ~20%. Considering the stoichiometry of Eq. S3, selectivity to H2 is relatively low. Since the selectivity to CO2 was <5%, the desired SRG reaction (Eq. S1) is unlikely. CH4 was also produced with the selectivity of 15–20%, suggesting the presence of CO methanation (Eq. S5a).In catalytic SRG, compared to the thermal case, the conversion of glycerol and yield of H2 over the encapsulated catalysts (of Ni@Si-1 and Ni@HolSi-1) increased as a result of the reaction temperature rise from 500 to 700 °C, as shown in Fig. 7a and b. The selectivity to various gaseous products in the two systems over the two encapsulated catalysts was rather comparable, but Ni@HolSi-1 showed better performance than Ni@Si-1, especially at high temperatures. At temperatures of 600–650 °C, the Ni@HolSi-1 catalyst (with glycerol conversions of >90% and hydrogen yield of >40%) outperformed the Ni@Si-1 (with glycerol conversions of <85% and hydrogen yield of <40%). This can be attributed to (i) the highly dispersed Ni NPs due to the encapsulation strategy (Fig. 4 and Table 3) and (ii) the presence of mesoporous hollow structures in the Ni@HolSi-1 catalyst (as indicated in Figs. 3 and 4), which could improve the molecular diffusion through the catalyst structure. Conversely, the impregnated Ni/Si-1 catalyst showed the comparatively lowest glycerol conversion of <75% and insignificant hydrogen yield of <30%, respectively. The gaseous product distribution of catalytic SRG (H2, CO2, CO and CH4) over different catalyst were investigated, as shown in Fig. 7c–7f. By comparing the selectivity to different gaseous products at 750 °C, the impregnated Ni/Si-1 catalyst presents the lowest selectivity to H2, CO2, CO and CH4 at ~48%, ~20%, ~24% and ~ 5%, respectively. Comparatively, the amount of H2 and CO2 increased significantly over the Ni@HolSi-1 catalyst with the relative proportion of the selectivity of H2 at ~65%, CO2 at ~23% and CO at ~14%, respectively. In accordance with the previous results, it was found that the use of the Ni@HolSi-1 catalyst could prevent the production of the unwanted side products of CO and CH4 (Fig. 7e and f), which demonstrates the benefits of the mesoporous hollow framework of Si-1 zeolite crystals to promote the diffusion of intermediates in SRG and convert CO to CO2 and H2 via water gas shift reaction (Eq. S4). Similarly, the product distribution of H2, CO2, CO and CH4 were clearly less affected by the reaction temperatures under study, which might be attributed to the endothermic and exothermic nature of different side reactions (from Eqs. (S2) to (S10)), which favours the formation of H2, CO2, CO and CH4 at varying temperatures.As shown in Fig. S4a, among all the catalysts under study, the Ni@HolSi-1 shows a stable molar ratio (of H2/CO2 ≥ 2.33) in the reformed mixture at temperatures of 500–750 °C, being close to the theoretical ratio based on the stoichiometry of the SRG reaction (Eq. 1). Under the conditions used, in accordance with SRG (Eq. 1), the existence of other reactions (Table S1) such as WGSR, Eq. S4 [40] was also likely since CO and CH4 were also detected in the catalytic SRG systems. It is essential to highlight the catalyst (i.e. Ni@HolSi-1) that showed the lowest CO/CO2 molar ratio of <1 at temperatures investigated, as shown in Fig. S4b, with the lowest yet insignificant selectivity to CH4 (Fig. 7f). In comparison, over the impregnated Ni/Si-1 catalyst, the highest CO/CO2 molar ratios of >1 were observed, suggesting that the large and aggregated Ni particles promoted side reactions. Interesting, the CO/CO2 molar ratio in the impregnated Ni/Si-1 catalyst shows a volcano shape as a result of rising the reaction temperature, and the values are significantly high at 600–650 °C. This phenomenon could be attributed to the carbon deposition on large particles of Ni according to the Boudouard reaction (Eq. S8), whilst the gradual decrease of the ratio could be attributed to the reduction of the generated CO into the carbon species (Eq. S9) [41]. The possible carbon formation on the catalyst's surface, as shown through the carbon balance (Fig. S4c) was calculated using Eq. S14 without the consideration of solid products. On the other hand, in all of the catalyst systems under study, the average total mass balance was mostly determined in the range of ~90–96 wt%, with the loss of about ~4–6 wt%, which could be attributed to the uncondensed gaseous products and unavoidable retention of some condensed liquids by the inner wall of the reactor and the lines in the experimental rig. An example of mass balance calculation over different catalytic systems was shown in Table S4, using Eq. S15.Specific reaction rate of glycerol conversion (rX G ) and hydrogen formation (rH 2 ) on the developed catalysts were calculated using the results from kinetic experiments, which are presented in Fig. 9 and Table S3, respectively. Specifically, kinetic experiments were performed at 300–450 °C, atmospheric pressure and GHSV = 10,360 (STP) h−1 to ensure that the influence of mass and heat transfer boundaries are avoided. The summary of kinetic measurements of SRG over the catalysts are presented in Table S3, and the Ni@HolSi-1 catalyst demonstrated the higher specific reaction rates for glycerol conversion (i.e. ~1.8 × 10−5 mol s−1 g−1 at 400 °C) compared to that of the impregnated Ni/Si-1 catalyst (i.e. ~0.8 × 10−5 mol s−1 g−1 at 400 °C), perhaps due to the well-dispersed small Ni NPs of the Ni@HolSi-1 catalyst, suggesting higher contact efficiency, and hence the high glycerol conversions. Comparatively, the specific reaction rate of glycerol conversion over Ni@S-1 at 400 °C was about 1.5 × 10−5 mol s−1 g−1, being lower than that of Ni@HolSi-1 which could be due to the lack of mesoporous structure in it. Similar phenomena were found as well for the production of hydrogen with improved specific rate. For example, Ni@HolSi-1 demonstrated an excellent hydrogen generation with specific rate of about 2.7 × 10−5 mol s−1 g−1 at 400 °C, surpassing that of Ni/Si-1 (at about 0.38 × 10−10 mol s−1 g−1) under the same conditions. Arrhenius plots were obtained based on the results from the kinetic experiments to determine the activation energy (Ea) for the glycerol conversion using Eq. S13, and all the catalysts demonstrated common Arrhenius behaviour, as shown in Fig. 8 . According to the highlighted results shown in Table S3, the Ni@HolSi-1 catalyst presents the lowest E a value of 19 kJ mol−1, not as much as that of the Ni/Si-1 catalyst (i.e., 46 kJ·mol−1), as well as the Ni@Si-1 catalyst (~27 kJ·mol−1), which demonstrates the advantage of the Ni@HolSi-1 catalyst for promoting SRG.Stability of the Ni/Si-1, Ni@Si-1 and Ni@HolSi-1 catalysts under study was evaluated during the SRG. The longivity experiments were performed by running the freshly reduced catalysts continuously in a stream for 100 h. The catalytst testings were run under the same experimental conditions including the reaction temperature at 750 °C and GHSV = 3120 h−1 (STP). Glycerol conversion and hydrogen yield for the catalysts in stream of 100 h are displayed in Fig. 9 . As indicated in Fig. 9a, for the impregnated Ni/Si-1 catalyst, continuous deactivation was measured. In detail, the initial conversion of glycerol and hydrogen yield were 73% and 45%, respectively. However, the performance of the Ni/Si-1 catalyst kept decresing gradually as a function of TOS, which confirms the significant deactivation of Ni/Si-1 during catalytic SRG. The continuous deactivation of the impregnated Ni/Si-1 catalyst could be attributed to the poorly dispersed and large Ni particles which are prone to coking during the SRG reaction. Regarding the catalysts prepared by the encapsulation strategy, i.e., Ni@Si-1 and Ni@HolSi-1, they showed very stable performance over the 100-h longevity tests. Specifically, glycerol conversions of 99 ± 1% and H2 yield of 60 ± 1% were achieved over the Ni@Si-1 catalyst, as indicated in Fig. 9c. The post-synthetic TPAOH treatment resulting in the creation of mesoporous void structure was beneficial to improve hydrogen production, as shown in Fig. 9d, in which (i) the glycerol conversion over Ni@HolSi-1 is comparable to that of Ni@Si-1, and (ii) the H2 yield is higher at 70 ± 1%. The findings also suggest that the post-treatment was able to form the mesoporous structure without jeopodising the highly dispersed Ni NPs from the in situ encapsulation strategy. The selectivity to H2, CO, CO2 and CH4 gases in different catalytic SRG systems over different catalysts is shown in Fig. 9b, d and f. As indicated in Fig. 9b, over the Ni/Si-1 catalyst, H2 selectivity was stable from 0 to 50 h, then decreased from 50 to 100 h, as a result of reduction of CO and CO2 with H2 (Eq. S9 and Eq. S10) with the associated carbon deposition on the Ni/Si-1 catalyst. Comparatively, as shown in Fig. 9d and f, the encapsulated Ni catalysts maintained stable selectivities of H2, CO, CO2 and CH4 over the ToS of 100 h. The relatively stable performance of the encapsulated Ni catalysts could be associated to the presence of highly dispersed small Ni NPs, which are most active for reforming reaction than large Ni particles [42].In Fig. S5, the results of H2/CO2 and CO/CO2 ratios as determined for all the respective catalysts under study over 100 h on stream was highlighted. The H2/CO2 molar ratio of the Ni/Si-1 is observably changing over time on stream, until the catalyst was gradually deactivated (Fig. S5a). In the cases of encapsulated Ni catalysts (Figs. S5b and S5c), the H2/CO2 molar ratios are rather stable, especially the Ni@HolSi-1 catalyst, suggesting that the encapsulated yet highly dispersed Ni NPs combining the mesoporous hollow structure could mitigate catalyst deactivation considerably. Regarding the molar ratio of CO/CO2, the Ni@HolSi-1 catalyst exhibited a steady performance at <1 over 100 h in the reaction stream at 750 °C (Fig. S5c). The encapsulated Ni@Si-1 catalyst (Fig. S5b) produced slightly more carbon monoxide and showed the molar ratios of CO/CO2 > 1. Conversely, the impregnated Ni/Si-1 catalyst presented the molar ratio of CO/CO2 at ~1.5–5.3 over the 100 h, which could be due to the fact that carbon deposits are formed on the catalyst during the longevity test of catalytic SRG.Post-reaction characterisation of the used catalysts was performed to assess the effect of longevity SRG experiments on them. Comparative TGA analysis of the fresh (black solid lines) and used (black dot dash lines) catalysts was shown in Fig. 10 . All the fresh and calcined catalysts (i.e., Ni/Si-1, Ni@Si-1 and N@HolSi-1 catalysts, respectively) were very stable during TGA tests, showing no weight loss. Mass loss at temperatures below 200 °C was measured for the used encapsulated catalysts of Ni@Si-1 and Ni@HolSi-1, as shown in Fig. 10b and c, which was associated with the dehydration of all physisorbed water molecules within the frameworks of Si-1 zeolite supported catalysts. Regarding the weight loss which occurs at 550–800 °C, as shown by the derivated weight loss (solid red lines in Fig. 10a), the used Ni/Si-1 catalyst demonstrated a large weight loss of ~15 wt% (after 100 h SRG on stream at 750 °C), suggesting significant effects of carbon formed on the impregnated Ni/Si-1 catalyst during the SRG [43]. Similarly, a substantial weight loss of ~8 wt% at 550–800 °C was formed on the used encapsulated Ni@Si-1 catalyst (Fig. 10b). Conversely, the TGA profile of the used Ni@HolSi-1 showed insignificant weigh loss of <0.5 wt% in the high temperature range, demonstrating the anti-coking ability, which can be attributed to the formed mesoporous hollow structure from the post-synthetic TPAOH treatment of the encapsulated Ni catalyst (i.e., Ni@Si-1). Comparatively, post-analysis of XRD structures in the used catalysts were conducted at the end of the longevity tests of catalytic SRG at 750 °C (Fig. S6). The Si-1 structures were maintained after the 100-h test in all the tested catalysts, showing that the siliceous silicalite-1 zeolite is the suitable stable support for developing relevant reforming catalysts for applications under harsh conditions. The diffraction peak associated with the Ni phase was clearly observed for the used Ni/Si-1 catalyst, being comparable with that of the fresh catalyst (Fig. 1). For the encapsulated catalysts, the small Ni peaks were measured, as highlighted in Fig. 1b, especially Ni@HolSi-1, suggesting that the post treatment could render some encapsulated Ni phases exposed, which is subject to sintering at high temperatures. This was further evidenced by Fig. S7, in which some carbon nanostructures were formed on the used Ni@HolSi-1 catalyst. However, based on the catalytic SRG and comparative TGA results (Figs. 9 and 10), the developed encapsulation strategy and the post-synthetic treatment were effective for promoting catalytic SRG processes.Reforming reactions are class of important catalysis for producing many chemicals/fuels such as hydrogen, and they commonly require harsh thermal conditions, and hence the strategies of mitigating catalyst deactivation are needed to make the reforming processes more sustainable. In principle, catalyst design with small metal nanoparticles and segregated metal dispersion is beneficial to reduce deactivation because such catalysts can afford high activity with less chances of metal sintering and coke formation. Herein, towards H2 production via catalytic steam reforming of glycerol (SRG), an encapsulation strategy was employed to develop the encapsulated ultra-small Ni catalysts in silicalite-1 zeolite (Ni@Si-1) with high Ni dispersion, which showed much better performance than the conventional impregnated Ni catalyst (Ni/Si-1) regarding the activity and anti-coking ability. More importantly, the encapsulated Ni@Si-1 can be treated further using a post-synthetic method employing TPAOH solution under hydrothermal condition, which resulted in the encapsulated Ni catalysts with the mesoporous hollow structure, i.e., Ni@HolSi-1. The obtained Ni@HolSi-1 catalyst preserves the highly dispersed small Ni nanoparticles, being responsible for the measured high activity in SRG, e.g., a stable glycerol conversion of 99 ± 1% and H2 yield of 70 ± 1% over 100 h on stream together with the H2/CO2 molar ratio of >2.33 and CO/CO2 molar ratio of <1 at 750 °C. Compared to its parent of Ni@Si-1, the mesoporous Ni@HolSi-1demonstrated a very good anti-coking ability with insignificant coke deposition after a 100-h longevity test, whilst the encapsulated Ni@Si-1 showed ~8 wt% coke deposition under the same condition. The findings of this work show that (i) encapsulation is a very effective strategy for making highly dispersed metal catalysts for catalysis under harsh conditions, (ii) siliceous silicalite-1 zeolite is a good support candidate for preparing highly stable catalyst, and (iii) the presence of mesoporous feature in the encapsulated catalyst benefits local mass transfer, which is highly desired for reducing coke deposition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 872102. A.I. thanks the financial support of Petroleum Technology Development Fund (PTDF) in Nigeria for merit PhD scholarship (PTDF/ED/OSS/PHD/IA/1209/17). H.C. thanks the financial support from the European Commission under the Marie Skłodowska-Curie Individual Fellowship (H2020-MSCA-IF-NTPleasure-748196), the funding support from the Jiangsu Specially-Appointed Professors Program, the Natural Science Foundation of Jiangsu Province (BK20200704), and the State Key Laboratory of Materials-Oriented Chemical Engineering (No. ZK202001). Acknowledgements are extended to Dr. Shaojun Xu (Cardiff University) and Dr. Shaoliang Guan (Cardiff University) for their help with the TEM and XPS measurements, respectively. Supplementary material. Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuproc.2022.107306.

Valorisation of crude glycerol via steam reforming, i.e., SRG, is a promising method to produce sustainable hydrogen. However, catalyst deactivation under harsh SRG conditions is still a main challenge which hinders the further development of practical SRG. In this work, the encapsulated Ni catalyst in siliceous silicalite-1 zeolite (Ni@Si-1) were developed to show the improved performance and enhanced anti-deactivation potentials in catalytic SRG as compared with the conventional impregnated Ni catalysts (i.e., Ni/Si-1). Importantly, the post-synthetic treatment of Ni@Si-1 using TPAOH solution formed the encapsulated Ni catalyst with the mesoporous hollow structure (i.e., Ni@HolSi-1), which demonstrate even better performance in SRG with glycerol conversion of >95%, H2 yield of ~70%, H2/CO2 molar ratio of >2.33 and CO/CO2 molar ratio of <1 at 750 °C. Specifically, highly dispersed ultrasmall encapsulated Ni particles were retained within the hollow crystals of siliceous silicalite-1, as confirmed by XPS and HRTEM characterisation. The activation energy for glycerol conversion over Ni@HolSi-1 (i.e., Ea = ~ 19 kJ mol−1) was much lower than that of Ni/Si-1 and Ni@Si-1. 100-h longevity tests over the three catalysts were investigated at 750 °C, and the Ni@HolSi-1 catalyst exhibited an excellent stability and activity, as well as insignificant coke deposition, which could be due to the enhancement of highly dispersed yet accessible Ni NPs within the hollow Si-1 crystals. The findings of the work show the promise of the encapsulation strategy and mesoporous zeolites for developing the future reforming catalysts.

Data will be made available on request.In heterogeneous catalysis, the interaction between (metal) nanoparticles and a support is a crucial factor for the catalytic performance. Stabilizing the nanoparticles is the main reason to use a support, resulting in a high particle dispersion during synthesis [1] and preventing them from sintering during catalysis. In addition, supports can affect the catalytic activity and selectivity, for instance via the absorption of reactants or intermediates, by influencing the particle size or shape, or by altering reaction pathways [2,3].All mentioned factors can affect the performance of catalysts in the Power-to-Gas process, where CO2 is hydrogenated to methane. This is a highly interesting reaction to allow the storage of renewable hydrogen in synthetic natural gas [4]. A wide range of metals, for example Ru, Rh and in particular Ni have been investigated for this reaction [5,6]. Compared to noble metals, Ni is relatively low-priced, active and abundant. Typical supports used for this reaction are SiO2 and Al2O3 [6,7], with a recent switch to reducible oxides, such as CeO2, TiO2 or ZrO2 because of their increased CO2 adsorption activity [8,9]. In this reaction, factors such as metal particle size, support and promoter effects are important to understand, but can at the same time be very challenging to disentangle.Recently the use of carbon, especially carbon nanotubes (CNTs) as support for CO2 hydrogenation catalysts has gained more attention for fundamental studies [10–12]. Carbon materials are interesting model supports, because of their relatively high surface area and tunable surface chemistry [13]. Furthermore, carbon supports can be used to diminish the formation of species that strongly interaction with the support, for example metal silicates or aluminates [14–16], or enhance the interaction between active metal and promoters [17].During methanation, catalyst deactivation is an important factor to consider. This can be caused by the formation of nickel carbonyl species at low temperatures, whereas particle growth usually occurs at high temperatures [2,18,19]. Another challenge is the formation of carbon deposits, blocking the active metal surface, although this can be prevented by working at elevated pressures [20]. Carbon offers a high heat conductivity [10], which is crucial to prevent the formation of local hot spots during the exothermic methanation reaction (ΔH⁰ = −165 kJ mol−1) [21]. Modifying the surface chemistry of a support can help to stabilize nanoparticles.Typical support surface groups introduced to carbon supports are oxygen and nitrogen containing groups, changing the chemical properties of the carbon surface without changing its structural properties [22–24]. As a result, it is possible to vary the point of zero charge (PZC) and consequently the acidity or basicity over a wide range. A reflux treatment of pristine carbon (in this case graphite nanoplatelets, GNP) in HNO3 typically results in the incorporation of carboxylic, lactone and anhydride surface groups [25–27], increasing the acidic character of the material. An amination treatment of the oxidized carbon (GNP-O) converts the oxygen- into nitrogen-containing surface groups (GNP-N), which increase the surface basicity [22,23,28]. Support surface groups are often found to influence the final metal particle size of fresh catalysts. They can enhance the wetting of the precursor solution or can anchor the metal precursor more strongly [29]. Both could result in smaller nanoparticles [30–35], or even single atoms or clusters [12,36].Functionalization of carbon supports can improve the catalyst stability, by preventing nanoparticle growth [37–39]. Besides, the catalytic activity can be modulated, for instance by introducing N-containing species, to increase the basicity of the support [10,21,40–42], allowing enhanced CO2 adsorption [23]. Gonҫalves et al. performed a systematic study on the effect of support surface treatment of nickel on active carbon for low pressure CO2 hydrogenation and found that the use of the most basic carbons resulted in the highest catalytic activity [40]. However, the interference of differences in nanoparticle size on catalysis and the effect of support modification on catalyst stability were not addressed in full detail.In this paper, we discuss the effect of support functionalization for high pressure CO2 hydrogenation using graphite carbon nanoplatelets (GNP) as model support for Ni nanoparticles. Both oxygen and nitrogen containing surface groups were introduced to the carbon support surface before deposition of the nickel. We kept other parameters, such as the initial Ni particle size, the same and discuss the effect of the support treatment on catalytic performance during CO2 hydrogenation at 300 °C and 30 bar, with main focus on catalyst stability.Nickel nitrate hexahydrate (Ni(NO3)2·6H2O, Sigma Aldrich, ≥ 97.0%), nitric acid (HNO3, Merck, 65%) and Silicon Carbide, (SIKA ABR I, F70) were used as received. Graphite nanoplatelets (GNP-500, XG Sciences, grade C ∼500 m2 g−1 surface area) were either used as received, referred to as GNP, functionalized with oxygen-containing support surface groups (GNP-O) or nitrogen-containing support groups (GNP-N) or washed (GNP-W). To prepare oxidized carbon, approximately 10 g of the pristine GNP-500 was heated in 400 mL 65% HNO3 to 80 °C for 2 h while stirring. Afterwards, the suspension was washed several times with 5 L demi water each time until a pH of 6 was reached. After the last washing step, the support was dried at 120 °C for at least 24 h and subsequently crushed. Nitrogen functionalities were introduced to the support by substitution of oxygen functionalities [43]. Typically, ∼3 g GNP-O was loaded into a tubular oven, purged for 15 min with N2 gas at room temperature (200 mL min−1) and subsequently exposed to a flow of NH3 gas at 600 °C (220 mL min−1, 5 °C min−1, 4 h). To prepare GNP-W, approximately 2 g GNP was washed in 50 mL 1 M HNO3 at room temperature while stirring for 2 h. Afterwards the suspension was washed several times with 100 mL demi water each time until a pH of 6 was reached and dried in the same way as GNP-O.Nickel was deposited on either GNP, GNP-O, GNP-N or GNP-W using incipient wetness impregnation. Typically, 1.0 g of carbon support was dried in a round-bottom flask for 120 min at 170 °C, while stirring under dynamic vacuum to remove water and air from the pores. Aqueous nickel nitrate solutions were prepared by dissolving 2.0 M Ni(NO3)2 in mili Q water. The solution was acidified with 0.10 M HNO3 to ensure a pH around 1. The dried carbon support was impregnated with 0.73 mL gsupport −1 (90% of the pore volume of pristine GNP, determined using N2 physisorption) under static vacuum while stirring, to ensure that the solution was homogeneously spread over the support. Subsequently the sample was dried overnight at room temperature under dynamic vacuum. To decompose the precursor, 1 g of the sample was transferred to a plug-flow reactor and heated to 350 °C in 200 mL min−1 N2 (3 °C min−1, 90 min) to decompose the nitrate. The reactor was then cooled down and the gas was switched to 5% H2/N2 (200 mL min−1), which was the gas atmosphere for the subsequent reduction at 350 °C (2 °C min−1, 90 min). After cooling down, the catalyst was slowly exposed to air to passivate the nickel nanoparticles. The catalysts are denoted as Ni/GNP-X, where GNP-X is the type of carbon used (GNP, GNP-O, GNP-N or GNP-W).The pore volume and surface area of carbon supports were analyzed using N2-physisorption. Isotherms were measured at − 196 °C on a Micromeritics TriStar II Plus apparatus. The samples were dried overnight under vacuum at 170 °C before the measurement. The specific surface area of the support was calculated using the BET equation (0.05 < p/p 0 < 0.25) and the total pore volume was derived from the amount of N2 adsorbed at p/p 0 = 0.995.The density of acidic and basic surface groups was determined by potentiometric titration using a TIM 880 Titralab Titration Manager. The carbon materials were suspended in 65 mL 0.1 M KCl solution and degassed under N2 flow and vigorous stirring. For both acid and base titrations ∼25 mg of carbon material was used. The titrations were performed with either a 0.01 M NaOH or 0.01 M HCl solution, both in 0.1 M KCl solution. The amount of surface groups per gram carbon material was calculated based on the equivalence points of the titration data. Combined with the BET surface area obtained from physisorption, the density of surface groups (# groups nm−2) was determined for the different supports. The point of zero charge (PZC) of the support was determined through mass titration of the carbon material. Increasing amounts of carbon material were suspended in 10 g of 0.1 KCl solution, increasing the weight percentage of the support in the liquid, while measuring the pH. It is assumed that the amphoteric behavior of the surface groups will lead to a system pH equal to the PZC.[44].The supports were imaged with scanning electron microscopy (SEM) on a Helios G3 UC at 2 or 5 kV. The images were measured in field-free mode with a current of 0.40 nA. EDX analysis was performed using an Oxford silicon drift detector and Aztec software.The catalysts were imaged with transmission electron microscopy (TEM) on a Thermo Fisher Talos L120C operated at 120 kV or a Thermo Fisher Talos F200X microscope operated at 200 kV. The catalyst sample was dispersed as a dry catalyst powder onto a Cu sample grid coated with holey carbon (Agar 300 mesh Cu). Because of the nature of the carbon, consisting of thin graphitic sheets, dispersion of the catalyst powder in a solution and subsequent sonication was not necessary during the sample preparation. At least 400 nickel nanoparticles were manually counted per catalyst sample on at least 8 different catalyst locations using ImageJ analysis software. The determination of the Ni particle sizes is described in Supporting Information Section 1.High resolution TEM imaging was performed on a Thermo Fisher Spectra 300 monochromated, double-aberration corrected microscope operated at 300 kV. High angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and integrated differential phase contrast (iDPC) images were acquired in parallel. The screen current was ca. 0.05 nA and the camera length 145 mm.Powder X-ray diffraction (XRD) was performed on a Bruker D2 X-ray diffractometer, equipped with a Co-Ka1,2 radiation source (λ = 1.790 Å) and a Lynxeye detector. All catalysts were measured with diffraction angles varying between 10° and 95° 2θ with a step size of 0.05° 2θ/step while the sample was rotated at a rate of 15 rpm. All diffractograms were normalized to the carbon (002) peak at 30.9°. The crystallite sizes were calculated by applying the Scherrer equation to the NiO (111) peak at 43° or the Ni0 (200) peak at 61°.Thermogravimetric analysis was performed on a Perkin Elmer TGA800 coupled to an Hiden Analytical HPR-20 MS system. For the bare supports, the weight of 4–8 mg sample was determined while heating in Ar (10 °C min−1) to identify the weight % of functional groups on the supports. This technique was also used to determine the Ni weight-loading of the catalysts before and after catalysis as reported before [45] and described in detail in the Supporting Information Section 1.A Kratos Axis Ultra DLD system was used to collect XPS spectra using a monochromatic Al Kα X-ray source operating at 168 W (12 mA x 14 kV). Data was collected with pass energies of 160 eV for survey spectra, and 20 eV for the high-resolution scans with step sizes of 1 eV and 0.1 eV respectively. The system was operated in the Hybrid mode, using a combination of magnetic immersion and electrostatic lenses, and acquired over an area of approximately 300 × 700 µm2. A magnetically confined charge compensation system using low energy electrons was used to minimize charging of the sample surface and all spectra were taken with a 90° take of angle. A pressure of ca. 5 × 10−9 Torr was maintained during collection of the spectra. All samples were mounted into recesses of a modified Kratos Axis Ultra standard sample bar and gently pressed flat with iso-propyl alcohol cleaned glass slides before insertion into the spectrometer. All data was analyzed using CasaXPS (v2.3.24) [46] after subtraction of a Shirley background and using modified Wagner sensitivity factors as supplied by the instrument manufacturer. Cure fits were performed using an asymmetric Lorentzian form (LA line shape in CasaXPS), whereas the line shape for graphitic, sp2 carbon, was based on a cleaved, oxygen free HOPG sample.Temperature-programmed desorption (TPD) was performed on a Micromeritics AutoChem II 2920 apparatus. For the bare supports, 80 mg support was dried at 120 °C in Ar for 15 min. The sample was cooled down to 40 °C and subsequently heated in Ar (15 mL min−1) with 5 °C min−1 to 900 °C. H2O was captured with a dry ice/isopropanol cold trap. The outgoing gas was analyzed using a mass spectrometer (MS) of Hiden Analytical equipped with a QGA Professional software package.H2 chemisorption was measured on a Micromeritics ASAP 2020 C apparatus using ∼100 mg of sample. Prior to the measurement, the sample was reduced in pure H2 (6.0, Linde) at 300 °C for 2 h (5 °C min−1), after which full reduction was assumed, based on TPR analysis. The sample was then evacuated and cooled to 35 °C, and H2 chemisorption was measured at that temperature. The Ni surface area was obtained from extrapolation of the linear range of the adsorption isotherm of H2 to a pressure of 0 kPa, giving the H2 uptake (μmol gcat −1). The determination of the experimental and theoretical Ni surface areas is described in Supporting Information Section 1.The CO2 methanation catalysis was performed in a high throughput gas-phase 16-parallel fixed bed reactor system (Avantium Flowrence). Prior to the catalytic test, the catalyst powders were pelletized using a hydraulic press and subsequently sieved into a fraction of 75–150 µm. 60 mg catalyst was diluted with 240 mg SiC (>150 µm) to prevent the formation of hotspots. The mixture of catalyst granules and SiC were loaded in stainless steel reactor tubes (2.6 mm inner diameter) on top of ∼0.5 cm SiC granules. This was topped off with SiC.The Ni/GNP-X catalysts were in situ reduced prior to the reaction in a flow of 10% H2/N2 at 300 °C (2 °C min−1) for 3 h. Subsequently the reactors were cooled down to 120 °C before the reaction mixture was added. The reaction mixture consisted of CO2:H2:He = 19:76:5, 120 mL min−1, and was divided over 16 reactors. The resulting GHSV was 7500 mL gcat −1 h−1 . The reactor was gradually pressurized to 30 bar and subsequently heated to 300 °C with 2 °C min−1. This temperature was determined with TPR, see Fig. S1. The catalysts were tested up to 100 h to study both the activity and stability. The products were analyzed directly with online gas chromatography (GC, Agilent 7890B) with a sampling time of 14 min. Thus when all 16 reactors were in use, each sample was analyzed every ∼4 h. For each catalyst, three reactors were loaded and tested. After confirming the reproducibility, the catalytic results were averaged.To test the selectivity at different conversions, after 100 h the GHSV was varied. The total flow over the 16 reactors was adapted (50, 75 and 150 mL min−1 total flow) while keeping the gas mixture the same. Each new flow was equilibrated for 1 h and at least 2 datapoints per catalyst were taken (with 4 h difference). After the reaction, the catalysts were flushed with He and left to cool down to 60 °C before exposing them to air. This resulted in controlled passivation for post-catalytic characterization, for which the contents of the three reactors were combined. The formulas to determine the conversion, selectivity and turnover frequency are described in Supporting Information Section 1. Table 1 shows the structural characteristics of the graphite nanoplatelets that were used as-received (pristine, GNP), after the oxidation treatment (GNP-O) and after the amination treatment (GNP-N). The BET surface area and total pore volume of pristine carbon were 456 m2 g−1 and 0.81 mL g−1, respectively. After surface modification, GNP-O and GNP-N exhibited a surface area of to 415 and 308 m2 g−1 and pore volume of 0.72 and 0.62 mL g−1, respectively. The N2 physisorption isotherms are shown in Fig. S2. A decrease in surface area and pore volume is common for this relatively harsh oxidation treatment [22,47,48] and is probably due to the removal of an amorphous carbon fraction (with high specific surface area) as well as some collapse of the ordered graphite pore structure.The pristine GNP contained oxygen and its overall surface chemistry was slightly acidic (Table 1, Fig. S3). With the introduction of more oxygen-containing surface groups, the acidity increased, as evidenced by a decrease of the point of zero charge (PZC) from 4.0 to 3.0 (Table 1, Fig. S4). With the introduction of nitrogen functionalities, the PZC was increased to 9.0 and only basic groups were detected with titration (Table 1).Support treatment did not lead to significant changes in the X-ray diffractograms between 2θ = 20 and 95° (Fig. S5). At lower angles an extra peak was present for GNP, which mostly disappeared upon surface treatment. This likely indicated that the treatment influenced the stacking of the carbon platelets. The D-parameter represents the ratio between sp2 and sp3 carbon and is derived from the differential of the carbon x-ray induced Auger peak in the XPS spectrum [49,50]. This value was similar for all carbons (between 21.5 and 22.5 ± 1.0 eV), corresponding to a sp3 carbon content of ca. 10% [49]. This value was in good agreement with the C1s fitting (Table S1). No significant differences in morphology between the highly graphitic GNP and GNP-O were identified with scanning electron microscopy (SEM) (Fig. S6).The nature of the oxygen functionalities was investigated by following the gas release of the supports with temperature programmed reduction coupled with mass spectrometry (TPD-MS) up to 900 °C in argon ( Fig. 1). We first consider the pristine and oxidized carbon supports. In both cases CO2 and CO were released, due to the presence of oxygen-containing surface groups. In the case of GNP-O about double the amount was released compared to GNP, in line with the differences in acidity (Table 1). In addition thermographic analysis (TGA) showed a larger weight-loss of GNP-O (9.6%) than GNP (5.4%) at 800 °C in Ar (Fig. S7), confirming the presence of more oxygen in the carbon.The formation of CO2 (Fig. 1A) was attributed to the decomposition of carboxylic acids (100 and 400 °C), anhydrides (200 – 600 °C) and lactone groups (400 – 900 °C) [25–27,51]. CO formation (Fig. 1B) at low temperatures (< 300 °C) was caused by to the decomposition of aldehyde or ketone groups [25,52]. At higher temperatures, the peaks are typically ascribed to the decomposition of anhydrides (350 – 600 °C), phenols (500 – 750 °C) and carbonyl or quinone groups (650 – 950 °C) [22]. Altogether, TPD-MS analysis showed that the oxidation treatment had resulted in the incorporation of a range of oxygen containing surface groups, that could be carboxylic acids, anhydrides and phenols.Interestingly, for GNP-N, Fig. 1A and B show that only minor amounts of CO2 and CO were released; only above 600 °C a peak was observed for m/z = 28. This peak could either represent the formation CO from relatively stable oxygen containing surface groups (carbonyls or quinones) or the formation of N2 from nitrogen containing surface groups. The absence of CO2 and CO release at lower temperatures indicates that with the amination treatment, (most) oxygen containing surface groups were successfully removed. The release of some NO (m/z = 30, Fig. 1C) implies that nitrogen-containing groups had been successfully introduced to the support and that (part of) the nitrogen functional groups also contained oxygen, in agreement with results reported by Arrigo et al. [53].XPS analysis (Table 1) showed the increase in oxygen content for GNP-O (7.7 at%) with respect to the GNP (4.6 at%). The amination treatment was observed to cause significant loss of oxygen (<1at% left in GNP-N), whilst there was a concomitant increase in nitrogen (2 at%). High resolution spectra analysis of both C1s and N1s regions was performed to understand the chemical functionality. Fitting of the O1s spectra of the GNP materials ( Fig. 2A) identified contributions of two major peaks located at ca. 531.5 eV and 533 eV, corresponding to oxygen doubly or singly bound to carbon, respectively,[6,40] whilst the peaks between 535 and 540 eV are characteristic of a shake-up structure for carbonyl containing species.The oxidation treatment doubled the CO content (1.4 at% in GNP to 3.0 at% in GNP-O), and also exhibited a corresponding increase in the C−O functionality from 2.3 to 3.5 at% (Table S1), whilst amination caused low levels of these species to remain (0.5 and 0.3 at% for CO and C−O respectively). The C1s spectra were more complicated to fit given the similar binding energies of some oxygen and nitrogen containing functions, together with the asymmetry of the graphitic carbon and the uncertainty in the shape of the photoelectron background [49]. Nevertheless, XPS confirmed for GNP-O a high amount of several types of oxygen functional groups, with the groups comprising of either a C−O or a CO bond being dominant over the COO− groups (Fig. S8, Table S1).The main peak in the XP N1s spectrum of GNP-N is located at ca. 398 eV and is attributed to pyridinic-type groups (Fig. 2B, Table S2), whilst the peak at ca. 400 eV corresponds to pyrrolic- or pyridonic-type nitrogen species [26,53] or absorbed NHx. The smaller peaks between 402 and 405 eV could originate from graphitic N (∼403 eV) and oxidized nitrogen (∼405 eV) [54], however given the signal at 398 eV and these higher energy signals, these are likely to be attributed to loss structure from nitrogen in conformations such as that found in g-C3N4 [55]. In short, our findings from the XPS analysis and the TPD-MS analysis are in agreement and confirm the presence of oxygen or nitrogen-containing surface groups on the carbon support. Hence with the oxidation and subsequent amination treatment of GNP, three supports were prepared with different acidity/basicity and different types and amounts of support surface groups; C−O and CO groups for GNP and GNP-O and mainly pyridinic N for GNP-N.The main goal was to deposit Ni nanoparticles on the different supports without changing other parameters such as Ni particle size or loading. Indeed, all catalysts contained a Ni loading of ∼8 wt% and Ni nanoparticles of about 5 nm in diameter ( Table 2). The nickel deposition lowered the BET surface areas, but to a similar extent for all catalysts (249, 238 and 218 m2 g−1 for Ni/GNP, Ni/GNP-O and Ni/GNP-N, respectively, Table S3). The properties for the used catalysts, also displayed in Table 2, will be discussed in detail in Section 3.5.The D-parameter (Table S4) of the fresh Ni/GNP and Ni/GNP-O catalysts, determined from the Auger peak in XPS, was 22 – 23 ± 1.0 eV, in agreement with ∼10% sp3 carbon determined from the fitting of the C1s spectra (Table S4). Thus, the nickel deposition yielded negligible difference in the graphitic nature of the support. At the elevated temperatures used during Ni deposition (350 °C), carboxylic acid groups were not stable. As a consequence, the ratio between the C−O and CO groups decreased upon Ni deposition (Fig. S9, Table S5). Hence CO functionalities were preferentially retained after Ni deposition. This could either mean that these are more stable than C-O containing groups or, less likely, that the Ni nanoparticles bind preferably to C−O surface groups. For the Ni/GNP-N, the presence of nitrogen was evidenced by TGA-MS analysis, as NOx was released between 300 and 700 °C while heating this catalyst in oxygen atmosphere, which was done to determine the Ni weight loading (Fig. S10). Fig. 3A-C show transmission electron microscopy (TEM) images of three catalysts prepared using pristine (A, blue), oxidized (B, orange) and aminated (C, green) carbon with corresponding particle size distributions (insets in A, B and C). Independent of the support used, the surface averaged particle diameter (d s ) was 5 nm (Table 2). Although the TEM particle size was similar for all catalysts, the metallic surface area, determined using H2 chemisorption (Table 2, Table S6 and Fig. S11) was higher for Ni/GNP-N than for Ni/GNP and Ni/GNP-O. The experimental metal surface area of Ni/GNP-N was in agreement with the theoretical surface area calculated from the TEM particle size for a spherical nanoparticle, whereas for the other catalysts clearly lower specific metal surface areas were measured.The NiO(111) crystallite sizes determined from the peak at 2θ = 43° were 4.2, 5.2 and 3.2 nm for Ni/GNP, Ni/GNP-O and Ni/GNP-N respectively (Fig. 3D, Table 2), roughly matching the TEM results. The peak at low angles and the increased background of the bare GNP had disappeared, indicating that the nickel deposition had caused changes in the morphology of the GNP. No crystalline Ni3C was observed with XRD. Whilst this is not definitive proof of the absence of Ni3C, because of the relatively small peak shift compared to Ni, XPS supports this finding. Carbides typically give a distinct and narrow peak or shoulder in the lower binding energy side of the C1s peak (between 282.5 and 283.5 eV) [56], which was not observed for our catalysts (Fig. S9). Thus, XPS analysis of Ni/GNP-O and Ni/GNP catalysts also indicated that the presence of Ni3C was unlikely. Hence, we prepared nickel on carbon catalysts with different support surface groups, but similar Ni particle sizes and loadings and specific metal surface areas.The effect of the catalytic properties of the Ni-based catalysts was investigated under industrially relevant CO2 hydrogenation pressure and temperatures (30 bar, 300 °C). Fig. 4 shows the CO2 conversion of the catalysts, which were tested at relatively low conversion (10–20%) to allow examination of their performance far from equilibrium conversion (close to 100% at 30 bar and 300 °C). All bare supports were inactive for CO2 hydrogenation under these conditions. Ni/GNP-N showed the highest weight based activity, e.g. normalized to Ni content (22.6% CO2 conversion) (Fig. 4). The initial conversion of Ni/GNP-O and Ni/GNP were 15.6% and 12.9%, respectively. Thus the trend in initial conversion was Ni/GNP-N > Ni/GNP-O > Ni/GNP. This trend was reproducible throughout different catalytic tests and for various batches of catalysts (Fig. S12).The turnover frequency (TOF) based on this active metal surface area of the fresh catalysts, as determined by H2 chemisorption, was similar for all catalysts at the start of catalysis (1.3 – 1.6 *10−2 s−1, Table 2). Hence the differences in weight-based activity might be explained by a different metal-support interaction and/or specific accessible Ni surface area for the different supports. Alternatively Gonҫalves et al. reported for activated carbon and carbon nanotubes [12,40], that the amination treatment increased the catalytic activity during CO2 hydrogenation, due to enhanced adsorption of CO2 [23,40,57]. The latter was ascribed to the increased basicity of the support, where the adsorbed reaction intermediates can spill over from the basic groups onto the metal. Fig. 5A compares the selectivity towards CH4 of the Ni catalysts on different supports. Ni/GNP-N gave the highest CH4 selectivity (initially 92%). The initial selectivity was slightly lower for Ni/GNP-O (86%) and lowest for Ni/GNP (82%). Over the course of 100 h on stream, the CH4 selectivity of Ni/GNP-O and Ni/GNP-N were relatively stable, while a substantial decrease in selectivity was observed for Ni/GNP (to 65% after 100 h CO2 hydrogenation). In all cases CO was the main side product.One must take into account that in the low conversion range the selectivity to CH4 increases with conversion [45]. After the 100 h stability test, the reactant flow (and as a result GHSV) was changed to vary the CO2 conversion (Fig. S13). This allowed the study of CH4 selectivity versus CO2 conversion (Fig. 5B). Interestingly, the curves for Ni/GNP-N and Ni/GNP-O completely overlap. For Ni/GNP the CH4 selectivity versus CO2 conversion was substantially lower than for Ni/GNP-N and Ni/GNP-O. Thus, the support surface treatments have a positive effect on the CH4 selectivity although it does not seem to matter which kind of surface groups are introduced.Nanoparticle size and/or the formation of nickel carbide or a carbon layer around the Ni particles might influence the selectivity. However, a slight increase in selectivity is expected for larger particles [45], hence particle size effects cannot explain the differences. Besides, particle growth was severe for Ni/GNP-N, without a great change in selectivity. Similar to the fresh catalysts, XPS showed no indication of the presence of nickel carbide in the used Ni/GNP-O and Ni/GNP (Fig. S9).For pristine GNP, traces of support with different morphology might have affected the catalysis. Small amounts of, for instance, alkali elements might act as promoter or poison for supported metal catalysts [16,58–60]. With SEM-EDX analysis, no impurities were detected except some SiO2 in both GNP and GNP-O (Fig. S5). Because promoters might be active in low concentrations [61], impurities with concentrations below the detection limit of SEM-EDX might still have affected the selectivity. Hence we gave the pristine carbon a mild treatment not to introduce any surface groups, but nevertheless mimicking the treatment for GNP-O and GNP-N by washing in 1 M HNO3 at room temperature. The surface area of GNP-W was similar to GNP (488 vs 456 m2 g−1 respectively) and the TPD-MS profile was barely affected (Fig. S14). The selectivity of Ni/GNP-W was greatly enhanced as a result of the washing, and now similar to the selectivity of Ni on functionalized carbon (see Fig. 7). At the same time, the washing did not affect the activity as the CO2 conversion was still similar to the conversion Ni/GNP (Fig. S15). This shows that most likely the CH4 selectivity is very sensitive to low concentrations of contaminants. The exact influence of small concentrations of contaminants is interesting for further study. In conclusion, support treatment had a positive effect on the CH4 selectivity, most likely explained by the removal of traces of impurities.The activity evolution, normalized to the activity at t = 0, is depicted in Fig. 6A. The activity loss of Ni/GNP-O was 19 ± 5% during 100 h on stream, whereas for Ni/GNP this was 28 ± 5%. The most severe deactivation occurred for Ni/GNP-N, which lost 37 ± 5% activity during 100 h CO2 hydrogenation under these conditions.The most likely explanation for the activity loss is the loss of Ni active surface, as TEM analysis after CO2 hydrogenation (Fig. 6B-D) revealed particle growth in all catalysts. However, the extent of the particle growth was distinctly different for the different catalysts. The least growth occurred for the Ni nanoparticles in the Ni/GNP-O catalysts (from d s = 5.1 ± 1.2 to 6.0 ± 2.0 nm). This was followed by Ni/GNP (from d s = 5.3 ± 1.4 to 7.6 ± 2.7 nm) while most severe particle growth had occurred for Ni/GNP-N (from d s = 5.0 ± 1.7 to 10.7 ± 5.6 nm. This trend was confirmed by XRD (Fig. S16) and H2 chemisorption (Table 2). Overall the turnover frequencies were quite similar for all three catalysts, both before and after catalysis (Table 2). Although the main peak of the histograms was located around 5–6 nm for all catalysts, the size distributions of Ni/GNP and Ni/GNP-N were more broad than for Ni/GNP-O, especially a longer tail of large particles was visible in the histograms. For Ni/GNP-N, 12% of the nanoparticles counted had a diameter > 10 nm, whereas this value decreased to 5% for Ni/GNP and only 1.4% for Ni/GNP-O. A modest increase of intrinsic activity is expected with increasing particle sizes up to ∼8 nm, at least for Ni/GNP-O [45]. However, if there is an optimum above this size, as for example is the case for Co in Fischer Tropsch catalysis [62,63], one would expect the nanoparticles that had grown substantially (> 10 nm) to be less active in catalysis due to the lower surface area. Fig. 7 shows high resolution STEM images of the catalysts after CO2 hydrogenation, acquired in both High Angle Annular Dark Field (HAADF-STEM) and integrated differential phase contrast (iDPC) mode. iDPC analysis allows the visibility of the light carbon in the same image as the heavier nickel nanoparticles [64,65], The HAADF-STEM images revealed the appearance of core-shell nanoparticles, with a metallic Ni core surrounded by a 1–2 nm NiO shell, as expected from passivation in air. Some small (<5 nm) nanoparticles were fully oxidized. Especially for GNP-N, the Ni nanoparticles on the edge of the carbon sheets appeared less embedded in the carbon compared to the other supports (Fig. 7E and F). These observations might illustrate a weaker interaction of the nickel metal with specifically the GNP-N support, which has a low density of functional groups and could explain both the initially higher CO2 conversion and the poorer stability of Ni/GNP-N.The HAADF-STEM images and the iDPC images of Fig. 7 further showed that, independent of the support used, several nanoparticles were partially covered by carbon. The combination of the two imaging modes allows identification of these thin carbon layers. Although no statistical information can be derived from the 2D TEM images, we did not observe clear indications that aminated carbon in particular prevented Ni surface coverage as suggested by Wang et al. [21] At the same time, for none of the supports it was observed that these layers of (graphitic) carbon fully covered the nickel surface. This is in line with the accessible metal surface area as measured by H2 chemisorption as well as the fact that upon exposure to air the nickel nanoparticles were oxidized. The latter would be prevented when they would be fully encapsulated in carbon [66].Besides particle growth, we checked whether changes in the support might have contributed to the activity loss. The support surface areas remained between 200 and 220 m2 gNi −1 (Table S3). The D-parameter of both Ni/GNP-O and Ni/GNP was 22 ± 1.0 eV, within error the same as before CO2 hydrogenation (23 ± 1.0 eV) ((Table S4). Also, the ratio between C−O and CO surface groups remained the same for both Ni/GNP and Ni/GNP-O (Table S5). Upon heating, the used Ni/GNP-N catalysts still released NOx in oxygen atmosphere (Fig. S7). Finally, for none of the catalysts, the Ni weight loading was affected (Table 2). This shows that under the catalytic conditions used, the catalyst supports were fairly stable, and there was no significant Ni leaching. At the same time, compared to Ni on oxidic supports [7,67] these catalysts were neither the most stable, nor the most active catalysts for CO2 hydrogenation. However, it was not our aim to improve existing industrial catalyst but rather to present a series of model catalysts, allowing fundamental studies on support effects. This could be extended to, for instance the addition of another metal, such as Fe [6,40], or metal oxide promoters [35,67–69] improve catalyst activity and/or stability.All data support the conclusion that the catalyst deactivation was related to a loss of Ni active surface area, due to particle growth, which was influenced by the support properties. Ni/GNP-O was clearly the most stable catalyst, followed by Ni/GNP and finally Ni/GNP-N. It is most likely that the Ni particles remain smallest, and hence the most active in the GNP-O support due the remaining surface groups on this support. The treatment of the carbon to introduce functional groups might also have created defects [38]. However, if these were responsible for anchoring the nanoparticles, these must have been removed during the amination treatment. It is interesting that the interaction of the Ni nanoparticles was stronger with GNP-O than with GNP or GNP-N, despite the fact that nitrogen-containing surface groups are reported to stabilize nanoparticles [21,39,40,70]. Carboxylic surface groups are unstable during the heat treatments and thus are least likely to be present during high pressure CO2 hydrogenation. Nevertheless, their presence during Ni deposition could have resulted in a higher degree of nanoparticle embedding in the carbon support. Altogether, the support surface groups that are stable up to higher temperatures (> 350 °C), such as anhydrides, phenols, lactones or quinones are most probable to have contributed to the higher stability of the Ni nanoparticles, either by anchoring them, or by blocking their movement over the support.We have demonstrated the effect of carbon support functionalization with introduction of both oxygen and nitrogen-containing surface groups for Ni supported catalysts for CO2 hydrogenation. The surface modifications did not severely affect the Ni nanoparticle size of the fresh catalysts. The treatment to introduce nitrogen-containing surface groups resulted in the initially most active, but also least stable catalyst. Both phenomena were likely caused by the weak interaction between the Ni particles and the support, caused by the low amount of surface groups present. A higher available metal surface area benefited the activity, without affecting the TOF. The introduction of oxygen-containing surface groups significantly enhanced the catalyst stability. The oxygen-containing surface groups that were stable above 350 °C, either anchored the nanoparticles or prevented them from moving over the support. Finally, we showed that the type of support surface groups did not affect the CH4 selectivity significantly, but it was important to remove trace contaminants. With this work we showed that initial improvement in activity is not always optimal for long term catalysis. When using carbon as a support, the introduction of oxygen-containing support surface groups is advised for the severe conditions needed for synthesis of Ni-based catalysts (at least 350 °C) and high pressure CO2 hydrogenation. Nienke L. Visser: Conceptualization, Methodology, Investigation, Formal analysis, Validation, Writing – original draft, Visualization, Project administration. Juliette C. Verschoor: Methodology, Investigation, Formal analysis, Validation. Luc C.J. Smulders: Investigation. Francesco Mattarozzi: Formal analysis. David J. Morgan: Investigation, Formal analysis. Johannes D. Meeldijk: Investigation. Jessi E.S. van der Hoeven: Supervision. Joseph A. Stewart: Conceptualization. Bart D. Vandegehuchte: Conceptualization. Petra E. de Jongh: Conceptualization, Resources, Supervision, Project administration, Funding acquisition. All authors: Writing – review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to thank Jan Willem de Rijk and Remco Dalebout for their support in the catalytic experiments. Suzan Schoemaker, Kristiaan Helfferich and Laura Barberis are thanked for performing the N2-physisorption experiments. Dennie Wezendonk is kindly acknowledged for performing the TGA-MS experiments. We thank Ali Kosari for his input on the HRTEM measurements and Claudia Keijzer for performing the SEM measurements. This project is part of the Consortium on Metal Nanocatalysis funded by TotalEnergies OneTech Belgium, under TOTB Contract Ref IPA-5443.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.cattod.2023.114071. Supplementary material .

The interaction between metal nanoparticles and a support is of key importance in catalysis. In this study, we demonstrate that the introduction of oxygen- or nitrogen-containing surface groups on a graphite nanoplatelet support influences the performance of nickel supported catalysts during CO2 hydrogenation. By careful design of the synthesis conditions, the Ni nanoparticle size of the fresh catalysts was not affected by the type of support surface groups. A combination of H2 chemisorption and high resolution TEM demonstrates that the available metal surface depends on the interaction with the carbon support. The amination treatment to introduce nitrogen-containing groups results in the weakest interaction between the Ni and the support, showing the highest initial Ni weight-based activity, although at the expense of nanoparticle stability. Hence initial enhancement in activity is not always optimal for long term catalysis. The use of carbon with a higher density of oxygen functional groups that are stable above 350 °C, is beneficial for preventing deactivation due to particle growth. Furthermore, small amounts of contaminants can have a substantial influence on the CH4 selectivity at low conversions.

Water splitting by electrolysis (2H2O → O2 + 2H2) provides a possible path for the conversion of clean, renewable energy to H2 fuel to power human civilization [1,2]. The efficiency of water electrolysis is partially limited by the high kinetic overpotential associated with driving the oxygen evolution reaction (OER) [1,3,4]. Therefore, development of efficient catalysts is indispensable to facilitate fast kinetics (i.e. low overpotential). An ideal OER catalyst would be composed of nontoxic earth-abundant elements, economical to manufacture, chemically and mechanically stable, and sufficiently electrically conductive [5–7].Development of improved catalysts can be accelerated by an enhanced understanding of the underlying electrocatalytic mechanism and its dependence on catalyst composition and structure. The paradigm for understanding heterogeneous OER catalysis that has emerged over a century of research is based on the application of the Sabatier principle. The OER occurs on the catalyst surface sites, M, via a series of intermediates (e.g. M-OH, M-O, M-OOH, M-OO) [3]. If all of the intermediates are bound by an M-O bond, plotting activity versus the M-O bond strength should, in principle, result in a volcano-shaped graph, i.e a so-called “volcano plot”. At either side of the apex of the volcano the bond strength is sub-optimal; surfaces with either too large or too small M-O bond strengths are poor catalysts, as both lead to rate-determining steps with free energies that are larger than the average free energy for the steps in the mechanism [3]. OER catalysts based on earth-abundant first-row transition metals is of particular interest, as these catalysts might be used in water electrolysis or photoelectrolysis systems at a scale commensurate with global energy use. Consequently, there have been many experimental and computational efforts to correlate OER activity to chemical or material parameters. Mn [8], Fe [9], Co [10], and Ni-based [11] metal oxides and (oxy)hydroxides have been broadly studied and benchmarked for OER catalysis.Ni and its bimetallic oxides, particularly with Fe, are state-of-the-art catalysts in alkaline medium [12–15]; a Ni0.9Fe0.1O x OER activity was reported to surpass that of IrO2 [11]. Early studies by Corrigan and co-workers [16,17] and more recent ones by Boettcher et al. [18] show that Fe impurities from the electrolyte are readily incorporated into the Ni(OH)2 and significantly enhance the activity, but the role of Fe is still being debated. While various types of NiFe with alloys have been reported [14], Cui et al. [19] reported that a porous monolithic NiFe structure prepared by dealloying NiFeAl alloy exhibited much higher OER activity than the NiFe alloy itself. The improved performance was attributed to a large number of active sites and fast electron/mass transfer induced by the porous structure.Despite the high activity of some metal oxides, reported so far, most metal oxides possess insufficient electric conductivity for electrocatalytic purposes, as a low conductivity impedes the electron transport inside the bulk of the catalysts and between neighbouring catalyst nanoparticles (NPs), compromising kinetics [20]. However, Stevens et al. [21] concluded that in case of electrodeposited thin layers, conductivity enhancements does not necessarily enhance the electrocatalytic activity. Moreover, transition-metal-oxide/transition-metal nanocomposites such as NiO/Ni, FeO/Fe, and CoO/Co are inherently magnetic, the magnetic properties varying with size, crystal structure, and morphology, thus showing a wide variety of intriguing phenomena [22,23]. In the present context, the main issue concerning magnetism is that it may adversely affect colloidal stability and lead to particle agglomeration. This, in turn, decreases the active surface area and consequently leads to lower catalytic activity [24,25]. However, it should be noted that the decrease in activity due to magnetism might not be palpable when the catalysts are prepared via electrodeposition or formed on a porous support [26] as the agglomeration happens just in the powder form, and not an issue when they are prepared on a substrate. Therefore, optimizing the electrical behavior of the transition metal oxides or hydroxides and enhancing their colloidal stability to maintain the desired high specific surface area are two main properties that need to be considered in designing efficient catalysts. In order to overcome the aforementioned issues, conductive additives, such as carbon, have been extensively used to support transition metals and semiconducting or insulating metal oxide nanoparticles [27,28]. However, corrosion of carbon materials under OER conditions is under debate, and the absence of a solution to this problem prevents the industry from considering them as additives or supports for anodes in water electrolysis systems. Moreover, anodic degradation of carbon materials may not only decrease the extent of metal oxide utilization during the OER, but also leads to an uncertainty in the determination of the OER activity if the corrosion contribution to the oxidation current is not considered explicitly [29–31].In this context, transition metal phosphides (TMPs) [32] and carbon-encapsulated materials [33,34] have been reported as promising candidates for efficient electrocatalysis with enhanced activity compared with transition metal or metal oxides, which can be ascribed to both their nonmagnetic nature [35] (which translates to higher active surface area), and optimizing the electron transport inside the bulk of the electrocatalyst [36]. Among, all the tested transition metal-based catalysts, TMPs have the lowest overpotentials demonstrated to date [37]. A number of studies show that TMPs undergo an in situ electrochemical transformation under anodic oxidation conditions, being irreversibly converted to transition metal (oxy)hydroxides (TMOHs). These TMHOs have been proposed to be the true catalytically active species for the OER [38]. On the other hand, no such transformation was observed in TMP electrocatalysts after the OER by Liu et al. [39] and Liang et al. [40]. In this respect, TMPs are mainly considered as “pre-catalysts”, i.e. a catalyst that transforms into the actual catalytic material under and as a consequence of the operating conditions, rather than “catalysts” that maintains its nascent structure under any relevant conditions [32,38].Interestingly, there are many reports showing that the electrocatalytic activity of TMPs is enhanced by in situ formation of TMHOs on the surface. In other words, TMHOs-TMP composites formed in situ exhibit a better apparent OER performance than the corresponding pristine TMOs or TMHOs synthesized directly [41,42]. Although the underlying mechanisms are not fully understood, many studies have provided clues that the electrochemical oxidation of TMPs would enable the exposure of high density catalytically active sites. Moreover, any TMP with superior conductivity underneath a TMHO surface layers would facilitate electron transfer at the interface as well as electron transport inside the bulk component [43,44]. In the past few years, large research efforts have been devoted towards developing various TMP pre-catalysts for use in catalyzing the OER.The concept of encapsulating nanoparticles of non-precious 3d TMs and their alloys in various carbon matrices as an alternative towards efficient catalysts for the OER, ORR, and HER has recently attracted substantial attention [34,45,46]. Depending on the purpose, the carbon shell in carbon encapsulated nanoparticles plays different roles or provide multi-functionality. For instance, carbon encapsulated Pt nanoparticles in which Pt is electrochemically active show high durability as a result of a protection provided by the carbon shell. On the other hand, the electronic properties of the carbon shell can be modulated by the metallic nanoparticle cores, allowing for the binding energies of reaction intermediates on the carbon surface to be tuned. In some cases, carbon encapsulated metal nanoparticles exhibit high activity simultaneously against a variety of electrochemical reactions (e.g., HER and OER), demonstrating a bi-functional catalyst [47–49].Different methods, such as chemical vapor deposition (CVD), the polymer coating method, the solvothermal method, and the high-temperature pyrolytic method, have been utilized to form a thin carbon shell to encapsulate metal nanoparticles. Among all these, the solvothermal method has been given the most interest due to several advantages including a low temperature process ( < 300 °C), morphology tuning, time-efficient, possible scale-up, possibility of engineering the carbon shell, and so forth [47].For the first time, Carenco et al. [50] reported synthesis of carbon-encapsulated Ni2P nanoparticles via a solvothermal method, in which amorphous Ni2P nanoparticles were synthesized with excess amount of trioctylphosphine (TOP) at 220 °C and then subsequently converted to carbon-encapsulated nanoparticles by heating in a Schlenk tube for 30 min at 400 °C, under N2. The carbon layer was formed due to the decomposition of an excess amount of TOP during an annealing procedure.Recently, Jung and co-workers [47,51] have reported the synthesis of various transition metal nanoparticles encapsulated by carbon shell through the solvothermal method, which involves decomposing metal acetylacetonates precursors in organic solvents with surfactants under inert atmospheres at temperatures below 300 °C, after which the products are processed and subjected to annealing under different gas conditions to yield different carbon encapsulated metal structures. The carbon layer formed through the annealing step, in which the carbon atoms absorbed inside the lattice of the metal alloys diffuse to the nanoparticle surface, producing a mono or bilayer-level uniform carbon shell at the sub-nm scale.In the present work, we report the fabrication of ternary Ni12−x Fe x P5 nanoparticles (x = 0, 1.2, 2.4, 3.6) via a colloidal synthesis route. By introducing Fe precursors to the synthetic solution, a self-generated carbon layer surrounds the particles as the native ligand covering the nanoparticles is decomposed and lead to the formation of a carbon layer. This is contingent on the decomposition of the precursors happening at a high enough temperature, 300 °C in this work, and which is possibly catalyzed by the Ni-Fe bimetallic system. The key aspect of this catalyst design is that the carbon layer can provide a large specific area and interconnected electrically conducting networks which promotes the electrocatalytic activity of NiFeP nanoparticles significantly. Moreover, the stability of the carbon layer and NiFeP catalyst after being subjected to OER conditions were evaluated by TEM and Raman spectroscopy.Oleylamine (OAm; technical grade, 70 %), tri-n-octylphosphine (TOP; 97 %), nickel(II) acetylacetonate (Ni(acac)2; 97 %), iron(III) acetylacetonate (Fe(acac)3; anhydrous, 95 %), toluene (anhydrous, 99.8 %), acetone (99.5 %), isopropanol (IPA; 99.5 %), potassium hydroxide (99.99 %), and (5 wt%) Nafion 117 solution. All chemicals were purchased from Sigma-Aldrich and used as received, without further purification.Deionized water (DI-water), generated by a Milli-Q water system 18.2 MΩ cm−1, was used for all measurements.For all the catalysts in this work, the entire synthesis was completed in a single reactor in a dry, oxygen-free, Ar atmosphere (99.9999 %) by the use of Schlenk lines and a glove box. The protocol developed to synthesize Ni12−x Fe x P5 nanoparticles is based on the method refined by Muthuswamy et al. [52] to synthesize discrete Ni12P5 phase-pure nanoparticles.Formation of Ni12P5 and nickel-iron phosphide nanoparticles was achieved by reaction of Ni(acac)2 or mixtures of Ni(acac)2 and Fe(acac)3, respectively, with TOP as the P source in the presence of oleylamine via a two-step process. The two-step procedure is comprised of the generation of Ni and Ni x Fe1−x precursor particles at 220 °C followed by further reaction and crystallization at 300 °C. In a typical synthesis, 50 mL of OAm (156 mmol) was added to a 250 mL three-neck round bottom flask and evacuated for 10 min at room temperature. In the next step the corresponding amount (overall 15.6 mmol) of the two metal precursors (Ni(acac)2 and Fe(acac)3) (Fe:Ni molar ratios were 0.1, 0.2, or 0.3), and 14 mL TOP (31.2 mmol) were added to the solution and kept at 50 °C (ramp rate of 3 ° C min − 1 ) for 5 min under Ar atmosphere (99.9999 %). Then the temperature was ramped to 220 °C at rate of 8 ° C min − 1 and kept at this temperature for 2 h. In the second step the flask was heated further until 300 °C and kept for 30 min at this temperature. Once the reaction had finished, the flask was left to cool to room temperature either gradually while it was kept inside the heating mantle or with the heating mantle removed immediately after synthesis. The nanoparticles were isolated and washed at least three times using a mixture of isopropanol, toluene, and acetone to remove the remaining reagents and organic matter. Black powder (1.2 g) was obtained, which corresponds to a 100 % yield of Ni12−x Fe x P5 nanoparticles.We will designate the Ni12−x Fe x P5 compositions as Ni10.8Fe1.2P5 for x = 1.2, Ni9.6Fe2.4P5 for x = 2.4, and Ni8.4Fe3.6P5 for x = 3.6 below.Powder X-ray diffraction (PXRD) was carried out on a Bruker D8 DaVinci X-ray Diffractometer with Cu Kα radiation (Billerica, Massachusetts, USA). Samples were deposited onto zero background silicon sample holders and analyzed in the 2θ range between 20° and 80° with a step size of 0.04° and a collection time of 6 s. Identification of phases was made by comparison to the powder diffraction files (PDFs) of the International Center of Diffraction Data (ICDD) using Eva 5.1 software. The background was subtracted using EVA software for easier phase identification.Rietveld analysis was carried out using the Bruker TOPAS version 6.0, using a pseudo-Voigt function model. Refinements of diffraction patterns were performed within space groups Fd-3 m:1, I4/m. The occupancies were set to nominal values and were not refined.Scanning transmission electron microscopy (S(T)EM) was carried out on a Hitachi S-5500 FESEM (Krefeld, Germany) equipped with an INCA 350 energy-dispersion X-ray (EDS) analysis unit. Acceleration voltages of 30 kV and 20 kV were used for the images and the analyses, respectively. All samples were prepared by dropping a toluene suspension containing uniformly dispersed nanoparticles on a carbon film supported on a 300-mesh copper grid.TEM bright-field, TEM high-angle annular dark-field imaging (HAADF), and TEM-EDS were performed using a spherical aberration-corrected field emission JEOL 2100F TEM operating at 200 kV. EDS mapping was performed using a JEOL Silicon Drift Detector.Raman spectroscopy was carried out using a WITec alpha 300 R Confocal Raman device equipped with a 532 nm laser. Raman spectra were obtained after 20 accumulations for 20 s from 100 to 1250 cm−1.Spectra were collected on an Axis Ultra (Kratos Analytical) equipped with a Mg Kα X-ray source operating at 280 W Physical Electronics radiation source. The samples were analyzed under ultra-high-vacuum conditions (2.5 × 10−10 Torr base pressure). After recording a broad range spectrum (pass energy, 100 eV), high-resolution spectra were recorded for the C 1s, Ni 2p, Fe 2p and P 2p core XPS levels (pass energy, 200 eV). The binding energies were calibrated with respect to the C 1s peak at 284.8 eV. Spectrum processing was carried out using the Casa XPS software package.Electrochemical characterization was carried out in a standard three-electrode rotating disc electrode (RDE) setup from Pine Instruments. Polished glassy carbon (GC) electrodes were used as working electrodes (A = 0.196 cm2, Pine Instruments) and a Pt mesh was used as a counter electrode. The working electrode potentials were measured versus a Hg∣HgO reference electrode filled with 4.2 mol dm−3 KOH from Pine Instruments. Polytetrafluoroethylene (PTFE) containers were used both for electrochemical experiments and electrolyte preparation. All measurements were controlled using a Bio-Logic Potentiostat/Galvanostat (Model VMP3) in 1 mol dm−3 KOH (Fe-free electrolyte, 99.99 % and 85 % trace metal basis).Cyclic and linear sweep voltammograms were collected at a rotation frequency of 1600 rpm. Polarization curves were collected using chronoamperometry with Eappl (applied potential) stepped from 1.4 to 1.7 V vs. RHE in 20 mV increments. At each potential step, steady-state data were collected at angular velocities (ω) corresponding to rotational frequencies of 2000 and 600 rpm. Data were also collected in the absence of disk rotation. Catalyst inks were prepared by dispersing 2.5 mg of the catalyst powders in a mixture of 750 μL of milli-Q water, 250 μL of 2-propanol, and 50 μL of Nafion (5 wt%). The inks were homogeneously dispersed by ultrasonication for 20 min and then 10 μL was drop-cast on the GC electrode to make up a final metal loading of 0.12 mg cm−2. All electrochemical data were corrected for uncompensated series resistance after data collection. The uncompensated resistance of the cell was measured with a single-point high-frequency impedance measurement, and IR drop was compensated at 85 % through positive feedback using the Bio-Logic EC-Lab software. Our electrochemical cell typically had R u ~ 4 Ω in 1 mol dm−3 KOH. Electrochemical impedance spectroscopy measurements were carried out at five different overpotentials (0.6, 0.61, 0.615, 0.62, 0.625 V vs. Hg/HgO from 10 mHz to 1 MHz with an amplitude of 10 mV.Prior to all catalytic tests, the electrode was first subjected to continuous potential cycling at 50 mV−1s in the potential range of 1.0 through 1.6 V vs.(RHE) until reproducible voltammograms were obtained.The electrochemically active surface area (ECSA) was estimated from the double layer capacitance [53]. The double-layer capacitance, in turn, was estimated by cyclic voltammetry (CV) in a potential region in which faradaic currents can be assumed absent. The CV measurements were conducted in a quiescent solution by sweeping the potential across this non-faradaic region from the more positive to negative potential and back at 7 different scan rates: 10, 30, 50, 70, 100, 200, and 300 mV−1s. The working electrode was held at each potential vertex for 10 s before beginning the next sweep [54,55]. The double–layer capacitance was estimated from the slope of the plots of the charging current i c vs. the scan rate ν as dictated by the equation (1) i c = C dl × ν in which C dl is the double-layer capacitance [55].For rotating ring-disk electrode (RRDE) experiments, electrodes with various loadings (12–48 μg cm−2) were employed. 500 μL of the ink described above was diluted with 500 μL of milli-Q water. An amount of the ink corresponding to the desired loading was drop-cast on to a working electrode. The working electrode was a RRDE with a GC disk (5 mm diameter) and a gold ring (7.5 mm outer diameter and 6.5 mm inner diameter) equipped with an MSR rotator system, both from Pine Research Instruments. The counter electrode was a smooth Pt wire and the reference electrode was a Hg/HgO electrode filled with 4.2 mol dm−3 KOH. All cyclic voltammograms (CVs) of the disk electrode were recorded at a sweep rate of 10 mV−1s. The RRDE collection efficiency (24.1 % at 900 rpm) was determined from the ring and disk current ratios in 1 mol dm−3 KOH + 10 mol dm−3 K 3 [ Fe ( CN ) 6 ] solution. The ring potential (+0.3 V vs. RHE) for RRDE studies of the OER was chosen based on previous reports for oxygen reduction reaction on a gold electrode [29,56]. Before each RRDE measurement, the gold surface of the ring electrode was cleaned by applying 100 potential cycles in the interval from 0.03 to 1.53 V at 100 mV s−1.The Faradaic efficiency was also calculated using an eudiometer set-up (Figure S.10) based on collecting the generated oxygen gas bubbles by applying 10 mA (51 mA cm−2) constant current. The amount of the generated O2 was calculated from the volume of gas evolved corrected for the water vapour pressure and relating the amount of oxygen to the measured volume through the ideal gas equation. The theoretical amount of O2 expected to be produced by applying 10 mA (51 mA cm−2) was calculated from the electrical charge passed through the electrode using the Faraday equation: (2) n ( moles of produced O 2 ) = I t 4 F The calibration of the Hg/HgO electrode was performed in a standard three-electrode system with polished Pt foil as the working and counter electrodes, and the Hg/HgO electrode as the reference electrode. Electrolytes were pre-purged and saturated with 99.999 % H2. Linear sweep voltammetry (LSV) was then performed at a scan rate of 0.5 mV s−1, and the potential at which the current crossed zero was considered to be the thermodynamic potential for the hydrogen electrode reaction [57]. For example, in 1 mol dm−3 KOH, the zero current point appeared at − 0.900 V, and so the potential with respect to the reversible hydrogen electrode (RHE) is given by E(RHE) = E(Hg∕HgO) + 0.900 V. Fig. 1 shows the XRD patterns for the different target compositions. The XRD diffractogram for the nickel phosphide composition without any Fe unambiguously matches that of the pure Ni12P5-phase (tetragonal) structure (PDF 03-065-1623). For all iron-containing compositions, the diffraction peaks were shifted to larger angles compared to the corresponding peaks in the Ni12P5 diffractogram. As shown in Fig. 1, the Ni10.8Fe1.2P5 nanoparticles crystallized in the same tetragonal phase as Ni12P5 nanoparticles, suggesting the formation of homogeneous Ni-Fe-P compositions with no detectable crystalline impurities.However, clear changes in the diffractograms can be discerned upon further increase in the Fe content to above x = 1.2. At an Fe content to above x = 1.2, the peak at 49.2°, corresponding to the (312) plane of Ni12P5, dwindled while the intensity of the peak at 47.1°, corresponding to the (420) plane of Ni12P5, increased. Also, while the diffractogram for the composition with x = 1.2 contained the same peaks as the Ni12P5 catalyst, new peaks have emerged for the compositions with x > 1.2. This suggests the development of a second phase for x = 2.4, i.e. when the Fe content is increased beyond x = 1.2. This second phase is most likely an Fe3O4 phase; the new peaks at 35.16° and 31.7° agree well with the (101) and (211) planes of Fe3O4, respectively. The peak corresponding to the (420) plane in Ni12P5 overlaps with the peak corresponding to the (202) plane in Fe3O4. Therefore, the increase in the intensity at 47.1° with increasing Fe content can be attributed to a growing Fe3O4 phase. The peaks at 57° and 62.6° in Ni8.4Fe3.6P5 belong to the (115) and (044) planes in Fe3O4. For Ni8.4Fe3.6P5 (x = 3.6), Fe2O3 is formed as a third phase, and the reason for the higher intensity of the 49.2° peak than in NiFeP@Fe3O4(x = 2.4) is likely to be due to its overlap with the (024) plane of Fe2O3. The emergence of the new peak at 32.7° is also attributed to the Fe2O3, viz. its (104) plane, which substantiates the suggested presence of an Fe2O3 phase.Based on the Vegard’s law for alloys, we would expect a linear relation between lattice parameters and the composition. However, such a linear behavior was not observed. This deviation from Vegard’s law has been previously reported for Fe x Ni2−x P bulk solid solutions and nanoparticles, and has been attributed to an unequal distribution of the two different metals in sites of different size in the lattice [58,59].The Ni12P5 tetragonal structure type has two metal coordination sites. The Ni atoms in the first site are surrounded by the four nearest P atoms at distances 2.194–2.467 Å and eight Ni atoms located at 2.526–2.725 Å. There are 11 atoms forming coordination polyhedra around the second Ni site, viz. two nearest P atoms at 2.260 Å – 2.283 Å, five Ni atoms at 2.517–2.575 Å, two P atoms at 2.619 Å, and two Ni atoms at 2.725 Å [60]. The atomic radius of Fe is slightly larger than that of Ni 125 pm vs. 121 pm. Among the two available sites, we would normally expect Fe to occupy the larger one in Ni12−x Fe x P5. This is the case for high Fe fractions. However, studies of bulk hexagonal structures, which also have two different metal sites, suggest that occupancy is dependent on composition; at low Fe metal fractions, the smaller site is preferentially occupied by Fe. Goodenough [61] has suggested that the preference of Ni for the larger site in Fe-poor compositions is due to electron transfer from Fe to Ni. X-ray photoelectron spectroscopy (XPS) and X-ray absorption near-edge spectroscopy (XANES) investigations of the Fe x Ni2−x P system [62] have revealed that the electron density of Ni atoms has been increased, presumably due to electron transfer from Fe to the more electronegative Ni atoms, consistent with this hypothesis.Bright-field TEM images of Ni12P5 show that the catalyst consists of quasi-spherical nanoparticles with average diameter of 15.20 ± 2.25 nm. An example is given in Fig. S.1 in the Supporting Information. Analysis of the high-resolution TEM (HR-TEM) image (Fig. S.1(b)) gives lattice-fringe spacings of about 2.1 Å, corresponding to the (400) lattice plane of tetragonal Ni12P5. Energy-dispersive spectroscopy (EDS) in TEM indicate uniform distributions of Ni and P across the Ni12P5 nanoparticles (Fig. 1.2). Based on the EDS maps performed in the TEM, the apparent ratio of P:Ni was estimated to 0.38.TEM images of Ni12−x Fe x P5 nanoparticles (Fig. S.2, 2, and S.3) show that the quasi-spherical Ni12P5 nanoparticles were converted to highly faceted nanoparticles with pentagonal cross-sections in TEM upon addition of Fe. The size distribution became broader with increasing Fe content.Analysis of the HR-TEM image of Ni10.8Fe1.2P5 nanoparticles (Fig. S.2(c)) revealed a d-spacing of 2.1 Å corresponding to the (400) plane of the Ni12P5 tetragonal crystal structure, which is consistent with the XRD results. Moreover, the EDS mapping confirmed the uniform distribution of Ni, Fe, P, and O across the particles. The EDS composition was in relatively good agreement with the targeted stoichiometry.Upon increasing the Fe content to x = 2.4 and 3.6, the particles became more faceted and irregular in shape. Results of energy-dispersive spectroscopy (EDS) indicate a core-shell structure for the Ni9.6Fe2.4P5 and Ni8.4Fe3.6P5 nanoparticles (Fig. 2 and S.3) where Ni and P are evenly distributed in the core while Fe and O that reside in the shell dominate over that in the bulk. This partial segregation is compatible with the XRD patterns, which indicates the evolution of Fe3O4 as the second phase. The TEM images demonstrate that the Fe3O4 phase forms a shell surrounding a Ni12−x Fe x P5 core, in which 1.2 < x (stoichiometry of Fe) < 3.6. The core is rich in Ni and P, while the shell is rich in Fe and O. For simplicity, we will refer below to these particles as NiFeP@Fe3O4(x = 2.4) for the sample of nominal composition x = 2.4 or NiFeP@Fe3O4(x = 3.6) for the sample of nominal composition x = 3.6, while referring to the compositions in the general sense as Ni12−x Fe x P5 as before when the catalyst architecture is not important.Close inspection of the TEM images of Ni12−x Fe x P5 nanoparticles reveals the existence of a relatively regular coating at least partly covering the NiFeP@Fe3O4 particles (Fig. 2(b) and (c)). The thickness of this layer varies from catalyst to catalyst and it is more developed (thicker) for NiFeP@Fe3O4(x = 2.4), and NiFeP@Fe3O4(x = 3.6) in comparison with Ni10.8Fe1.2P5.Raman spectra of the synthesized Ni12−x Fe x P5 nanoparticles are shown in Fig. 3. The peak positions are listed in Table 1. All the recorded spectra were subjected to a Voight-based deconvolution analysis.The data in Table 1 show that when the Fe content increases all peaks below 350 cm−1 are blue-shifted and those with wavenumbers higher than 1105 cm−1 are red-shifted. As indicated in Fig. S.4 some of the observed peaks were attributed to NiO and FeO x species [63,64]. In all Raman spectra of Ni12−x Fe x P5 nanoparticles, Fig. 3, two peaks at around 1582 and 1360 cm−1 can be clearly seen. For comparison, similar peaks were also observed at the glassy-carbon electrode used for the electrochemical measurements, see Section 3.3 below Table 2.XPS survey spectra recorded for Ni12−x Fe x P5 (see Figure S.5 in the Supporting Information) show clear peaks corresponding to Fe which are not present in the spectrum for the pure Ni12P5 phase. This indicates the successful incorporation of Fe in the former samples. Fig. 4 (a) shows the Ni 2p XPS core-level spectra of the synthesized nanoparticles. The Ni 2p spectrum contains two main peaks, resulting from the spin-orbit splitting of the p orbital that are assigned as Ni 2p 3∕2 (850–865 eV) and Ni 2p 1∕2 (865–885 eV). The Ni 2p 3∕2 region was further deconvoluted into three peaks for Ni12P5, NiFeP@Fe3O4(x = 2.4), and NiFeP@Fe3O4(x = 3.64). However, since the satellite and oxidized Ni was quite well-separated for Ni10.8Fe1.2P, the Ni 2p 3∕2 region was therefore deconvoluted into four peaks. The peak at 853 eV can be related to both Ni and Ni-P [13]. Unfortunately, an unambiguous separation of the contributions from these two species through XPS is challenging. A previous study by Li et al. attributed both Ni and Ni-P to the same BE of 853.1 eV, [65] while others have tabulated Ni(0) at 852.7 eV and Ni2P at 852.9 eV, only 0.2 eV apart [66]. We therefore made no attempt at separating the two contributions here. However, for NiFeP@Fe3O4(x = 3.6) the peak at 854.4 eV can be exclusively assigned to Ni-P [67].Regarding the shift in Ni 2p peaks with addition of Fe, there is evidence in the literature [61,62] showing that electron transfer from Fe to Ni will take place in nickel iron phosphide compounds, which in turn increases the electron density of Ni atoms. Considering the fact that Ni atoms have higher electron density upon addition of Fe, we would expect a shift to lower binding energies in Ni. This is in accordance with our experimental results.The XPS spectra for the as-synthesized Ni12−x Fe x P5 nanoparticles (Fig. 4) could be fitted to an Fe 2p 3∕2 peak at 712.15 eV and an Fe 2p 1∕2 peak at 724.18 eV. This indicates that two distinct Fe species are present in the samples. The Fe 2p 3∕2 peak can, in turn, be decomposed into two peaks approximately at 706 and 713 eV, respectively originating from the iron(0) and oxidized iron [68].For all the catalysts except Ni10.8Fe1.2P5, the P 2p region was deconvoluted into four peaks. For the Ni12P5 catalyst, components at 129.4, 130.4, 132.6 and 133.3 eV, corresponding to phosphide, P(0), P(III) and P(V) species, respectively [69,70], proved to fit the spectra well. The values were in good agreement with the corresponding values reported for Ni12P5 in the literature [71]. The P(V) and P(III) components have been interpreted as surface phosphate and phosphite [70], respectively. These may have formed as a result of the exposure of the nanoparticles to air while being stored at the ambient conditions. Upon addition of Fe, a noticeable shift is observed in all the components, possibly due to the interaction of P with Fe. It is also worth noting that the fraction of oxidized phosphorous is larger in Fe-containing nanoparticles than in Ni12P5, which indicates that the addition of Fe makes particles more vulnerable to oxidation. The ratio of oxidized to non-oxidized phosphorous species increases in the order of Ni10.8Fe1.2P5 > NiFeP@Fe3O4(x = 3.6) > NiFeP@Fe3O4(x = 2.4), which is opposite of the order in terms of the thickness of the self-generated carbon layer. Therefore, it is reasonable to conclude that the carbon layer to some extent protects the particles from oxidation. For the Ni10.8Fe1.2P5 nanoparticles with the thinnest carbon layer, the metal-P (phosphide) and P(0) species were barely detectable, indicating negligible carbon-layer protection and extensive surface oxidation of nanoparticles. All the parameters obtained from fits to the XPS data are presented in Table S.1. Fig. 5 shows cyclic voltammograms of Ni12−x Fe x P5 catalysts. For comparison, an Fe-free Ni12P5 catalyst was also tested as a benchmark compound to explore the effect of the addition of Fe on the electrocatalytic activity. The CVs for all the tested catalysts contained redox peaks at potentials below the onset of the oxygen evolution reaction, attributed to Ni3+∕Ni2+. However, the peak position differs depending on the composition of the catalyst.For the Ni12P5, the anodic redox peak appears at E anodic = 1.36 V in the CV and the cathodic peak at E cathodic = 1.28 V. Interestingly, upon addition of Fe the anodic redox peak is shifted towards positive potentials. The anodic peak in Ni10.8Fe1.2P5 is split into two peaks ( ~ 1.34 and 1.40 V) whereas the cathodic peak is observed at the same potential as Ni12P5.Splitting of the anodic peak is also observed for the NiFeP@Fe3O4(x = 3.6) catalyst. A wider separation of the peaks was observed in this case, however, with peak positions at E anodic = 1.34 and 1.42 V. The splitting of the anodic peak suggests two types of Ni sites in the particles, one corresponding to Ni sites in Ni12P5 and another at which Ni interacts with Fe. The absence of any cathodic split maybe related to sluggish kinetics. The cathodic peak shifts to the more positive potential of 1.34 V. Finally, the cyclic voltammogram of the NiFeP@Fe3O4(x = 2.4) catalyst shows a redox peak without any splitting at E anodic = 1.41 V and E cathodic = 1.34 V.The shift in the Ni3+∕Ni2+ redox peak to more positive potentials upon addition of Fe is well documented, and has generally been attributed to the stabilization of the Ni2+ state in the presence of Fe [72–74]. The larger peak current in the case of NiFeP@Fe3O4(x = 2.4) is opposite of what is normally reported in the literature [75,76], and the effect of Fe is usually that of reducing the peak current density. Fig. 6 shows the linear sweep voltammograms for the Ni12−x Fe x P5 catalysts. Fig. 6 also includes polarization curves recorded by chronoamperometry, which are in excellent agreement with those recorded by linear sweep voltammetry. The overpotential needed for all tested catalysts to deliver 10 and 50 mA cm−2 (i.e. η 10 and η 50) are tabulated in Table 3. To reach the benchmark current density η 10 at the Ni12P5, an overpotential of 301 mV is needed, while NiFeP@Fe3O4(x = 2.4) merely requires an overpotential of 220 mV, showing a significant improvement in the OER activity. The apparent OER activity per mass for all the tested catalysts follows the order: Ni12P5 < Ni10.8Fe1.2P5 < NiFeP@Fe3O4(x = 3.6) < NiFeP@Fe3O4(x = 2.4).The polarization curves presented in Fig. 6 all show an up-turn at high overpotentials, which is a common feature of plots of electrode potential vs. the logarithm of current for the OER as presented in the literature [77–80,81,82]. Such changes in the slope d E ∕ d log i with increasing potential are most often attributed to either a change in the rate-determining step (rds) within a given pathway [79] or to saturation or depletion of intermediates at the surface [77]. The degree of consistency between the data recorded by LSV and CA, suggests that the dual-slope behavior is mechanistically significant and not due to electrode blocking, mass-transport limitations or ohmic effects. Kinetic parameters, including Tafel slopes (i.e. d E ∕ d log i ), determined from the lower overpotential region (below the up-turn) are presented in Table 3.The results of EIS measurements, plotted as Tafel impedance (Z t ), at different overpotentials for Ni12−x Fe x P5 with (x = 0, 1.2, 2.4, 3.6) are shown in Fig. 7. Z t was computed from the impedance by multiplication of the latter with the steady-state current density as [83], (3) Z t = E ˜ i ˜ i s s where E ˜ is the potential amplitude, i ˜ the current-density amplitude, and i ss is the steady-state current density. (The ohmic resistance, as assessed from the high-frequency intercept of the impedance-plane plot with the real axis, was subtracted from all data prior to the conversion to Tafel impedance). As can be seen, the low-frequency intercept increases slightly as the overpotential is increasing. In these plots, the dE∕dlogi slope can be read off as the value of the low-frequency intercept with the real axis [84,85]. For all samples the diameter of the arc in the Tafel-impedance plane plot are in reasonable agreement with the slopes from the steady-state curves, Fig. 6. However, due to some ambiguity in determining the appropriate region to use for fitting the steady-state data, we consider the Tafel slopes obtained through impedance to represent the more accurate of the two sets of values. The Tafel slopes from the impedance data cluster around 40 mV for all the iron-containing samples (Ni10.8Fe1.2P5, NiFeP@Fe3O4, and NiFeP@Fe3O4), whereas the Tafel slope for Ni12P5 is significantly higher, 60 mV).Data from which the double-layer capacitances (C dl ) were evaluated and the ECSA were estimated, are given in the Supporting Information(Fig. S.6). C dl values of 3.35, 3.56, 2.85 and 2.26 mF cm−2 were obtained for NiFeP@Fe3O4(x = 2.4), NiFeP@Fe3O4(x = 3.6), Ni10.8Fe1.2P5, Ni12P5 respectively. In general, the double layer capacitances and hence the ECSA for the iron-containing Ni12−x Fe x P5 catalysts were found to be larger than those for the Ni12P5 catalyst. In effect, the C dl value is increasing along with the thickness of the carbon shell. Fig. 8 compares the mass activity and overpotential of the NiFeP@Fe3O4(x = 2.4) catalyst in this work with data for other catalysts based on non-precious metals as collected by Kibsgaard and Chorkendorff [86]. As can be seen from the plot, the NiFeP@Fe3O4(x = 2.4) catalyst is among the best catalysts reported so far, displaying a mass activity of 0.1 A mg−1 and an overpotential of 220 mV at 10 mA cm geo 2 . Faradaic efficiencies of ~ 95 % and ~ 97 % (see the Supporting Information) were estimated from measurements of the volume of the collected gas and by use of a ring-disc electrode (see Section 2.9).In addition to the high OER catalytic activity, the NiFeP@Fe3O4(x = 2.4) catalyst also showed a high stability under OER conditions, as measured by 500 potential cycles between 1.1 and 1.7 V at a scan rate of 10 mVs−1 ( Fig. 9(a)). From cycle 10 to cycle 500 the current at 1.525 V decreased from 140 mA cm−2 to 100 mA cm−2. There is no noticeable decrease in the charge associated with the anodic redox peak at 1.43 V and the corresponding cathodic peak at 1.35 V. There is, however, a slight shift to lower potentials with increasing number of scans. It is well-known that the addition of Fe to Ni catalysts will shift the redox peak to higher potentials. Therefore, we associate the shift in the peaks to lower overpotentials to a slight change in the surface composition and a concomitant change (8 %) in the catalytic activity also visible in the figure. The chronoamperometric measurement involved applying a constant current of 50 mAcm−2) for 10 h (Fig. 9(b)) in 1 mol dm−3. No noticeable increase in the potential was observed after 10 h, indicating that NiFeP@Fe3O4(x = 2.4) is very stable. Fig. 10 and Fig. S.7 exhibit TEM images of the semi-spherical NiFeP@Fe3O4(x = 2.4) nanoparticles and the corresponding EDS mappings after they had been subjected to a constant 10 mA current for 5 h. The TEM-EDS mapping of NiFeP@Fe3O4(x = 2.4) nanoparticles (Fig. S.7) shows that phosphorus remains a part of the catalyst after exposure to the OER conditions. The bulk Ni:P ratio was 2.5, essentially similar to that of the as-prepared nanoparticles, with the Ni:P ratio of 2.7 prior to the test. Fig. 10(b) shows the HR-TEM images of two adjacent NiFeP@Fe3O4(x = 2.4) nanoparticles. The images are similar to those in Fig. 2(b) and (c), and the coating covering the particles that is visible in Fig. 2(b) and (c) is still intact after exposure to the electrolyte and high electrode potentials associated with the OER. The HR-TEM image of the particles also shows crystalline domains at their center, but somewhat less crystalline domains at their periphery. Fig. 3 shows the Raman spectra of as-prepared NiFeP@Fe3O4(x = 2.4)/GC (glassy carbon) electrodes and the same sample after immersion for 10 min, and after the sample had been subjected to 100 cycles between 1 and 1.7 V and a constant current of 10 mA (50 mA cm−2) for 2 h. A large peak at 1100 cm−1 in the NiFeP@Fe3O4(x = 2.4) powder in the Raman spectra prior to mixing the ink (Fig. 3) is no longer present in the spectra of the same catalyst on the GC electrode, i.e. post mortem (Fig. 3). We associate this with dissolution of phosphate/phosphite species during the ink preparation. Apart from that, no other change was observed related to the changing/reorganization of the NiFeP@Fe3O4(x = 2.4) catalyst after OER.Apart from their high mass activity and current efficiency for the OER, the most prominent feature of the Ni12−x Fe x P5 catalysts is the presence of a coating both in the pristine catalysts as in Fig. 10(b), and post mortem as in Fig. 2(b) and (c). Regarding the fact that nanoparticles were synthesized in the presence of organic compounds (i.e. oleylamine and TOP), it is likely that the layer consists of carbon which has been generated upon decomposition of organic moieties adhered on the nanoparticles during the synthesis [69]. However, according to Jung et al. [51] and considering the fact that carbon atoms can be absorbed inside the lattice of the metal nanoparticles where metal acetylacetonate is used as a metal precursor [87], another possibility for the formation of the carbon layer could be the diffusion of carbon atoms from the interior of the metal nanoparticles to their surfaces in the phosphidation step at 300 °C. Moreover, the fact that the layer is invisible in the high-angle annular dark-field image of NiFeP@Fe3O4(x = 2.4) (Fig. 2(d)) is consistent with the layer being composed of a lighter element, such as carbon, than those of the catalyst particle itself.The bands peaking at wavenumbers 1582c−1 and 1360 cm−1 in Fig. 3 are consistent with the G- and D-bands, respectively, for carbon samples [88], and supports the suggestion that a carbon layer has been formed at the particle surfaces. The G-band is associated with an ordered graphite structure and the D-band with defects, respectively. The peak height ratios I D:I G are 0.98 for NiFeP@Fe3O4(x = 2.4) (Fig. 3), 0.96 for NiFeP@Fe3O4(x = 3.6), and 0.95 for the Ni10.8Fe1.2P5. The higher ratio found for NiFeP@Fe3O4(x = 2.4) indicate a more defective nature and porous structure of the carbon layer [88] for this sample.The fact that the particle size is quite narrow, Fig. S5, makes it likely that at least the majority of the particles have been coated by carbon. A narrow particle size distribution indicates that the oleylamine and TOP were efficient in preventing particle growth, and therefore organic residues will coat all particles within the dominating size range. It is these residues that would be converted to carbon in the heating step, hence coating all particles within the size range indicated in Fig. S5.We therefore conclude that the coating covering the catalyst particles is a self-generated layer of carbon coming from ligand decomposition, with some possible doping by nitrogen or phosphorus from the ligands or even Fe [89] or Ni from the metal precursor. This layer only forms with iron present in the nanoparticles. Therefore, it is probably catalyzed by iron and therefore only present in the bimetallic system. The process of formation is therefore somewhat analogous to that suggested for the growth of carbon nanotubes on NiP amorphous nanoparticles, in the absence of any Fe, by annealing at the substantially higher temperature of 400 °C in an inert atmosphere [69,90].The differences in the ECSA (and in the peak heights in the voltammograms in Fig. 5 between the different catalysts are a likely manifestation of the carbon coating. This is because a carbon layer may help keeping catalyst particles apart and prevent agglomeration. As has been reported previously, one of the main advantages of carbon encapsulation of nanoparticles is the increase of the active surface area as a consequence of reduction in the agglomeration of nanoparticles [33,91,92]. The NiFeP@Fe3O4(x = 2.4) catalyst showed the highest C dl and also has the thickest self-generated carbon layer, whereas the Ni12P5 nanoparticles showed the lowest C dl among all tested catalysts and which have negligible carbon coverage. Presumably, carbon layers separate particles from each other and provide more area due to an increased access to some “inner surfaces” and leads to higher C dl. These observations suggest that the presence and thickness of the carbon coverage display a prominent role in the obtained value for C dl and consequently in the electrochemical active surface area.Oxygen evolution at catalysts covered by a carbon layer would require transport of reaction products and reactants either directly through the layer itself or through pinholes in the layer. Reaction through pinholes is not likely due to the very high activity of these catalysts; the catalytic activity would have to be rather extreme to explain this, and this is not compatible with the stability measurements indicating that the carbon layer does protect the catalysts. According to the Pourbaix diagrams, most transition metal phosphides will not be stable under OER conditions [32]. The fact that our catalysts are stable, indicates that the layer does keep the phosphides from disintegrating.) A direct influence of the metal on the carbon as suggested by Cui et al. for carbon monolayers [34] is not likely in view of the thickness of the carbon layers in this work. Also, the observation of pronounced pre-catalytic redox peaks attributed to Ni3+∕Ni2+ (Fig. 5) rules out carbon as being the only electrochemically active site. Other options are diffusion of iron into the carbon creating a carbon iron catalyst [89] or exfoliation of the carbon layer, providing electrolyte access to the metal sites underneath [45]. However, our post-mortem TEM images clearly shows that the carbon layer is completely preserved after being exposed to the OER conditions, excluding exfoliation of the carbon layer as a possibility.The quite intense Ni3+∕Ni2+ redox peak shows that the surface of the Ni12−x Fe x P5 particles caged inside the carbon layer is electrochemically active. Therefore, this Ni12−x Fe x P5 surface is likely to contribute significantly to the catalytic activity in the OER potential region as well. We tentatively propose a mechanism in which the hydroxide anions are transported through the carbon layer, possibly in a fashion similar to intercalation. Slow reaction steps are catalyzed at the Ni12−x Fe x P5 surface, and reaction intermediates formed in steps downstream of the rate-determining step are transported by diffusion in the graphitic or disordered carbon layers, again in a similar fashion to intercalation. The exact details of theses processes will, however, have to await further investigation beyond the scope here.A solution phase synthetic method for discrete Ni12−x Fe x P5 (x = 0, 1.2, 2.4, 3.6) nanoparticles was developed. The ternary Ni10.8Fe1.2P5 nanoparticles have a tetragonal crystal structure corresponding to that of Ni12P5, indicating the formation of a uniform Ni-Fe-P alloy. However, the XRD results showed that for the x > 1.2, particles with a core-shell structure were formed, in which a NiFeP alloy forms the core and Fe3O4 the shell (NiFeP@Fe3O4). A detailed inspection of the TEM images revealed that in effect a self-generated porous carbon layer covers Ni12−x Fe x P5 nanoparticles. This carbon layer is formed as a result of the decomposition of organic precursors during synthesis. We observed no such carbon layer for Ni12P5 nanoparticles synthesized under the same conditions, i.e with Ni12P5 catalysts not containing iron. This suggests that the decomposition of organic compounds are catalyzed by the bimetallic system (i.e. NiFe). Encapsulation of the particles with carbon was further substantiated with Raman spectroscopy, in which the two characteristic peaks of carbon at 1395 and 1520 cm−1 were clearly observed and are attributed to the carbon D and G-band, respectively. Based on the TEM images, the self-generated porous carbon layer was thickest (~ 5 nm) for NiFeP@Fe3O4(x = 2.4) nanoparticles and thinnest (~ 1 nm) for Ni10.8Fe1.2P5 nanoparticles. All the as-synthesized nanoparticles were applied as electrocatalysts for the OER. The activity for the OER increases in the order Ni12P5 < Ni10.8Fe1.2P5 < NiFeP@Fe3O4(x = 3.6) < NiFeP@Fe3O4(x = 2.4). NiFeP@Fe3O4(x = 2.4) nanoparticles showed an extraordinary electrocatalytic activity by achieving 10 mA cm−2 at 220 mV. A difference in the Tafel slopes between catalysts containing iron and Ni12P5 indicates that the reaction mechanism for the OER changes as iron is included in the composition. Post-mortem TEM characterization of NiFeP@Fe3O4(x = 2.4) showed that the carbon layer is very stable and is preserved after OER, consistent with in-situ Raman spectra which did not show any significant structural change upon exposure to the potentials at which the OER proceeds (1.6 V∕51 mA cm−2) for 5 h.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors acknowledge the financial support from the Norwegian University of Science and Technology (NTNU) (project no. 81771154). Fatemeh Poureshghi: Conceptualization, Investigation, Methodology, Writing – original draft, Writing – review & editing. Frode Seland: Supervision, review, Methodology. Jens Oluf Jensen: Funding acquisition, Supervision, review, Methodology. Svein Sunde: Funding acquisition, Conceptualization, Supervision (investigation), Methodology, Writing – review & editing, Project administration.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.apcata.2022.118786. Supplementary material .

Rational design of efficient, earth-abundant, and durable electrocatalysts to accelerate the oxygen evolution reaction (OER) is critical for hydrogen ion by water electrolysis. In the present work, nanostructured Ni12−x Fe x P5 (x = 1.2, 2.4, 3.6) OER electrocatalysts synthesized by a colloidal method is reported. For x = 1.2, an alloy of Ni, Fe, and P is formed. For x = 2.4 or x = 3.6, a core-shell NiFeP@Fe3O4 structure is formed. The nanoparticles are encapsulated in a self-generated carbon layer. The carbon layer is formed during synthesis from synthesis residues. The carbon-encapsulated Ni9.6Fe2.4P5 catalyst offers the outstanding mass activity of 0.1 A mg−1 and overpotential of 220 mV at 10 mA cm−2, assigned to a combination of enhanced electrical conductivity provided by the carbon shell, a large surface area, and a high specific catalytic activity. Post-mortem characterization indicates that the carbon encapsulation remains intact under conditions of the OER.

n-Hexane from petroleum and gasoline industries is a typical volatile organic compound (VOC), which has been widely used as polymerization reaction media, cleaning agent, and a solvent in industries, with the high volatility and high toxicity at low concentrations. The emissions of n-hexane can affect human health and contribute to air pollution [1,2]. Among various removal technologies, catalytic combustion is thought to be one of the predominant strategies for VOCs elimination because of its low cost, simple treatment, no secondary pollution, and maximum efficacy [3]. Supported precious metal-based catalysts (e.g., Pt, Au, Pd, Ru, and Rh) show the outstanding performance for VOCs oxidation, but their scalable industrial applications are limited owing to the expensiveness and easy sintering of precious metals [4,5]. Thus, it is necessary to develop the promising catalysts with both low cost and high efficiency for VOCs removal.Considering the alternative to noble metals, transition-metal oxides have been focused in recent years because of their earth abundance, inexpensive cost, and high activity [3]. Among them, chromium oxide attracts a great attention for catalytic VOCs oxidation due to its strong oxidizing ability, insolubility, and chemical stability [6]. For example, Xing et al. generated the CrO x /γ-Al2O3 catalysts, and found that the surface Cr6+ species were the active sites, and chromia with monolayer dispersion presented the optimum catalytic activity for benzene oxidation [7]. Tian et al. used the sol–gel method to synthesize the chromium oxides (CrO x ) catalysts, and claimed that Cr-300 showed the best performance for the oxidative dehydrogenation of propane (ODP) to propene owning to the smallest crystallite grain size and the highest Cr6+/Cr3+ and Olatt/Oads atomic ratios. In addition, the DFT calculations reveal that the Cr–O site is the leading active site in the ODP reaction [8]. Working on the ODP over the Cr-MSU-x catalysts, Baek et al. pointed out that the initial composition of the soft Cr(VI) in the total Cr(VI) was a major dominant factor governing the catalytic performance [9]. Actually, in order to enhance catalytic activity, polymetallic oxides usually exhibit better performance than individual metal due to the synergistic effect between the different metals of the former. Specially, the cobalt and nickel species with relatively low cost and high activity have been applied in various industries. For instance, Greluk et al. prepared the CeO2-supported Co and Ni catalysts for the steam reforming of ethanol, and thought that the cobalt/nickel terrace was the preferential reaction sites but the edge/steps sites favored the cleavage of the C–C bond; moreover, good dispersion and strong metal−oxide interactions between Co or Ni and CeO2 could modify chemical properties of the catalyst [10]. After loading bimetallic Co–Ni on alumina for methane combustion, Choya et al. found that simultaneous loading of Co and Ni improved redox property of the catalyst due to partial inhibition of the interaction between alumina and Co3O4 and favorable generation of NiCo2O4 [11]. Li et al. designed the atomic Co/Ni and Co–Ni alloy nanoparticles (NPs) in N-ZIF-67 for bifunctional oxygen electrocatalysis, and claimed that the atom-level Co/Ni dual active sites exhibited a high electrocatalytic activity than single noble-metal-free catalyst because of the synergistic impact of the atomic Co/Ni–N–C bonds and microstructure in the sample [12]. Nowadays, single-atom catalysts (SACs) have been deemed as a promising material in various fields due to its maximum atom utilization efficiency, excellent performance, and strong metal−support interaction [13,14]. To the best of our knowledge, nevertheless, preparing bimetallic cobalt−nickel single-atom catalysts for VOCs combustion have been rarely reported in the literature. Herein, we developed a facile approach to prepare the mesoporous chromic oxide (meso-Cr2O3)-supported bimetallic cobalt−nickel single-atom (Co1Ni1/meso-Cr2O3) and bimetallic Co and Ni nanoparticle (CoNPNiNP/meso-Cr2O3) catalysts, measured their physicochemical properties, evaluated their catalytic performance for n-hexane combustion, and clarified the involved catalytic mechanisms.Mesoporous silica (KIT-6) template was fabricated according to the procedures stated in the literature [15]. Three-dimensionally (3D) ordered mesoporous Cr2O3 (denoted as meso-Cr2O3) was synthesized with KIT-6 as hard template. The synthesis steps are as follows: 1.0 g of KIT-6 was added to 20 mL of ethanol solution, followed by the ultrasonic treatment for 0.5 h. 2.0 g of Cr(NO3)3⋅9H2O was then added to the above KIT-6- and ethanol-containing mixture solution, followed by the ultrasonic treatment until ethanol was completely volatilized to obtain the precursor@KIT-6 composite. The above obtained composite was dried in an oven at 60 °C for 10 h. After that, the precursor@KIT-6 composite was calcined at a ramp of 2 °C min−1 in a muffle furnace from room temperature (RT) to 400 °C and maintained at 400 °C for 5 h. Thus, the meso-Cr2O3 support was generated after KIT-6 template removal by washing with a sodium hydroxide aqueous solution (2.00 mol/L) at 80 °C for 2 h three times and drying at 60 °C for 10 h.The meso-Cr2O3-supported Co and Ni NPs (CoNPNiNP/meso-Cr2O3) catalyst was fabricated by the one-pot polyvinyl alcohol (PVA)-protecting method. Typically, 13.4 mg of Co(NO3)2⋅6H2O and 13.7 mg of Ni(NO3)2⋅6H2O were added to the ethanol−water mixed solution (ethanol/water volumetric ratio = 2:3) with 0.01 g of PVA under stirring for 10 min, then 0.08 g of NaBH4 was added to the above mixed solution and stirred for 10 min. After that, 0.5 g of meso-Cr2O3 was added to the above mixture and stirred for 12 h. Finally, the CoNPNiNP/meso-Cr2O3 catalyst was obtained after being filtered and dried in an oven at 60 °C for 12 h, and calcined at a ramp of 2 °C min−1 in the muffle furnace from RT to 400 °C and maintained at 400 °C for 5 h.The Co1/meso-Cr2O3 or Ni1/meso-Cr2O3 catalyst was prepared using the one-pot polyvinyl pyrrolidone (PVP)-protecting method with vitamin C as reducing agent. Typically, 26.7 mg of Co(NO3)2⋅6H2O or 29.7 mg of Ni(NO3)2⋅6H2O was added to the ethanol−water mixed solution (ethanol/water volumetric ratio = 2:3) with 0.01 g of PVP under stirring for 30 min, then 0.17 g of vitamin C was added and stirred for 2 h. After that, 0.5 g of meso-Cr2O3 was added to the above mixture and stirred for 6 h. Finally, the Co1/meso-Cr2O3 or Ni1/meso-Cr2O3 catalyst was generated after filtration and drying in an oven at 60 °C for 12 h, and calcination at a ramp of 2 °C min−1 in the muffle furnace from RT to 400 °C and maintaining at 400 °C for 5 h.The Co1Ni1/meso-Cr2O3 catalyst was prepared adopting the one-pot PVP-protecting method with vitamin C as reducing agent. Typically, 6.2 mg of Co(NO3)2⋅6H2O, 5.3 mg of Ni(NO3)2⋅6H2O, and 0.01 g of PVP were added to an ethanol−water mixed solution (ethanol/water volumetric ratio = 2:3) under stirring for 30 min. Then, 0.15 g of vitamin C was added to the above mixed solution and stirred for 2 h. After that, the mixture containing Co(NO3)2⋅6H2O and Ni(NO3)2⋅6H2O were mixed, and 0.5 g of meso-Cr2O3 was afterwards added to the above mixture and stirred for 6 h. Finally, the Co1Ni1/meso-Cr2O3 catalyst was obtained after filtration and drying in an oven at 60 °C for 12 h, followed by calcining at a ramp of 2 °C min−1 in the muffle furnace from RT to 400 °C and keeping at 400 °C for 5 h.Physicochemical properties of all of the as-prepared catalysts were measured using the techniques as follows: inductively coupled plasma−atomic emission spectroscopy (ICP−AES), small- and wide-angle X-ray diffraction (XRD), high-resolution transmission electron microscopy (HRTEM) and elemental mapping, high-angle annular dark field−scanning transmission electron microscopy (HAADF-STEM), and X-ray absorption spectroscopy (XAS), N2 adsorption−desorption (BET), X-ray photoelectron spectroscopy (XPS), oxygen temperature-programmed desorption (O2-TPD), hydrogen temperature-programmed reduction (H2-TPR), n-hexane temperature-programmed desorption (n-hexane-TPD), n-hexane temperature-programmed surface reaction (n-hexane-TPSR), density functional theory (DFT) calculations, gas chromatography−mass spectrometry (GC–MS), and in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS). The detailed measurement procedures are presented in the Supplementary material.Catalytic activities of the samples for n-hexane combustion were evaluated in a continuous flow fixed-bed quartz tubular microreactor (i.d. = 6.0 mm) at 1 atm. 50 mg of the catalyst was mixed with 0.25 g of quartz sand to be loaded in the microreactor. The total flow rate of the (1000 ppm n-hexane + 20 vol% O2 + N2 (balance)) gas mixture was 33.3 mL min−1, giving a space velocity (SV) of 40,000 mL g−1 h−1. 5.0 vol% CO2 and/or 10.0 vol% H2O were introduced to the reaction system, so that their effects on catalytic activity were examined. The reactants and products were detected online by a gas chromatograph. The n-hexane conversion was defined as (C inlet − C outlet)/C inlet × 100%, where the C inlet and C outlet are the inlet and outlet n-hexane concentrations in the feed stream, respectively.The actual metal contents in the as-prepared samples are listed in Table 1 . The actual Co contents in Co1/meso-Cr2O3, CoNPNiNP/meso-Cr2O3, and Co1Ni1/meso-Cr2O3 were 0.96, 0.45, and 0.17 wt%, respectively. The actual Ni contents in Ni1/meso-Cr2O3, CoNPNiNP/meso-Cr2O3, and Co1Ni1/meso-Cr2O3 were 1.04, 0.47, and 0.14 wt%, respectively. The XRD analysis was performed to identify the crystal phases of the samples. There were diffraction peaks of each sample at 2θ = 24.5°, 33.6°, 36.2°, 39.7°, 41.5°, 44.2°, 50.2°, 54.9°, 57.1°, 58.4°, 63.4°, 65.1°, 72.9°, 73.3°, 76.8°, and 79.1° (Fig. 1 ), matching well with the (012) (104) (110) (006) (113) (202) (024) (116) (211) (122) (214) (300) (1010) (119) (220), and (306) lattice planes, respectively, which were due to the rhombohedral Cr2O3 phase (JCPDS PDF no. 38–1479). The results indicate that each sample exhibits a rhombohedral Cr2O3 crystal structure. In the meanwhile, no reflections of the Co and/or Ni phases were detected, probably owing to lower Co and Ni loadings in Ni1/meso-Cr2O3, CoNPNiNP/meso-Cr2O3, Co1/meso-Cr2O3, and Co1Ni1/meso-Cr2O3. Crystallite sizes (D c) of all of the samples were calculated using the (116) crystal plane of Cr2O3 according to the Scherrer equation, and their results are summarized in Table 1. The D c value (17.8 nm) of meso-Cr2O3 was bigger than those (15.4–17.7 nm) of the other samples. In addition, diffraction peak intensity of the other samples was weaker compared with that of meso-Cr2O3, indicating the decrease in crystallite size after loading of Co and/or Ni. Fig. 2 presents TEM images in various regions of meso-Cr2O3 and Co1Ni1/meso-Cr2O3. The meso-Cr2O3 sample exhibited a high-quality 3D ordered mesoporous (3DOM) architecture with a mesopore diameter of ca. 11 nm. As expected, the well-ordered mesoporous structure remained perfectly after the loading of bimetallic Co and Ni single atoms, and lattice spacing of the Cr2O3 (012) crystal plane was ca. 0.36 nm. This result indicates that reduction treatment did not induce a remarkable change in 3DOM structure. There was a diffraction peak at 2θ = 0.85–1.00° in low-angle XRD pattern of each sample (Fig. S1), and a H2-typed hysteresis loop in the relative pressure (p/p 0) range of 0.70–1.00 was observed in the isotherm of each sample, together with a peak in pore-size distribution of each sample (Fig. S4). These results demonstrate that the ordered mesopores were generated in each sample. Furthermore, surface areas of these samples were 75–92 m2/g (Table 1). The meso-Cr2O3 with a regularly porous channel structure and a larger surface area can be regarded as an ideal support, which was beneficial for the adsorption, diffusion, and activation of reactant molecules, and could effectively prevent the migration and aggregation of Co and/or Ni atoms, expose more active sites, and enhance the interface effect between meso-Cr2O3 and Co or Ni atoms. Fig. 3 A−C displays TEM images of the Co1/meso-Cr2O3, Ni1/meso-Cr2O3, and Co1Ni1/meso-Cr2O3 samples, from which the Co or Ni NPs and/or nanoclusters were hard to be observed on the meso-Cr2O3 surface, Moreover, the energy-dispersive spectroscopic (EDS) element mappings analysis (Fig. 3a−c) reveals that the Cr, O, Co and/or Ni elements were homogenously distributed in each of the samples. More importantly, the high-angle annular dark field− STEM (HAADF−STEM) technique was used to disclose the dispersion of single Co and Ni atoms (Fig. 3D, E, and S2(A-C)). As a result, a number of small isolated bright spots were clearly observed, assignable to the Co and Ni single atoms due to their Z-contrasts higher than those of the Cr and O atoms. Furthermore, the single-atom EEL spectroscopy (Fig. S2(E)) was used to better distinguish the Co and Ni single atoms, but no significant signals were observed in the sample. On the one hand, the atomic numbers of Co, Ni, and Cr are similar; on the other hand, although the extranuclear electrons of Co and Ni atoms were excited, the meso-Cr2O3 support had a certain thickness. The excited extranuclear electrons might be extinguished in the substrate and could not penetrate the substrate, which was not received by the detector. In addition, we also made the statistics of the distance between two adjacent single atoms in blue square region in Fig. S2(C), and the results are presented in Fig. S2(D). The XAS measurements were conducted to further reveal the coordination environment in the Co1Ni1/meso-Cr2O3 sample. However, the absorption edge energies of Co, Ni, and Cr were about 7709, 8333, and 5989 eV, respectively. On the one hand, when the cobalt and nickel were co-existed in the sample, the edge collision absorption phenomenon could occur during the test. On the other hand, the support material possessed a high content of Cr, which resulted in the self-absorption phenomenon after light irradiation. Hence, as shown in Fig. S3, the above-mentioned factors led to the inability to measure the results of Ni signals. Furthermore, it was also difficult to determine the results of Co signals.The XPS experiments were done to identify elemental compositions and chemical states of the samples, and their XPS spectra and quantitative analysis results are shown in Fig. 4 A−D and Table 1, respectively. The Cr 2p3/2 spectrum (Fig. 4A) of each sample was deconvoluted into four components that were associated with the surface Cr2+ (binding energy (BE) = 575.1–575.3 eV), Cr3+ (BE = 576.3–576.6 eV), Cr5+ (BE = 577.6–577.9 eV), and Cr6+ (BE = 578.9 eV) species [16–18]. The (Cr5+ + Cr6+)/(Cr2+ +Cr3+) molar ratio increased in the order of meso-Cr2O3 (0.27) < Ni1/meso-Cr2O3 (0.37) < CoNPNiNP/Cr2O3 (0.41) < Co1/meso-Cr2O3 (0.43) < Co1Ni1/meso-Cr2O3 (0.48). It was reported that the Cr species with a higher chemical state (Cr5+ or Cr6+) (i.e., higher electronegativity) was beneficial for the redox reaction [19]. The O1s spectrum of each sample was divided into three components at BE = 529.5–529.9, 531.1–531.6, and 532.7–532.8 eV (Fig. 4B), which was related to the surface lattice oxygen (Olatt) in the form of M−O (M = Cr, Ni or Co) bond, adsorbed oxygen (Oads), and adsorbed molecular water ( O H 2 O ) or carbonate species [20], respectively. The components of Co 2p3/2 spectrum at BE = 782.2 and 786.6 eV corresponded to the surface Co2+ species and the satellite signal (Fig. 4C), respectively. For CoNPNiNP/meso-Cr2O3, the component at BE = 778.4 eV was related to the surface metallic Co0 species, the one at BE = 780.2 eV of Co1/meso-Cr2O3 in Co 2p3/2 spectrum belonged to the surface Co3+ species, and the one at BE = 779.1 eV of Co1Ni1/meso-Cr2O3 in Co 2p3/2 spectrum was associated with the surface Co δ+ (3 > δ > 2) species [21]. In the Ni 2p3/2 spectra (Fig. 4D) of the Ni1/meso-Cr2O3, CoNPNiNP/meso-Cr2O3, and Co1Ni1/meso-Cr2O3 samples, there were four components at BE = 853.7, 855.7–855.9, 859.2–859.9, and 861.8–864.3 eV, ascribable to the surface metallic Ni0, Ni2+, and Ni3+ species and the satellite signal [22], respectively. The results indicate that the Co1 or/and Ni1 in Ni1/meso-Cr2O3, Co1/meso-Cr2O3 or Co1Ni1/meso-Cr2O3 existed in an oxidized valence state, whereas the CoNP and NiNP in CoNPNiNP/meso-Cr2O3 were present in the form of combined oxidized with metallic valence states. Compared with meso-Cr2O3, Cr 2p3/2 peaks of the Ni1/meso-Cr2O3, CoNPNiNP/meso-Cr2O3, Co1/meso-Cr2O3, and Co1Ni1/meso-Cr2O3 samples were shifted to the positions with higher BEs, suggesting a strong synergistic effect between Ni and Co atoms or NPs and meso-Cr2O3, and their corresponding O 1s peaks also exhibited higher BE values in contrast to those on meso-Cr2O3. It can be inferred that electron transfer may proceed along a route of meso-Cr2O3 → Ni and/or Co atoms or NPs in those samples, where meso-Cr2O3 was regarded as electron acceptor while Ni and/or Co atoms or NPs as electron donor. Furthermore, the component of O 1s spectrum on the Co1Ni1/meso-Cr2O3 sample displayed a higher BE than that on the Ni1/meso-Cr2O3, CoNPNiNP/meso-Cr2O3, and Co1/meso-Cr2O3 samples, and the component of Ni 2p3/2 spectrum on Co1Ni1/meso-Cr2O3 also showed a higher BE than that on Ni1/meso-Cr2O3 or CoNPNiNP/meso-Cr2O3. This result indicates that loading bimetallic Co and Ni single atoms was more conducive to accumulating electrons on the surface of the double active sites to optimize the ΔG H∗, thus improving the catalytic performance [23].The O2-TPD experiments were conducted to measure the types and mobility of oxygen species of the samples, and their profiles are shown in Fig. 5 . The peaks at < 220 °C, 220–420, and above 420 °C were assigned to desorption of the surface adsorbed oxygen (Oads), surface lattice oxygen, and bulk lattice oxygen (Olatt) [20], respectively. The loading of the highly dispersed Co or Ni atoms might generate the lattice defects and facilitate mobility of the Olatt species on/in the meso-Cr2O3 support. The surface Olatt desorption temperature (377 °C) and bulk Olatt desorption temperature (803 °C) from the CoNPNiNP/meso-Cr2O3 sample were higher than those (373 and 581–703 °C) from the Co1/meso-Cr2O3 sample and those (363 and 596–716 °C) from the Ni1/meso-Cr2O3 sample, respectively, implying a lower lattice oxygen mobility in/on the former. Desorption peaks of the lattice oxygen species from the Co1Ni1/meso-Cr2O3 sample were observed at lower temperatures than those from the other samples, demonstrating that loading of the highly dispersed bimetallic Co and Ni active sites can significantly activate the bond between surface metal sites and surface Olatt to promote the mobility of lattice oxygen. It is generally accepted that surface Oads species plays a vital role in catalytic reactions governed by a suprafacial catalytic process [24]. The Co1Ni1/meso-Cr2O3 sample with the lowest desorption temperature of the surface Oads species exhibited the highest catalytic activity, indicating the importance of the surface Oads species in n-hexane combustion. In addition, mobility of the Olatt species was in favor of a redox reaction [25], thus promoting the enhancement in catalytic n-hexane combustion activity.The H2 temperature-programmed reduction (TPR) technique was applied to assess reducibility of all of the samples, and their H2-TPR curves are depicted in Fig. 6 A. The meso-Cr2O3 support exhibited a four-stage reduction feature with a weak reduction peak at 320 °C, a strong shoulder at 391 °C, a distinct reduction peak at 640 °C, and a weak reduction peak at 858 °C. The first peak was attributed to the reduction of Cr6+ to Cr5+ (or Cr3+) [26–28], the second one was assigned to the reduction of Cr5+ to Cr3+, the third one was attributed to the reduction of Cr3+ to Cr2+ [29,30], and the last one was associated with the direct reduction of bulk chromia to Cr2+ [30], accompanied by the removal of the surface Oads, surface Olatt, and bulk Olatt species, respectively. After loading of Ni single atoms, there were a sharp peak centered at 197 °C, a weak shoulder between 280 and 337 °C, a broad peak at 459 °C, a weak peak around 694 °C, and a peak at 851 °C, which corresponded to the reduction of Cr6+ to Cr5+ (or Cr3+) with removal of the Oads species, Cr5+ to Cr3+ with removal of the weak surface Olatt species, Cr3+ to Cr2+ and Ni3+ to Ni2+ with removal of the partial bulk Olatt species [31], Ni2+ to Ni0 with removal of the partial bulk Olatt species [32], and bulk chromia to Cr2+ with removal of the deep bulk Olatt species, respectively. As for the Co1/meso-Cr2O3 sample, there were an obvious peak located at 233 °C, a weak and broad peak at 396 °C, a wide peak at 637 °C, and an extremely weak peak at 867 °C, which represented the reduction of Cr6+ to Cr5+ (or Cr3+) with removal of the Oads species, Cr5+ to Cr3+ and Co3+ to Co2+ [33] with removal of the weak surface Olatt species, Cr3+ to Cr2+ and Co2+ to Co0 [34] with removal of the partial bulk Olatt species, and bulk chromia to Cr2+ with removal of the deep bulk Olatt species, respectively. The CoNPNiNP/meso-Cr2O3 sample exhibited a distinct peak at 231 °C, a band at 293 °C, a weak shoulder between 390 and 490 °C, a broad peak at 593 °C, and a peak at 823 °C, which were ascribed to the reduction of Cr6+ to Cr5+ (or Cr3+) with removal of the weakly adsorbed Oads species, Cr5+ to Cr3+ with removal of the partial surface Olatt species, Ni3+ to Ni2+ with removal of the partial bulk Olatt species, Cr3+ to Cr2+ and Ni2+ to Ni0 or Co2+ to Co0 with removal of the partial bulk Olatt species, and bulk chromia to Cr2+ with removal of the deep bulk Olatt species, respectively. The Co1Ni1/meso-Cr2O3 sample showed a distinct peak at 213 °C, a tiny peak at 392 °C, a wide peak at 490 °C, and an especially faint peak at 860 °C, attributable to the reduction of Cr6+ to Cr5+ (or Cr3+), Cr5+ to Cr3+ and Co δ+ (3 > δ > 2) and/or Ni3+ to Co2+ and/or Ni2+, Cr3+ to Cr2+ and Co2+ and/or Ni2+ to Co0 and/or Ni0, and bulk chromia to Cr2+, respectively, which was also accompanied by the consumption of the corresponding oxygen species. Additionally, H2 consumption of the samples are summarized in Table 2 . Obviously, the total hydrogen consumption followed a declined sequence of meso-Cr2O3 (1.14 mmol gcat −1) > Co1/meso-Cr2O3 (0.99 mmol gcat −1) > Co1Ni1/meso-Cr2O3 (0.92 mmol gcat −1) > Ni1/meso-Cr2O3 (0.88 mmol gcat −1) > CoNPNiNP/meso-Cr2O3 (0.86 mmol gcat −1). It should be mentioned that hydrogen consumption of the first peak obeyed a decreased order of Co1/meso-Cr2O3 (0.59 mmol gcat −1) > CoNPNiNP/meso-Cr2O3 (0.44 mmol gcat −1) > Co1Ni1/meso-Cr2O3 (0.39 mmol gcat −1) > Ni1/meso-Cr2O3 (0.31 mmol gcat −1) > meso-Cr2O3 (0.30 mmol gcat −1). The first reduction signal was associated with the reduction of chromia with the high-valence states as well as the removal of the Oads species, and the higher H2 consumption and the shift to lower temperatures would be beneficial for the redox reaction. It can be seen that the strong interaction between the highly dispersed Ni atoms and meso-Cr2O3 was more conducive to the improvements in mobility of the surface Olatt species and redox ability (from Cr6+ to Cr5+ (or Cr3+) and Cr5+ to Cr3+). Loading Co atoms tended to generate higher contents of chromium with the high-valence states and surface oxygen species. Apparently, the low-temperature reducibility declined in the order of Ni-meso/Cr2O3 > Co1Ni1/meso-Cr2O3 ≈ Co1/meso-Cr2O3 > CoNPNiNP/meso-Cr2O3 > meso-Cr2O3. Besides, the initial H2 consumption rate (at which less than 25% oxygen in the sample is consumed for the first reduction peak) presented in Fig. 6B also possessed such a changing trend.The n-hexane-TPD technique were used to probe n-hexane adsorption behaviors of the samples. The desorption signals of C6H14 (m/z = 57), CO2 (m/z = 44), H2O (m/z = 18), CO (m/z = 28), acrylic acid (m/z = 15), and 2-methyloxirane (m/z = 31) were recorded in n-hexane-TPD profiles of the samples. From Fig. 7 A, we can observe that there is a sharp n-hexane desorption peak at 82 °C for Co1Ni1/meso-Cr2O3 and an obvious broad peak at about 276 °C for CoNPNiNP/meso-Cr2O3, which are related to desorption of the physically or weakly chemically adsorbed n-hexane and the strongly chemically adsorbed n-hexane, respectively. Extremely weak desorption peaks were observed for the Co1/meso-Cr2O3 and Ni1/meso-Cr2O3 samples. Besides, the n-hexane adsorption capacity followed an increased order of Co1/meso-Cr2O3 (1.3 × 10−7 μmol gcat −1) < Ni1/meso-Cr2O3 (2.1 × 10−7 μmol gcat −1) < Co1Ni1/meso-Cr2O3 (59.2 × 10−7 μmol gcat −1)< CoNPNiNP/meso-Cr2O3 (89.6 × 10−7 μmol gcat −1) (Table 2). Significantly, Co1Ni1/meso-Cr2O3 showed a n-hexane desorption peak at the lowest temperature among all of the samples and a larger n-hexane adsorption capacity than the Co1/meso-Cr2O3 and Ni1/meso-Cr2O3 samples, suggesting that this sample possesses a stronger capability to adsorb n-hexane than the other samples. Since no gaseous oxygen was present in the experiments, the generated CO2 (Fig. 7B), H2O (Fig. 7C), CO (Fig. 7D), acrylic acid (Fig. 7E), and 2-methyloxirane (Fig. 7F) were attributed to the products due to interaction of the Oads and Olatt species with the adsorbed n-hexane on the sample surface. Additionally, the desorption peaks at < 220, 220–420, and >420 °C were associated with the reactions of the Oads, surface Olatt, and bulk Olatt species with the adsorbed n-hexane, respectively. It can be seen that the adsorbed n-hexane reacted with the surface Oads (and possibly a small amount of active surface Olatt species) to generate the intermediates, such as CO and acrylic acid. The large amount of CO2 and H2O generation suggests that the Olatt species mainly participate in the oxidation of adsorbed n-hexane and intermediates. Apparently, a lower desorption peak temperature was observed for the Co1Ni1/meso-Cr2O3 sample, furthermore a larger amount of CO2 was generated over this sample, indicating that the Co1Ni1/meso-Cr2O3 sample possesses a stronger ability to oxidize n-hexane than the other samples.The subsequent n-hexane-TPSR technique was used to investigate the relevance of the active oxygen species and n-hexane dissociation, and their profiles are depicted in Fig. 8 . Interestingly, when gaseous oxygen was introduced into the system, a n-hexane desorption peak appeared at 83 °C for the Co1/meso-Cr2O3 sample (Fig. 8A), which might be due to the fact that gaseous oxygen can more easily supply the surface Oads species to enhance the generation of oxygen vacancies, thus improving the ability to adsorb n-hexane on the surface of Co1/meso-Cr2O3. Additionally, the n-hexane desorption peak of Co1Ni1/meso-Cr2O3 decreased in intensity, and almost disappeared for CoNPNiNP/meso-Cr2O3 compared with that shown in Fig. 7A, demonstrating that there was a competitive adsorption between gaseous oxygen and n-hexane. Obviously, as presented in Fig. 8B−F, desorption peaks of the generated CO2, H2O, CO, acrylic acid, and 2-methyloxirane were shifted to lower temperatures, and increased in intensity as compared with the results of n-hexane-TPD characterization. The results indicate that gaseous oxygen can supplement the active oxygen species, thus accelerating the n-hexane oxidation rate [35,36]. It can be also viewed that the meso-Cr2O3 sample exhibits the higher desorption peak temperatures of MS signals due to the involved intermediates and products than the other samples, suggesting that loading Co1 and/or Ni1 as well as CoNPNiNP can effectively promote the active oxygen migration, thus increasing the catalytic performance of n-hexane combustion, in good accordance with the O2-TPD characterization results. Furthermore, the amounts of CO2 and other by-products generated over the Co1Ni1/meso-Cr2O3 sample were more than those formed during the n-hexane-TPSR process (consistent with the n-hexane-TPD process), indicating a strong ability of Co1Ni1/meso-Cr2O3 to adsorb and activate n-hexane in the presence of oxygen.To better clearly elucidate the adsorption of n-hexane on the sample surface, the projector augmented wave (PAW) method with the Vienna Ab Initio Simulation Package (VASP) based on the density functional theory (DFT) was applied to do the theoretical calculations. The DFT-optimized structures of model and related energies are presented in Fig. 9 and Table 4. The adsorption energies of n-hexane on the surface of the CoNPNiNP/meso-Cr2O3 and Co1Ni1/meso-Cr2O3 catalysts were compared. Besides, due to the fact that the particle size of nanoparticles is much larger than that of clusters, the stable sections of metal (Co (0001) or Ni (111)) are usually used for calculations in microscopic view. The adsorption energy (ΔE ads) was calculated according to the equation: ΔE ads = E total − E slab − E n-hexane, where E total is the total energy of the surface slab with n-hexane adsorption, E slab is the energy of the Co1Ni1/meso-Cr2O3 (116), Co (0001) or Ni (111) surface, and E n-hexane represents the energy of the isolated n-hexane molecule in the gas phase. It can be seen that the model of Co (0001) or Ni (111) exhibits a higher adsorption energy (−0.895 eV or −1.060 eV) than Co1Ni1/meso-Cr2O3 (116) (−1.371 eV), indicating that Co1Ni1/meso-Cr2O3 possessed a stronger n-hexane adsorption ability than CoNPNiNP/meso-Cr2O3.Catalytic activities of the as-prepared samples for n-hexane combustion are presented in Fig. 10 A and Table 3 . Apparently, n-hexane conversion over each sample increased with increasing the temperature, and catalytic activity below 220 °C dropped in the order of Co1Ni1/meso-Cr2O3 > Co1/meso-Cr2O3 > Ni1/meso-Cr2O3 > CoNPNiNP/meso-Cr2O3 > meso-Cr2O3, while that above 220 °C decreased in the sequence of Co1Ni1/meso-Cr2O3 > Co1/meso-Cr2O3 > CoNPNiNP/meso-Cr2O3 > Ni1/meso-Cr2O3 > meso-Cr2O3. Moreover, the temperatures (T 10%, T 50%, and T 90%) reaching 10, 50, and 90% n-hexane conversions are also used to compare catalytic performance of the samples, respectively. Obviously, the Co1Ni1/meso-Cr2O3 sample displayed the best catalytic activity: the T 10%, T 50%, and T 90% were 200, 239, and 263 °C at SV = 40,000 mL g−1 h−1, respectively, which were lower than those (207, 249, and 281 °C) over Co1/meso-Cr2O3, those (214, 260, and 289 °C) over CoNPNiNP/meso-Cr2O3, those (208, 270, and 303 °C) over Ni1/meso-Cr2O3, and those (276, 383, and 470 °C) over meso-Cr2O3, respectively. Such a result may be due to the fact that the strong synergistic interaction between bimetallic Co and Ni single atoms and meso-Cr2O3 support is more beneficial for generating the larger amount of higher-valence chromium ions and easier mobile active Olatt species to efficiently activate the C–H bonds in n-hexane; in the meanwhile, the stronger n-hexane adsorption and activation ability of the Co1Ni1/meso-Cr2O3 sample can accelerate the n-hexane combustion process. In addition, the CoNPNiNP/meso-Cr2O3 sample displayed a better catalytic activity than the Ni1/meso-Cr2O3 sample at higher temperatures due to the stronger chemisorption of n-hexane on the former, but a worse catalytic activity at lower temperatures owing to the fact that the Ni1/meso-Cr2O3 sample possesses better low-temperature reducibility than the CoNPNiNP/meso-Cr2O3 sample.Apparent activation energies (E a) were obtained from the Arrhenius plots of ln k versus inverse temperature of the samples, and their results are shown in Fig. 10B and Table 3. The sequence in E a value increased according to Co1Ni1/meso-Cr2O3 (54.7 kJ mol−1) < Co1/meso-Cr2O3 (58.4 kJ mol−1) < Ni1/meso-Cr2O3 (62.2 kJ mol−1) < CoNPNiNP/meso-Cr2O3 (93.5 kJ mol−1) < meso-Cr2O3 (100.6 kJ mol−1), which was in consistency with that in n-hexane conversion above 220 °C. Moreover, the calculated specific reaction rate at 260 °C (4.3 × 10−7 mol gcat −1 s−1) over Co1Ni1/meso-Cr2O3 was higher than that (3.4 × 10−7 mol gcat −1 s−1) over Co1/meso-Cr2O3, that (2.5 × 10−7 mol gcat −1 s−1) over CoNPNiNP/meso-Cr2O3, that (1.9 × 10−7 mol gcat −1 s−1) over Ni1/meso-Cr2O3, and that (0.2 × 10−7 mol gcat −1 s−1) over meso-Cr2O3 (Table 3). Furthermore, the specific reaction rate at 260 °C for n-hexane combustion over Co1Ni1/meso-Cr2O3 was much higher than those over 0.4 Mn/Pt-1 nm (D) [37] and SS/ZrO2/Pt (SP) [38], but inferior to those over Mn0·7Ce0·3/Al2O3 [39] and 0.12Pt/0.4Mn/Al2O3 [40] reported in the literature (Table S1).Effect of the SV on catalytic performance of Co1Ni1/meso-Cr2O3 was investigated in the range of 5000–80,000 mL g−1 h−1. As shown in Fig. S5, catalytic activity dropped with the increased SV due to the reduction in residence time of the reactant gases on the sample surface [41]. It can be obviously observed that Co1Ni1/meso-Cr2O3 also exhibited a better catalytic activity (T 90% = 276 °C) even at a higher SV of 80,000 mL g−1 h−1. To better examine thermal stability of the sample under the kinetically controlled reaction conditions [42], 20-h on-stream n-hexane combustion was performed over the Co1Ni1/meso-Cr2O3 sample at two different temperatures (239 and 263 °C) and SV = 40,000 mL g−1 h−1. As illustrated in Fig. 10C, there was no remarkable loss in catalytic activity during the durability test process, demonstrating good thermal stability of Co1Ni1/meso-Cr2O3 under the adopted reaction conditions. Considering the possibility of CO2 and/or H2O presence in industrial VOCs emissions, we examined the influence of 5.0 vol% CO2 and/or 10.0 vol% H2O on catalytic activity of Co1Ni1/meso-Cr2O3 for n-hexane combustion, as shown in Fig. 10D. Apparently, n-hexane conversion was not distinctly changed after 5.0 vol% CO2 was added to the reaction feedstock, indicating the good resistance to CO2 of the Co1Ni1/meso-Cr2O3 sample. The introduction of 10 vol% H2O to the reaction system exerted a minor negative effect on activity, with the conversion of n-hexane being dropped by ca. 4% because of the competitive adsorption of H2O and reactants molecules [43]. After both 5.0 vol% CO2 and 10 vol% H2O were introduced into the reaction system at a T 90% for 8 h, there was no obvious changes in catalytic activity. Furthermore, crystalline phase and surface topography of the Co1Ni1/meso-Cr2O3 sample exhibited almost no differences. The sample (Co1Ni1/meso-Cr2O3-(H2O + CO2)) treated in both 5.0 vol% CO2 and 10.0 vol% H2O at 263 °C for 8 h, as illustrated in Figs. 1 and 3(F, G) showed the good stability and CO2- and H2O-resistant performance. Consequently, the Co1Ni1/meso-Cr2O3 sample with good CO2 and H2O resistance can be regarded as a potential catalyst for n-hexane combustion.In situ DRIFTS spectra of n-hexane adsorption with the increased temperatures were collected on the Co1/meso-Cr2O3, Ni1/meso-Cr2O3, and Co1Ni1/meso-Cr2O3 samples to compare reactivity of the Olatt species on/in the samples, as illustrated in Fig. 11 A, C, and E. The adsorption process was as follows: The samples were treated in a (1000 ppm n-hexane + N2 (balance)) mixture flow of 33.3 mL min−1 at 180 °C for 0.5 h (denoted as ad-saturation) after being pretreated in an O2 flow of 20 mL min−1 at 250 °C for 1 h and purging in a N2 flow of 20 mL min−1 at 250 °C for 1 h. Noticeably, the absorption bands at 2850–2980 (2966, 2934, and 2877 cm−1), 1466, and 1379 cm−1 were recorded on all of the samples, which were ascribable to the ν(C–H), δas (CH3), and δs (CH3) in adsorbed n-hexane [44,45], respectively. Besides, the bands in the range of 3100–3600 and 2300−2400 cm−1 were due to the ν(O–H) and gas CO2 [44,46], respectively. For the Co1Ni1/meso-Cr2O3 sample, there were three new bands at 1717, 1626, and 1287 cm−1 owing to the vibration modes of ν(CO), ν(CC), and δ(C–H), respectively, which indicates the generation of 3-hexanone, 2-hexanone, and olefins (which were deduced after taking into consideration of the GC–MS results shown in Fig. S6). This result suggests that n-hexane was partially preferentially oxidized by surface labile lattice oxygen species. Obviously, several new characteristic bands were observed and the accumulation of CO2 in all of the spectra increased with a rise in temperature. The bands at 1637, 1626, 1539, 1532, and 1521 cm−1 were due to the vibration of ν(CC) in olefins; the ones at 1419 and 1326, 1383, and 1290 cm−1 were owing to the vibration of δ(O–H), δs (CH3), and δ(C–H) [47], respectively; and the ones at 1206 and 1217, 1086 and 1074, 1139, and 1246 cm−1 were ascribable to the vibration of ν(C–O) in alcohols, ν(C–O) in five-membered cyclic ether, ν(C–C), and ν(C–O) in epoxide ethers or alcohols, respectively. Additionally, the bands at 968, 920, 891, 771, 751, and 744 cm−1 were due to the vibration of the γ(C–H). Combining the results of GC–MS, it is evident that 2,5-dimethyltetrahydrofuran, 3-hexyl hydroperoxide, and olefins may be regarded as the main intermediates on the surface of Co1/meso-Cr2O3 or Ni1/meso-Cr2O3 with increasing temperature. In the case of the Co1Ni1/meso-Cr2O3 sample, there were not only the abovementioned intermediates but also 2-methyloxirane or 2-ethyl-oxetane. This result indicates that the adsorbed n-hexane reacts with the surface labile lattice oxygen species on each sample, in which loading bimetallic Co and Ni single atoms can effectively promote the activation of lattice oxygen to accelerate the n-hexane oxidation process.To further reveal the n-hexane combustion mechanism over the as-prepared samples, 20 vol% O2 was first introduced for 30 min (defined as O2-30 min) after the saturated adsorption of n-hexane, and in situ DRIFTS spectra (Fig. 11B, D, and F) were then recorded with the increased temperatures. When O2 was introduced to the Co1/meso-Cr2O3 or Ni1/meso-Cr2O3 sample with a rise in temperature, new adsorption bands at 1722 and 945 cm−1 due to the vibration of ν(CO) and γ(C–H) were recorded as compared with the feed gas without O2, respectively, and the ones at 1641, 1599, and 1612 cm−1 were owing to the ν(CC) of olefins, demonstrating generation of 3-hexanone and 2-hexanone (also according to the GC–MS results). As for the Co1Ni1/meso-Cr2O3 sample, the bands at 1626, 1539, 1419, 1246, 968, 891, and 744 cm−1 disappeared, and new bands at 1717, 1546, 1483, and 948 cm−1 might be associated with the vibration of ν(CO), ν as (COO), ν s (COO), and γ(C–H) [48], respectively, which implies that 2-methyloxirane, 2-ethyl-oxetane, and 3-hexyl hydroperoxide decrease in amount and the acrylic acid intermediate appears, while the band intensity of 3-hexanone, 2-hexanone, 2,5-hexanedione, and CO2 enhanced gradually (connecting with the GC–MS results). On the basis of the above results, the possible n-hexane combustion pathways over the as-obtained samples can be described as follows: n-hexane → olefins or 3-hexyl hydroperoxide → 3-hexanone, 2-hexanone or 2,5-dimethyltetrahydrofuran → 2-methyloxirane or 2-ethyl-oxetane → acrylic acid → CO x → CO2 and H2O. To sum up, n-hexane could be oxidized in the absence of gas-phase O2, demonstrating that the labile lattice oxygen species participate in the reaction. Gaseous oxygen could be activated to the active oxygen species adsorbed at oxygen vacancies caused by the consumption of labile lattice oxygen, suggesting the Mars−van Krevelen (MvK) mechanism. Significantly, loading bimetallic Co and Ni single atoms can markedly accelerate this process due to the facilely activated lattice oxygen species, resulting in the predominant catalytic performance.The 3DOM chromium oxide-supported Co and/or Ni single-atom (Co1/meso-Cr2O3, Ni1/meso-Cr2O3, and Co1Ni1/meso-Cr2O3) and CoNPNiNP/meso-Cr2O3 catalysts were prepared using the one-pot PVP- and PVA-protecting methods, respectively. As a result, catalytic activity below 220 °C decreased in an order of Co1Ni1/meso-Cr2O3 > Co1/meso-Cr2O3 > Ni1/meso-Cr2O3 > CoNPNiNP/meso-Cr2O3 > meso-Cr2O3, and that above 220 °C decreased in a sequence of Co1Ni1/meso-Cr2O3 > Co1/meso-Cr2O3 > CoNPNiNP/meso-Cr2O3 > Ni1/meso-Cr2O3 > meso-Cr2O3, which was due to the fact that CoNPNiNP/meso-Cr2O3 shows the strong chemisorption of n-hexane at higher temperatures but Ni1/meso-Cr2O3 possesses the better low-temperature reducibility. The Co1Ni1/meso-Cr2O3 catalyst showed the best activity (T 50% and T 90% were 239 and 263 °C at SV = 40,000 mL g−1 h−1, respectively), the lowest E a (54.7 kJ mol−1), and the highest specific reaction rate at 260 °C (4.3 × 10−7 mol gcat −1 s−1). Long-term stability and CO2 or H2O resistance tests over the Co1Ni1/meso-Cr2O3 sample can also be considered as a promising catalyst for VOCs combustion. The good catalytic performance was associated with the fact that the strong synergistic effect between Co1 and Ni1 and meso-Cr2O3 makes Co1Ni1/meso-Cr2O3 possess a larger amount of higher-valence chromium ions (Cr5+and Cr6+) and easily activated lattice oxygen species, which can efficiently promote the enhancement in n-hexane adsorption and activation ability and the breaking of C–H bonds in n-hexane. The combustion of n-hexane occurs via the MvK mechanism, and its possible pathways are as follows: n-hexane → olefins or 3-hexyl hydroperoxide → 3-hexanone, 2-hexanone or 2,5-dimethyltetrahydrofuran → 2-methyloxirane or 2-ethyl-oxetane → acrylic acid → CO x → CO2 and H2O.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the National Natural Science Committee of China−Liaoning Provincial People's Government Joint Fund (U1908204), National Natural Science Foundation of China (21876006, 21976009, and 21961160743), Foundation on the Creative Research Team Construction Promotion Project of Beijing Municipal Institutions (IDHT20190503), Natural Science Foundation of Beijing Municipal Commission of Education (KM201710005004), and Development Program for the Youth Outstanding−Notch Talent of Beijing Municipal Commission of Education (CIT&TCD201904019).The following is the Supplementary data to this article. Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.gee.2022.12.008.

Developing the alternative supported noble metal catalysts with low cost, high catalytic efficiency, and good resistance toward carbon dioxide and water vapor is critically demanded for the oxidative removal of volatile organic compounds (VOCs). In this work, we prepared the mesoporous chromia-supported bimetallic Co and Ni single-atom (Co1Ni1/meso-Cr2O3) and bimetallic Co and Ni nanoparticle (CoNPNiNP/meso-Cr2O3) catalysts adopting the one-pot polyvinyl pyrrolidone (PVP)- and polyvinyl alcohol (PVA)-protecting approaches, respectively. The results indicate that the Co1Ni1/meso-Cr2O3 catalyst exhibited the best catalytic activity for n-hexane (C6H14) combustion (T 50% and T 90% were 239 and 263 °C at a space velocity of 40,000 mL g−1 h−1; apparent activation energy and specific reaction rate at 260 °C were 54.7 kJ mol−1 and 4.3 × 10−7 mol gcat −1 s−1, respectively), which was associated with its higher (Cr5+ + Cr6+) amount, large n-hexane adsorption capacity, and good lattice oxygen mobility that could enhance the deep oxidation of n-hexane, in which Ni1 was beneficial for the enhancements in surface lattice oxygen mobility and low-temperature reducibility, while Co1 preferred to generate higher contents of the high-valence states of chromium and surface oxygen species as well as adsorption and activation of n-hexane. n-Hexane combustion takes place via the Mars−van Krevelen (MvK) mechanism, and its reaction pathways are as follows: n-hexane → olefins or 3-hexyl hydroperoxide → 3-hexanone, 2-hexanone or 2,5-dimethyltetrahydrofuran → 2-methyloxirane or 2-ethyl-oxetane → acrylic acid → CO x → CO2 and H2O.

Biomass is a valuable source to produce liquid fuels; however, the bio-oils obtained from it have high oxygen and water contents and require further upgrading [1]. Hydrodeoxygenation (HDO) is a preferred process to decrease the oxygen content for bio-oil upgrading [2]. HDO reactions involve the presence of a catalyst under a hydrogen atmosphere at temperatures between 200 °C and 400 °C and high pressure [3], in which oxygen is eliminated as water or carbon oxide(s) [4]. A diversity of materials such as metal sulfides [5], oxynitrides [6], phosphides [7], metal oxides [8], molecular sieve-supported metal catalysts [9], bifunctional catalysts [10], and carbides [11], have been used for bio-oil upgrading via HDO reactions [12]. In particular, molybdenum-based catalysts (MoS2, Mo2C, and MoO3) have shown good activity [13], with molybdenum sulfide being the most active one; however, it has low stability at the reaction conditions and could present sulfur loss. Hence, the preparation of stable MoS2 or alternative catalysts for HDO are highly desired [14]. Molybdenum carbides seem to be an excellent alternative to unstable molybdenum sulfide-based catalysts.Molybdenum carbides are attractive active phases due to their low cost, corrosion resistance, high melting point, and catalytic activity [11,15,16]. Nowadays, there is a particular interest in the control of different parameters as morphology, particle size, and the phase of molybdenum carbides [17]. Their catalytic activity is attributed to the permeation of carbon atoms into the molybdenum metal lattice, which lengthens the metal-metal distance, increasing the d-band electron density at the Fermi level of molybdenum [18].Recent studies have demonstrated that molybdenum carbides are active catalysts for a range of oxygenated compounds found in biomass-derived bio-oils [2,15]. When oxygen is eliminated as water during HDO, this could induce the deactivation of the catalysts by the formation of molybdenum oxides [19]. However, it has been found that molybdenum carbides could be stable at the HDO reaction conditions; i.e., in the presence of water [20].Ni-modified Mo2C catalysts show superior activity than Mo2C, Ni has high hydrogenation activity [21,22], but low electrophilicity in comparison with molybdenum, making it less favorable for the activation and direct scission of C=O and C–O bonds [23]. Mo2C catalysts have high selectivity to deoxygenation without the hydrogenation of furanic or aromatic rings, as it specifically facilitates the η2(C, O) adsorption of oxygenated compounds, for this reason, a direct scission of C=O or C–O bond is possible [16,24–26]. Wang et al. [27] studied Ni–Mo–C catalysts deposited on various supports in the hydroprocessing of soybean oil. The authors found that Ni–Mo–C active phase supported on mesoporous supports such as γ-Al2O3 and Al-SBA-15 showed an increased yield of hydrocarbons containing mainly C15–C18 associated with decarboxylation, decarbonylation, and hydrodeoxygenation reactions. On the other hand, the carbon nanotubes used as support could also serve as the carbon source for the Mo2C formation [28]. Mai et al. [20] reported that β-Mo2C supported on carbon nanotubes was an efficient catalyst for the selective conversion of levulinic acid into γ-valerolactone in the aqueous phase. In a turnover frequency (TOF) basis, the activity of the catalyst was similar to that obtained for a ruthenium catalyst evaluated under the same conditions.According to the literature, differentiated catalytic performance is observed when metal particles are located inside or outside of CNTs [29]. Moreover, the preparation of MoxC carbide supported inside multi-wall carbon CNTs and its catalytic performance in hydrodeoxygenation are yet not well known. This work proposed a controllable synthesis of CNTs-supported Ni-Mo carbide catalysts. Additionally, the Ni/Mo molar ratio was evaluated for HDO using benzofuran as a model compound for some bio-oil fractions; indeed, benzofuran is a good model [7,29–31] because the majority of organo-oxo compounds have either a phenolic or a furanic structure [32].Multi-walled carbon nanotubes (CNTs) with 10–20 μm length, external diameter 10–20 nm, and purity >98 wt%, were obtained from Timesnano Company (Chengdu Organic Chemicals Co. Ltd), ammonium heptamolybdate tetrahydrate ((NH4)6Mo7O24∙4H2O, 99%) was obtained from Sigma-Aldrich, and nickel nitrate hexahydrate (Ni(NO3)2∙6H2O, 98%) was obtained from Alfa Aesar.In a typical procedure, 1.0 g of CNTs was refluxed in HNO3 (65 wt%, 20 mL per gram of CNTs) using an oil bath for 10 h, then the CNTs were washed with deionized water until pH ~ 7, and then dried at 100 °C for 12 h. The catalysts were prepared by wet impregnation of CNTs with an aqueous solution obtained by dissolving (NH4)6Mo7O24∙4H2O and Ni(NO3)2∙6H2O with variable proportions (namely, Ni/Mo = 0.2, 0.3, 1.0, 3.0 M ratio) to get 1.3 mmol of metal (Ni + Mo) per gram of CNTs. During impregnation, the samples were sonicated for 1 h, and then the water was slowly evaporated at 25 °C and kept at 200 rpm using magnetic stirring, and then dried at 100 °C for 12 h. The precursors were then submitted to temperature-programmed carburization (TPC) under a stream of 20% CH4/H2 (100 mL min−1) using a heating ramp of 2 °C min−1 and kept at 700 °C for 2 h. The system was then cooled down under inert atmosphere (Ar), and passivated with an air/Ar mixture (10/90) at room temperature. Catalysts containing only Ni or Mo were prepared following the same procedure.Structure features of the catalysts were determined by X-ray diffraction (XRD) using a PANalytical X'pert PRO MPD diffractometer with Cu Kα radiation (λ = 1.5406 Å), in the 2θ angle range of 20° to 70° with a step size of 0.026° and period of 50 s. TEM micrographs were obtained using a Tecnai F20 Super Twin TMP instrument; samples were dispersed in acetone and sonicated for 15 min before being dropped on a carbon-coated copper grid. The specific surface area was determined using the N2 adsorption at −196 °C performed with an Autosorb-iQ from Quantachrome and using the Brunauer-Emmett-Teller (BET) model to estimate the specific surface areas. X-ray photoelectron spectra (XPS) were performed in an ultrahigh vacuum (UHV) using a SPECS multi-technique analysis instrument equipped with a monochromatic Al-Kα source (1486.7 eV, 13 kV, 100 W) and an electron analyzer PHOIBOS 150 1D-DLD. The step was 1 eV and 0.1 eV for the general and the high-resolution spectra, respectively.Preliminary tests were carried out to evaluate the catalytic performance of the different catalysis. Typically, for each run 300 μL (0.33 g) of benzofuran, 30 mg of catalyst, and 300 μL of tetradecane (as internal standard) in 10 g of hexadecane, were loaded into a 50 mL batch autoclave reactor equipped with an electromagnetic stirrer. The reaction was carried out under an H2 environment at 5 MPa with a stirring speed of 400 rpm at 280 °C for 4 h. Additionally, the temperature (200 °C–320 °C) and reaction time (2 h − 8 h) effect, and catalyst recycling, were evaluated using the catalyst with the best catalytic performance in the preliminary tests.At the end of each experiment, the products were separated from the catalyst by filtration and the identification and quantification were performed in a GC/MS Shimadzu (2010-Plus) equipped with a DB-5 capillary column (30 m × 0.25 mm × 0.25 μm) using n-tetradecane as the internal standard. The initial oven temperature was 50 °C for 1 min. The temperature was programmed to increase from 50 to 220 °C at 2 °C min−1 and hold for 10 min. The conversion of benzofuran (XBF), the yield of each product (Yi) and the selectivity (Si) were calculated in molar basis relative to the feed following Eqs. (1–3): (1) X BF = 1 − C BF C BF 0 × 100 % (2) Y i = C i C BF 0 × 100 % (3) S i = Y i X BF × 100 % Where CBF|0 is the initial concentration of benzofuran in the reactor feed, CBF and Ci is the molar concentration of benzofuran and the product species after reaction, respectively. Fig. 1a shows the preliminary results of benzofuran conversion and product distribution. Monometallic Ni/CNTs catalyst exhibited a higher conversion of 21.0% in comparison with 0.7% for Mo/CNTs. On the other hand, the products over Ni/CNTs were hydrogenated products, 2,3-dihydrobenzofuran (2,3-DHBF, yield: 18.1%), and octahydrobenzofuran (OHBF, yield: 2.6%), this is related to the high hydrogenation activity attributed to metallic Ni particles dispersed on the surface of CNTs [1,33]. The lowest conversion of Mo/CNTs, when molybdenum is found mainly as amorphous carbides (MoxC) according to XRD results, may be related to the smaller surface area determined by BET (Table S1), the lower activity of MoxC carbide, and the reaction conditions (reaction temperature of 280 °C). This results is consistent with literature reports which indicate that Mo2C-based catalyst showed activity for HDO reactions at temperatures higher than 300 °C for guaiacol and 350 °C for stearic acid and for the catalytic upgrading of residual biomass derived bio-oil [34,35].Ni and Mo have been used to prepare catalysts for a high number of reactions, in which the Ni/Mo ratio is a key parameter for the catalysts performance [36–39]. The results for NiMo-x/CNTs catalysts show an important effect of Ni/Mo molar ratio on the conversion and products distribution in the benzofuran HDO. As previously mentioned, without Ni, the conversion on Mo/CNTs was only 0.7%. After introducing both Ni and Mo, conversion, hydrogenation, and deoxygenation degree were increased. Reduced Ni acted as hydrogen activation sites supplying reactive hydrogen species for the hydrogenation reaction [40]. For example, 73.2% conversion was obtained on NiMo-0.2/CNTs, with 67.6% yield to 2,3-DHBF. For NiMo-0.3/CNTs, the BF conversion was nearly 100%, though the yield to 2,3-DHBF decreased to 64.6%; this catalyst showed an increase in the formation of OHBF with a yield of 19.9%, where OHBF is the product of the hydrogenation of 2,3-DHBF (see Fig. 1b). Additionally, the yield for HDO products, ethylcyclohexane (ECH) and methylcyclohexane (MCH), increased to 5.1%, which suggests that the reaction path begins with the hydrogenation of BF followed by the hydrogenation 2,3-DHBF. In contrast, for the NiMo-3/CNTs catalyst, the BF conversion decreased to 43.9%, with a 39.4% yield to 2,3-DHBF.On the other hand, reactions carried out in the presence of unsupported Ni-Mo2C-0.3 showed a BF conversion lower than 1.0% (Fig. S1). The low conversion of BF over unsupported catalyst clearly indicated a positive effect of CNTs as catalyst support, and the results showed in Fig. 1 suggest a synergistic effect when Ni and Mo are deposited in CNTs. Although the Ni/Mo ratio of 0.3 has a higher activity to OHBF and HDO products, the optimum Ni/Mo ratio can be different for other reactions. Recently, Smirnov et al. [41] prepared carbide catalysts with different Ni/Mo ratios (0, 0.5, 1, 2, and 6) using a method based on the Pechini process. The catalysts were evaluated in the hydrodeoxygenation of anisole and ethyl caprate when the Ni2MoC (Ni/Mo = 2) catalyst showed the highest activity, which is attributed to the presence of Ni-Mo–C active sites. Under adopted reaction conditions (280 °C), the main products were 2,3-DHBF and OHBF; thereby, high hydrogenating properties for these catalysts are inferred. The high activity of Ni/CNTs in comparison to Mo/CNT is noteworthy; some properties such as particle size and the evaluation of other reaction conditions could lead to obtaining valuable products, however, this is not further considered in this work. Fig. 1c shows the XRD patterns of the catalysts. The peaks at 2θ = 44.5°, 2θ = 51.8° for Ni/CNTs corresponds to (111, 002) planes for metallic Ni (ICSD 64989). In the case of the Mo/CNTs, diffraction peaks are roughly observed with the peak at 2θ = 37° and 2θ = 39° showing low intensity. On the other hand, NiMo/CNTs catalysts show peaks at 2θ = 34.5°, 2θ = 37.9°, and 2θ = 39.5° corresponding to (010, 002, 011) planes of β-Mo2C (ICSD # 77158). The results show that the Ni addition improves the β-Mo2C phase formation, which can be attributed to methane dissociation on metallic Ni particles, generating chemisorbed carbon species suitable for the formation of molybdenum carbide [42,43]. Samples prepared with molar ratios Ni/Mo in the range 0.2–1.0 show representative signals of the β-Mo2C hexagonal phase, namely, 2θ: 34.5° (010), 37.9° (002), 39.5° (011), and 52.2° (012). However, for the sample with a Ni/Mo = 3 (Ni-Mo/CNTs-3) any diffraction peak corresponding to β-Mo2C phase was detected, possibly due to the lower loading of Mo (about 3 wt%), as compared to other samples. On the other hand, all the samples showed peaks corresponding to metallic nickel phase 2θ: 44.5° (111) and 51.8° (002). For comparison, an unsupported NiMo-0.3 solid was prepared under similar conditions (i.e., 700 °C and 20% CH4/H2), and the XRD results corroborate the formation of β-Mo2C (Fig. S1), which is consistent with the results reported by Jin et al. [44]. The formation of bimetallic Mo3Ni2C or NixMo1-x phases were not observed in the prepared catalysts, which may be due to the different preparation methods employed in this work. Other parameters can also influence the formation of β-Mo2C; for example, Liang et al. [40] prepared β-Mo2C/CNTs at a lower temperature, using a more complex procedure. Namely, a calcination in air at 500 °C was performed for 3 h, followed by reduction under H2 atmosphere at 700 °C for 4 h; however, it was found that β-Mo2C particles were mainly located outside of CNTs.In this work transmission electron microscopy was employed to identify the position of the particles, as well as their size and morphologic features. Representative micrographs are presented in Fig. 2 for NiMo-0.3/CNTs. STEM results (Fig. 2a-c) indicate that most of the particles are inside of CNTs, as expected according to the methodology for the preparation of the catalysts. Atomic quantification of one of these particles showed the presence of Ni of 43.9% and Mo 56.1% free of C and O (Fig. 2c). Fig. 2d-f show TEM images with particles inside and some outside of CNTs, the particles observed outside of CNTs present an interplanar distance of 2.13 Å (Fig. 2g), which can be assigned to the (002) plane of MoC phase. The particle inside of the CNTs with interplanar distance of 2.36 Å (Fig. 2h) corresponding to the (002) plane of β-Mo2C phase. The diameter of CNTs limited the diameter of formed Ni-Mo2C particles; as is shown in Fig. 2i, most particles are less than 12 nm, and most of particles with an approximate size of 5 nm contrast with the particle size of those present outside the CNTs, measuring about 28 nm.The effect of reaction temperature on HDO performance over the catalyst NiMo-0.3/CNTs has also been investigated. As shown in Fig. 3 , BF conversion was 4.7% at 200 °C. The conversion increased to ~100% at temperatures higher than 280 °C, where hydrogenated products (i.e., 2,3-DHBF and OHBF) were predominant up to 280 °C (yield up to 64.2%). The yield to hydrogenated products (2,3-DHBF, ECHOH) decreased up to 6.2% at 320 °C, and the hydrodeoxygenated products EB ECH, and MCH increase approximately to 3.9%, 74.4% and 15.6%, respectively (93.8% of deoxygenated products at 320 °C), which indicated that the hydrodeoxygenation selectivity was improved with the increase in reaction temperature. Other products as ethylphenol (2-EtPh) were detected at all temperatures, which should be formed by the rupture of the C–O bond of 2,3-DHBF before the ring hydrogenation; however, its yield was low throughout the temperature range in comparison with OHBF.The present results from HDO of benzofuran over NiMo-x/CNTs catalysts are consistent with the reaction network proposed in the literature [45,46], where the initial step is the hydrogenation of BF to 2,3-DHBF. As expected, higher temperatures (320 °C) caused an increase in BF conversion and led to higher yield to HDO products, particularly saturated products, such as ECH and MCH.The influence of reaction time on deoxygenation of benzofuran was studied and the results are shown in Fig. 4 . Once the reactor reached the reaction conditions (i.e., 300 °C) the measured conversion was 17.5%, with hydrogenation being the main reaction taking place by forming exclusively 2,3-DHBF; after one hour the conversion reached 100% and the consecutive hydrogenation reactions occurred with 2,3-DHBF and OHBF as the main products. Prolonged reaction time favored hydrodeoxygenation reactions and then at 6 h of reaction the total deoxygenated product yield was 69.6%, with ECH (56%) as the main product, indicating that incorporation of both Ni and Mo could not only increase the HDO conversion over NiMo-x/CNTs catalysts but also promote the desired HDO reaction over NiMo-0.3/CNTs.The detected deoxygenated products yield over NiMo-0.3/CNTs catalyst after 6 h decreased in the order: ECH > MCH > EB > ECHE. In addition, the yield toward O-containing intermediates decreased in the order: OHBF >2,3-DHBF > CHEOH. The presence of OHBF indicated that the NiMo-0.3/CNTs catalyst had a high hydrogenation properties, even higher than that reported over Ni2P/Al-SBA-15 catalyst where 2,3-DHBF is transformed into 2-EtPh [47].Based on the results concerning the effect of temperature and the formed products with time, the plausible reaction route is shown in Fig. 5 . The first step involves hydrogenation of the furan ring, which led to the formation of 2,3-DHBF, which is further converted into OHBF by hydrogenation of the benzene ring. Then, ECHOH is obtained from OHBF by a C–O bond cleavage of the heterocyclic ring through hydrogenolysis. Finally, ECHOH is transformed into ECH by dehydration, followed by demethylation of ECH to yield MCH.The catalyst follows mainly the route BF → 2,3-DHBF→OHBF→ECHOH→ECH, which is different of that reported for Nd-Ni2P or Y-Ni2P [46], Ni-Cu/γ-Al2O3 [48], Ni2P/Al-SBA-15 [7] catalysts, where OHBF is not detected and the reaction follow the pathways: BF → 2,3-DHBF→2-EtPh →EB → ECH. The present results suggest that the catalyst NiMo-0.3/CNTs have higher hydrogenation properties, which are similar to those reported for reduced Mo and Ni–Mo/ γ-Al2O3 [49] and silica-alumina-supported Pt, Pd, and Pt − Pd catalysts [50,51], where OHBF is detected. Fig. 6 shows the catalyst recycling of NiMo-0.3/CNTs in the BF hydrodeoxygenation. After each reaction, the catalyst was washed with dichloromethane, dried at 100 °C overnight, and reused for the next reaction. The conversion decreased from 97% in the first cycle to 51% in the fifth cycle. In general, the yield decreased in the order 2,3-DHBF>2-EtPh>OHBF>ECH > MCH. Upon reuse, the amount of 2-EtPh was higher with respect to OHBF, which suggests that the hydrogenating properties of the catalyst decreased.The surface characterization of NiMo-0.3/CNTs catalyst was further probed by XPS. From the survey spectrum displayed in Fig. S2, elements of C, O, Ni, and Mo can be identified. Fig. 7 shows that the high-resolution spectra of Mo 3d can be deconvoluted into three doublets for Mo6+ (89.6%), Mo4+(5.8%) and Mo2+(4.6%). The peaks at binding energies 233.0 eV and 236.1 eV (Mo 3d5/2 and 3d3/2) for Mo6+ and 230.9 eV and 232.6 eV (Mo 3d5/2 and 3d3/2) for Mo4+, in molybdenum oxidized phases [11,52,53], which was probably caused by surface oxidation of Mo2C due to air contact during the passivation process. The small peaks centered at 228.7 eV and 231.9 eV (Mo 3d5/2 and 3d3/2) were assigned to Mo2+, which can be attributed to the carbide phase, indicating the presence of Mo2C [54]. Additionally, a small Ni 2p XPS signal was detected (Fig. 7d), which can be assigned to Ni in Ni-MoxC particles. The deconvolution of the C 1 s peak is shown in Fig. 7b, peaks at 284.8 eV, 286.3 eV, and 291.1 eV can be assigned to C=C/C–C/C–H, C − OH/C–O–C and O–C=O bonds respectively [55]. Additionally, the oxygen region (Fig. 7c) shows peaks at 530.5 eV for M–O and 532.1 eV for C-O bonds.It should be noted that the thickness of more than 10 nm for CNTs makes it difficult to observe the actual surface of the Ni-Mo2C particles inside of CNTs. However, this analysis allows us to observe the accumulation of organic material, likely coke precursor on the CNTs, as has been reported for Ni-Mo based catalysts [56]. After five reaction cycles, the XPS signals change significantly (Fig. 7e-h). The most significant change is observed for C 1 s and O 1 s signals, the increase in 286.0 eV and 532.7 eV and 534.1 eV signals indicate the presence of bonds C–OH, C–O–C, and C=O. The increased formation of oxidized species (C–O, C=O, C–OH) changes from 6.49% to 58.61%, which could be due to the presence of hydrocarbon residues or coke precursors deposited on the catalyst surface during the reaction [57]. Several studies reported the formation of coke and the deactivation of the catalysts [58–60], however further experiments are required to determine the true reason for the deactivation of NiMo/CNTs catalysts. According to high-resolution XPS, Mo6+ decreased after five reaction cycles from 89.6% to 52.7%, whereas the signal for Mo carbide increased from 4.64% to 9.70%.The results showed that the proposed methodology improved the deposition of the Ni-Mo2C particles inside of CNTs. Ni/CNTs plays higher activity than Mo/CNTs, however Ni improves the formation β-Mo2C, possibly by the C–H bond dissociation of CH4, forming C species during the activation process. All Ni- Mo2C/CNTs catalysts showed higher conversion and yield toward hydrogenation and hydrodeoxygenation products than mono-metallic Ni/CNTs and Mo/CNTs. Nickel-modified molybdenum carbide particles inside CNTs with Ni/Mo molar ratio of 0.3 showed the highest catalytic performance, much better than unsupported NiMo-0.3 catalyst. The results suggest that carbon nanotubes limit the size of the formed Ni-Mo2C particles inside CNTs, exposing a higher number of active sites for the reaction. The NiMo-x/CNTs catalysts showed high hydrogenation properties, and the main pathway for benzofuran HDO was BF → 2,3-DHBF → OHBF → ECHOH → ECH → MCE. This reaction route is favored at high temperatures (320 °C), where 93.8% yield to deoxygenated products were obtained after 4 h of reaction, with ECH (74% of yield) as the main product.Nevertheless, it must be mentioned that a progressive lost in activity was observed upon catalyst reusing for 5 cycles (after solvent washing). The decrease of the catalyst activity is likely due to the accumulation of oxygenated species, possibly coke precursors on the catalyst surface; therefore, further work is required to improve stability for their reuse.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to thank the financial support from the National Key R&D Program of China (2018YFB1501403), and the National Natural Science Foundation of China (22078220 and 51776134). S.P would like to express his gratitude to the 2016 China-LAC Young Scientist Exchange Program for the financial support to perform studies at the Key Laboratory of Coal Science and Technology (Taiyuan University of Technology). D.L and A.M thank the Universidad de Antioquia UdeA (Colombia). Supplementary material Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuproc.2022.107416.

This work proposed a controllable synthesis of Ni-Mo catalyst supported inside multi-wall carbon nanotubes (CNTs). The results indicate that Ni improved the β-Mo2C formation and markedly promoted the benzofuran (BF) hydrogenation and hydrodeoxygenation activity of the catalysts. The synergistic interaction between Ni and Mo reached the maximum at a Ni/Mo molar ratio of 0.3, which could be favored by the proximity between the Ni and β-Mo2C particles inside the CNTs reaching a 99.5% of BF conversion to hydrogenated and deoxygenated products as 2,3-dihydrobenzofuran, octahydrobenzofuran, and ethylcyclohexane; in contrast, BF conversion on unsupported Ni-Mo2C-0.3 was only 0.7%. Deoxygenated products are favored under different conditions, such as the time, and mainly with the temperature achieving 93.8% of yield toward deoxygenated products with 100% of BF conversion at 320 °C. However, the catalyst activity is lost through reuse cycles, likely due to the deposition of high molecular weight compounds (coke) on the catalyst surface.

Data will be made available on request. Data will be made available on request.The classical Hirao reaction involves the P–C coupling between vinyl- or arylbromides and dialkyl phosphites performed in the presence of palladium-tetrakistriphenylphosphine (Pd(PPh3)4) as the catalyst [1,2]. Shortly after the invention of this elegant method for the synthesis of vinyl- or arylphosphonates, Pd- or Ni-salts (e.g. Pd(OAc)2 or NiCl2, respectively) were applied with different mono- and diphosphines as the P-ligands, and the reaction was extended to different aryl- and heteroaryl derivatives, and alkyl H-phosphinates along with secondary phosphine oxides [3–10]. Our research group developed a microwave (MW)-promoted method, in which Pd(OAc)2 or NiCl2 was the catalyst precursor, and the excess of the >P(O)H reagent provided the P-ligand via its trivalent tautomeric form (>P-OH) [11,12]. The fine mechanism was explored experimentally, and by high level quantum chemical calculations [13,14]. As regards the Ni-catalyzed P–C coupling reactions, surprisingly a Ni(II) → Ni(IV) transition was found instead of the generally believed Ni(0) → Ni(II) formula [15–17].The application of Cu(I) salts, this occasion, together with N-ligands is another option. A series of N-containing compounds, such as DMEDA [18–21], 2,2′-bipyridine, 1,10-phenanthroline, TMEDA, 1-methyl-1H-imidazole [22], N-methylpyrrolidine-2-carboxamide [23], proline and pipecolic acid [24], (S)-α-phenylethylamine [25,26], or 1-pyrrodinylphosphonic acid monophenyl ester [27] were described. Due to the lower activity of the Cu-catalysts, in most cases iodoarenes were the starting materials in reaction with P-reagents. It was also a possibility that aryl bromides were prereacted with potassium iodide [24,27]. Interestingly, “ligand-free” Cu-catalyzed protocols were also proposed [28,29]. We were the first, who investigated the mechanism of the Cu(I)-salt catalyzed P–C couplings. Moreover, in this study no conventional ligands were added, only a diarylphosphine oxide and triethylamine was present as the reactant and as the base, respectively [30].In this article, we overview our new results including the use of Cu(II)-salts as catalyst precursors. Cu(II) salts have a few advantages against Cu(I), such as better solubility, chemical stability, and a lower commercial price.As it was shown, a number of methods were described for the Hirao reaction. We wished to find the best P–C coupling methods for the synthesis of diethyl phenylphosphonate ((EtO)2PhP(O) (1a)) and triphenylphosphine oxide (Ph3P(O) (1b)) as the product of the simplest model reactions. However, the syntheses were carried out in different laboratories and by different hands, not speaking about the MW reactors, whose activity may have not been the same, as it decreases with aging. For this, we decided to reproduce the relevant experiments in a single MW reactor in our laboratory.First of all, the model reaction of bromobenzene (PhBr) with diethyl phosphite ((EtO)2P(O)H) was investigated under different conditions. In all cases triethylamine (NEt3) was used as the base. We were not successful in reproducing the original method using Pd(PPh3)4 as the catalyst at 90 °C for 2.5 h in the absence of any solvent. Phosphonate 1a was obtained in a yield of 55% that was in contrast with the outcome of 92% reported earlier [31] (Table 1 /entry 1). On the one hand, this catalyst is a sensitive species, we could identify Ph3P(O) as a by-product. The reproduction of the experiment using Pd(OAc)2/PPh3 in ethanol (EtOH) at 80 °C for 16 h was more successful. Our yield of 89% was not far from the reported 94% [32] (Table 1/entry 2). Here again, Ph3P(O) was a contaminant. In both previous cases, the Ph3P(O) may have come from the PPh3 ligand. The next case studied was the MW-assisted protocol involving Pd(OAc)2 as the catalyst precursor and some excess of the P-reagent applied as the ligand. Using EtOH at 120 °C for 0.5 h, the 71% yield could be somewhat exceeded [33] (Table 1/entry 3). However, when acetonitrile (MeCN) was the solvent, there was need for a longer reaction time of 1 h instead of 0.5 h. In this way, the 61% outcome could be exceeded by a better yield of 83% [13] (Table 1/entries 4 and 5). In the third, solvent-free variation at 150 °C, the 93% yield could not be reproduced applying a 5 min’ irradiation time, but prolonging the time of exposure, after 0.5 h, we could reach an outcome of 80% [34] (Table 1/ entries 6 and 7).As regards the P–C coupling of PhBr and diphenylphosphine oxide (Ph2P(O)H), using the previous approach comprising EtOH as the solvent at 120 °C, the 83% yield of Ph3P(O) (1b) could be reproduced [14] (Table 1/entry 8). At the same time, performing the synthesis in MeCN at 150 °C, there was need for prolonged irradiation of 1 h (instead of 0.5 h) to achieve a better yield of 75%, as compared to the reported 67% [13] (Table 1/entries 9 and 10).In summary, the best general method for the P–C coupling of PhBr with (EtO)2P(O)H and Ph2P(O)H is the MW-promoted accomplishment involving 5–10% of Pd(OAc)2 as the catalyst precursor and 1.15–1.30 equiv. of the >P(O)H reagent as the P-ligand in EtOH using NEt3 at 120 °C.As regards the use of NiCl2 as the catalyst precursor, the coupling of PhBr and (EtO)2P(O)H led to similar results, no matter if it was performed using K2CO3 in MeCN, or NEt3 without any solvent. After an irradiation at 150 °C for 1 h and 2 h, respectively, the outcomes were similar. The literature procedure reported a yield of 70% and 67%, respectively, while our own reproductions provided yields of 62% and 58%, respectively (Table 2 /entries 1 and 2).Reproduction of the reaction of Ph2P(O)H with PhBr at 150 °C applying Cs2CO3 or K2CO3 in MeCN gave closer results. Our yields were 74% and 86%, as compared to the literature values of 79% and 91%, respectively (Table 2/entries 3 and 4).One may conclude that, in general, the Pd-catalyzed P–C couplings are somewhat more efficient than the Ni-promoted ones.After clarifying the best variations with the Pd(II)- and Ni(II)-salts, we wished to test the applicability of the cheaper and more practical Cu(II)-salts in the P–C coupling reactions of Ph2P(O)H with halobenzene. In these cases, PhBr was not expected to reveal a suitable reactivity, for this, iodobenzene (PhI) was applied [30]. Using 20% of anhydrous CuSO4, 1 equiv. of Ph2P(O)H and the same amount of NEt3 at 165 °C in EtOH, the conversion was not complete, and the proportion of Ph3P(O) (1b) was only 57% (Table 3 /entry 1). Increasing the quantity of Ph2P(O)H to 1.4 equiv., almost a similar result was obtained (Table 3/entry 2). At the same time, when 1 equiv. of Ph2P(O)H was applied together with 2 equiv. of NEt3, the formation of the phosphine oxide (1b) was clear-cut, having manifested in a quantitative conversion, and in a preparative yield of 85% (Table 3/entry 3). Changing for Cu(OAc)2 · H2O as the precursor, the trend remained the same, and the overall yields were somewhat higher (50/59%, Table 3/entries 4 and 5), or slightly lower (75%, Table 3/entry 6). One could conclude that applying CuSO4 or Cu(OAc)2 · H2O, NEt3 must have a role in the ligation of the center Cu.Applying CuCl2 · 2H2O and CuSO4 · 5H2O as the metal salt component of the catalyst, and Ph2P(O)H and NEt3 in a 1:2 ratio, the useful conversions and yields were lower (78/72% and 48/56%, respectively, Table 3/entries 7 and 8) than in the previous cases.Then the P–C coupling reaction was extended to diarylphosphine oxides. Applying bis(4-methylphenyl)phosphine oxide and either CuSO4 or Cu(OAc)2 · H2O as the catalyst precursor, there was a smaller difference between the experiments using Ar2P(O)H and NEt3 in a ratio of 1:1 or 1:2. However, in all cases the latter ratio was more favorable (Table 4 /entries 1–4). The coupling with bis(3,5-dimethylphenyl)phosphine oxide was also quantitative with two equivalents of NEt3, no matter if CuSO4 or Cu(OAc)2 · H2O was the precursor (Table 4/entries 5 and 6).According to the experimental findings, catalytic amount of a Cu(II) salt can effectively promote the Hirao reaction of iodobenzenes with Ph2P(O)H in protic solvents. Earlier, a plausible reaction mechanism was explored for the Cu(I)-catalyzed Hirao reactions, where the metal is complexed by both the P-reactant and the excess of the NEt3 base [30].An analogous reaction mechanism was assumed using Cu(II) (Fig 1 ). Hypothetically, the starting species of the reaction sequence is the complexed form of Cu(II), which may be easily deprotonated by the base (NEt3) present in the reaction mixture. This is followed by the P-Cu-P → P-Cu-O isomerization. Analogously to the Cu(I) mechanism, a PhBr molecule (selected instead of PhI to simplify the calculations) was complexed via the pi-system of its aromatic ring. In the next step, the C–Br bond would be cleaved, resulting in a covalent Cu–C bond, meanwhile the Cu2+ would be oxidized formally to Cu4+, leading to intermediate-II. However, according to a careful theoretical investigation, this step did not prove to be feasible, as there was no real TS on the potential energy surface, which could link the two sides, intermediate-I and intermediate-II with a continuous pathway. It was confirmed by a stepwise bond scanning and IRC investigation. Due to the strong electron deficient central Cu(II) ion, intermediate-II represents a rather distorted structure. A similar situation was observed in an earlier study for the Ni2+-catalyzed Hirao reaction [16]. Formally, the “remaining part” of the mechanism could lead to the desired product. In this case, the second TS represents a rather low, almost negligible (ca. 5 kJ mol–1) enthalpy barrier, so the C–C bond formation could happen.The invalid catalytic cycle is shown in Fig 2 .In order to explain the failure of the oxidative addition with Cu(II), the mechanism assumed should be compared with that computed for Cu(I) earlier [30]. The process may be simplified to two subprocesses described by the change in the oxidation state of Cu. In the course of the oxidative addition step, the oxidation number of Cu is increased by two. In the case of Cu(I), the oxidation leads formally to Cu(III) that step is connected with the reduction of the α carbon atom of the phenyl group affording Ph–. The activation enthalpy of this step was found quite low (+45.7 kJ mol–1). In contrast to this, the analogous oxidation of Cu(II) to Cu(IV) is not feasible, as the formal negative charge in the phenyl ring of the Cu(IV)–Ph complex would be greater than in the previous Cu(I) complex. As a consequence of this, Cu(II) cannot be oxidized to Cu(IV). In other words, in the case under discussion, there is no continuous pathway leading to intermediate-II.The reductive elimination (that assumes +88.0 kJ mol–1 for the Cu(III) → Cu(I) case) would practically have no barrier for the Cu(IV) → Cu(II) process. See Table 5 and Fig 3 .In search for a valid mechanism with the Cu(II) precursor, we looked for an alternative explanation. Here, we propose a process, where Cu(II) is reduced to Cu(I) by the secondary phosphine oxide added to the reaction in a slight excess, meanwhile the P(III) atom is oxidized to P(V). A plausible red-ox reaction is suggested in Scheme 1 , where Cu(II) is transformed to Cu(I) in the presence of NEt3 as the base and in ethanol as the solvent. The reaction enthalpy is exothermic exhibiting 811 kJ mol–1 meaning a reasonable driving force.Once the Cu(I) ion was formed, the reaction sequence may proceed further as it was proposed earlier [30]. It is assumed that the in situ formed Cu(I) species may be advantageous, as the on site formation provides a better solubility, and consequently a faster rate.The reactions were carried out in a CEM® Discover (300 W) focused microwave reactor equipped with a stirrer and a pressure controller using 80–100 W irradiation under isothermal conditions. The reaction mixtures were irradiated in sealed borosilicate glass vessels (with a volume of 10 mL) available from the supplier of CEM®. The reaction temperature was monitored by an external IR sensor.The 31P, 13C and 1H NMR spectra were taken in CDCl3 solution on a Bruker AV-300 spectrometer operating at 121.5, 75.5 and 300 MHz, respectively. The 31P chemical shifts are referred to H3PO4, while the 13C and 1H chemical shifts are referred to TMS. The couplings are given in Hz. The exact mass measurements were performed using an Agilent 6545 Q-TOF mass spectrometer in high resolution, positive electrospray mode.To 0.047 mmol (0.011 g) or 0.094 mmol (0.021 g) of the Pd(OAc)2 catalyst in 1 mL of ethanol or acetonitrile or without any solvent were added 0.95 mmol (0.10 mL) of bromobenzene, 1.09 mmol (0.14 mL) or 1.42 mmol (0.18 mL) or 1.23 mmol (0.15 mL) of diethyl phosphite and 1.04 mmol (0.15 mL) of triethylamine. Then, the mixture was irradiated in a closed vial in the microwave reactor at 120 °C or 150 °C for the times (0.5 or 1 h) shown in Table 1. The reaction mixture was diluted with 3 mL of the corresponding solvent or EtOH, filtrated, and the residue obtained after evaporation was passed through a thin (2–3 cm) layer of silica gel using ethyl acetate as the eluent. The crude mixture was analyzed by 31P NMR spectroscopy, then it was purified by column chromatography (silica gel and ethyl acetate as the eluent). For the results see Table 1/entries 3, 5 and 7.To 0.022 mmol (0.0049 g) of the Pd(OAc)2 in 1 mL of ethanol were added 0.43 mmol (0.045 mL) of bromobenzene, 0.49 mmol (0.10 g) of diphenylphosphine oxide and 0.47 mmol (0.066 mL) of triethylamine. Then, the mixture was irradiated in a closed vial in the microwave reactor at 120 °C for 1 h. The reaction mixture was diluted with 3 mL of EtOH, filtrated, and the residue obtained after evaporation was passed through a thin (2–3 cm) layer of silica gel using ethyl acetate as the eluent. The crude mixture was analyzed by 31P NMR spectroscopy, then it was purified by column chromatography (silica gel and dichloromethane-methanol 97:3 as the eluent) to provide 84% (0.11 g) of phosphine oxide 1b.To 0.049 mmol (0.0064 g) of the NiCl2 in 1 mL of acetonitrile were added 0.49 mmol (0.055 mL) of bromobenzene, 0.64 mmol of >P(O)H-reagent [diethyl phosphite: 0.082 mL or diphenylphosphine oxide: 0.13 g] and 0.49 mmol (0.068 g) or 0.59 mmol (0.082 g) of potassium carbonate. Then, the mixture was irradiated in a closed vial in the microwave reactor at 150 °C for 60 or 45 min. The reaction mixture was diluted with 3 mL of MeCN, filtrated, and the residue obtained after evaporation was passed through a thin (2–3 cm) layer of silica gel using dichloromethane-methanol 97:3 as the eluent. The crude mixture was analyzed by 31P NMR spectroscopy, then it was purified by column chromatography (silica gel and ethyl acetate or dichloromethane-methanol 97:3 as the eluent) to give 62% (0.065 g) of phosphonate 1a and 86% (0.12 g) of phosphine oxide 1b, respectively.To 0.099 mmol of the catalyst (CuSO4: 0.016 g, Cu(OAc)2 · H2O: 0.018 g) in 1 mL of ethanol were added 0.49 mmol (0.055 mL) of iodobenzene, 0.49 mmol (0.10 g) of diphenylphosphine oxide and 0.99 mmol (0.14 mL) of triethylamine. Then, the mixture was irradiated in a closed vial in the microwave reactor at 165 °C for 3 h. The reaction mixture was diluted with 3 mL of EtOH, filtrated, and the residue obtained after evaporation was passed through a thin (2–3 cm) layer of silica gel using dichloromethane-methanol 97:3 as the eluent. The crude mixture was analyzed by 31P NMR spectroscopy, then it was purified by column chromatography (silica gel and dichloromethane-methanol 97:3 as the eluent) to furnish phosphine oxide 1b in a yield of 85% (0.11 g) and 75% (0.10 g), respectively.To 0.087 mmol of the catalyst (CuSO4: 0.0138 g, Cu(OAc)2 · H2O: 0.016 g) in 1 mL of ethanol were added 0.048 mL (0.43 mmol) of iodobenzene, 0.43 mmol of diarylphosphine oxide [bis(4-methylphenyl)phosphine oxide: 0.10 g or bis(3,5-dimethylphenyl)phosphine oxide: 0.11 g] and 0.87 mmol (0.12 mL) of triethylamine. Then, the mixture was irradiated in a closed vial in the microwave reactor at 165 °C for 3 h. The reaction mixtures were diluted with 3 mL of EtOH, filtrated, and the residue obtained after evaporation was passed through a thin (2–3 cm) layer of silica gel using dichloromethane-methanol 97:3 as the eluent. The crude product was analyzed by 31P NMR spectroscopy, then it was purified by column chromatography (silica gel, and dichloromethane-methanol 97:3 as the eluent). For the results see Table 4/entries 2, 4, 5 and 6.Appearance: colorless oil; 31P NMR (CDCl3, 121.5 MHz) δ 18.8, δP [12] (CDCl3, 121.5 MHz) 19.7, δP [24] (CDCl3, 121 MHz) 19.4; 13C NMR (CDCl3, 75.5 MHz) δ 16.2 (d, J = 6.5 Hz, CH3), 62.0 (d, J = 5.4 Hz, CH2), 128.3 (d, J = 187.9 Hz, C1), 128.4 (d, J = 15.0 Hz, C2)a, 131.7 (d, J = 9.9 Hz, C3)a, 132.3 (d, J = 3.0 Hz, C4), amay be reversed, δC [35] (CDCl3, 75.5 MHz) 16.3 (d, J = 6.5 Hz), 62.0 (d, J = 5.3 Hz), 128.3 (d, J = 187.8 Hz), 128.4 (d, J = 14.9 Hz), 131.7 (d, J = 9.8 Hz), 132.3 (d, J = 3.1 Hz); 1H NMR (CDCl3, 300 MHz) δ 1.29 (t, 6H, J = 7.1 Hz, CH3) 3.97–4.21 (m, 4H, OCH2), 7.37–7.56 (m, 3H, ArH), 7.71–7.84 (m, 2H, ArH), δH [35] (CDCl3, 300 MHz) δ 1.40 (t, 6H, J = 7.1 Hz), 4.10 (m, 4H), 7.42–7.48 (m, 2H), 7.54 (m, 1H), 7.77–7.84 (m, 2H), δH [24] (CDCl3, 300 MHz) δ 1.32 (t, 6H, J = 6.87 Hz), 4.10–4.14 (m, 4H), 7.42–7.48 (m, 3H), 7.81–7.85 (q, 2H); [M + H]+ = 215.0838 C10H16O3P requires 215.0837.Appearance: white crystals, mp 156–157 °C, mp [36] 156.6–157.4 °C; 31P NMR (CDCl3, 121.5 MHz) δ 29.1, δP [36] (CDCl3, 162 MHz) 29.5, δP [12] (CDCl3, 121.5 MHz) 30.3; 13C NMR (CDCl3, 75.5 MHz) δ 128.6 (d, J = 12.1 Hz, C2)a, 132.0 (d, J = 2.8 Hz, C4), 132.2 (d, J = 9.9 Hz, C3)a, 132.7 (d, J = 103.8 Hz, C1), amay be reversed, δC [36] (CDCl3, 100 MHz) 128.4 (d, J = 12.1 Hz), 131.9 (d, J = 2.2 Hz), 132.5 (d, J = 9.9 Hz), 132.8 (d, J = 104.6 Hz); 1H NMR (CDCl3, 300 MHz) δ 7.38–7.48 (m, 6H, ArH), 7.48–7.56 (m, 3H, ArH), 7.59–7.72 (m, 6H, ArH), δH [36] (CDCl3, 400 MHz) δ 7.43–7.48 (m, 6H), 7.52–7.56 (m, 3H), 7.64–7.70 (m, 6H); [M+H]+ = 279.0934 C18H16OP requires 279.0939.Appearance: white crystals, 31P NMR (CDCl3, 121.5 MHz) δ 27.8, δP [12] (CDCl3, 162 MHz) 29.4, δP [37] (CDCl3, 162 MHz) 30.5; 13C NMR (CDCl3, 75.5 MHz) δ 21.6 (CH3), 128.5 (d, J = 12.1 Hz, C2′)a, 129.3 (d, J = 12.5 Hz, C2)b, 129.4 (d, J = 106.6 Hz, C1), 131.8 (d, J = 2.7 Hz, C4′), 132.1 (d, J = 9.8 Hz, C3′)a, 132.1 (d, J = 10.3 Hz, C3)b, 133.1 (d, J = 104.1 Hz, C1′), 142.4 (d, J = 2.8 Hz, C4′), a,bmay be reversed, δC [37] (CDCl3, 100 MHz) 21.7, 128.6 (d, J = 11.8 Hz), 129.4 (d, J = 12.6 Hz), 129.4 (d, J = 106.9 Hz), 131.9 (d, J = 3.2 Hz), 132.0 (d, J = 8.7 Hz), 132.2 (d, J = 10.2 Hz), 133.0 (d, J = 102.5 Hz), 142.6 (d, J = 2.9 Hz); 1H NMR (CDCl3, 300 MHz): δ 2.39 (s, 6H, CH3), 7.18–7.32 (m, 4H, ArH), 7.37–7.47 (m, 2H, ArH), 7.47–7.61 (m, 5H, ArH), 7.61–7.73 (m, 2H, ArH); δH [37] (CDCl3, 400 MHz) 2.38 (s, 6H), 7.24 (dd, J = 8.4 Hz, 2.4 Hz, 4H), 7.48 (m, 1H), 7.53 (dd, J = 11.8 Hz, 8.0 Hz, 4H), 7.62–7.68 (m, 2H);, [M+H]+ = 307.1252 C20H19OP requires 307.1252.Appearance: white crystals, 31P NMR (CDCl3, 121.5 MHz) δ 29.6, δP [37] (CDCl3, 162 MHz) 30.9, 13C NMR (CDCl3, 300 MHz) δ 21.4 (CH3), 128.4 (d, J = 12.0 Hz, C2′)a, 129.7 (d, J = 9.8 Hz, C2), 131.7 (C4′), 132.1 (d, J = 9.9 Hz, C3′)a, 132.4 (d, J = 105.3 Hz, C1), 133.1 (d, J = 103.1 Hz, C1′), 133.7 (d, J = 2.8 Hz, C4), 138.1 (d, J = 12.7 Hz, C3), amay be reversed, δC [37] (CDCl3, 100 MHz) 21.56, 128.6 (d, J = 11.7 Hz), 129.8 (d, J = 10.0 Hz), 131.9 (d, J = 2.2 Hz), 132.3 (d, J = 9.7 Hz), 132.4 (d, J = 102.6 Hz), 133.1 (d, J = 102.7 Hz), 133.9 (d, J = 2.3 Hz), 138.3 (d, J = 12.2 Hz); 1H NMR (CDCl3, 300 MHz) 2.31 (s, 12H, CH3), 7.15 (s, 2H, ArH), 7.28 (d, J = 12.2 Hz, 4H, ArH), 7.39–7.55 (m, 3H, ArH), 7.62–7.73 (m, 2H, ArH), δH [37] (CDCl3, 400 MHz) 2.31 (s, 12H), 7.15 (s, 2H), 7.26 (d, J = 12.4 Hz, 4H), 7.42–7.47 (m, 2H), 7.51–7.55 (m, 1H), 7.63–7.68 (m, 2H); [M+H]+ = 335.1566 C22H23OP requires 335.1565.All computations were carried out with the Gaussian16 program package (G16) [38], using standard convergence criteria for the gradients of the root mean square (RMS) Force, Maximum Force, RMS displacement and maximum displacement vectors (3.0 × 10–4, 4.5 × 10–4, 1.2 × 10–3 and 1.8 × 10–3). Computations were carried out at M06–2X level of theory [39]. The basis set of 6–31G(d,p) was applied for C, H, O, P, N, Cl, Br, Cu and SDD/MWB46 for iodine [40]. The vibrational frequencies were computed at the same levels of theory, in order to confirm properly all structures as residing at minima on their potential energy hypersurfaces (PESs). Thermodynamic functions U, H, G and S were computed at 398.15 K. Beside the vacuum calculations, the IEFPCM method was also applied to model the solvent effect, by using the default settings of G16, setting the ε = 24.852 [41]. See the Supporting Information for details.While the Pd- and Ni-catalyzed Hirao reactions have been studied in detail, Cu-catalysis is a somewhat neglected field. After a short summary of the Pd- and Ni-promoted P–C couplings, the Cu(II)-salt (CuSO4 and Cu(OAc)2 · H2O) catalyzed reaction of iodobenzene with diethyl phosphite and secondary phosphine oxides (diarylphosphine oxides) was investigated under microwave irradiation. The P–C couplings were the most efficient, when the >P(O)H reagent and triethylamine used as a base were applied in a 1:2 molar ratio suggesting that the amine is also involved in the intermediate complexes as a ligand. A detailed study on the mechanism revealed that, as a matter of fact, the in situ formed Cu(I) species was the primary species in the oxidative addition step towards the P–C couplings.The whole research was supervised and directed by GK. The manuscript was written by GK, ZM and BH. The synthetic work was performed by BH and RSz. The theoretical calculations were carried out by ZM. The fund was collected by GK. All authors have approved the final version of the manuscript.NMR spetra of the products prepared and details of the quantum chemical calculations.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This project was supported by the National Research, Development and Innovation Office (K134318) (G.K.) and Bolyai Research Scholarship (BO/799/21/7) (Z.M).Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jorganchem.2022.122526. Image, application 1

After surveying the most important results described on the Pd(0)/Pd(II)- and Ni(II)-catalyzed P–C couplings of bromobenzene and diethyl phosphite or diphenylphosphine oxide, the simple Hirao reactions catalyzed by different Cu(II) salts were investigated. In these instances, the more reactive iodobenzene had to be used. CuSO4 and Cu(OAc)2 · H2O were found suitable catalyst precursors in case they were used together with two equiv. of triethylamine suggesting the role of the amine in the complexation. Theoretical calculations suggested that the corresponding Cu(I) species formed from the initial Cu(II) salt is involved in the catalytic cycle.

Coal tar, as a main by-product in the process of coal pyrolysis, contains a large number of aromatic hydrocarbons (mainly bicyclic and tricyclic aromatic hydrocarbons) (Ardakani and Smith, 2011; Hodoshima et al., 2003; Liang et al., 2009; Zhao et al., 2010). The aromatics in low temperature coal tar can be converted into hydrogenated aromatics or cycloalkanes through catalytic hydrogenation, which can be used as the ideal component of jet fuel (Kim et al., 2017; Martin et al., 2020). Therefore, how to obtain high performance aromatic hydrogenation catalyst is the key to produce environmental-friendly fuels.At present, transition metal sulfide catalysts are the main commercial catalysts for aromatic hydrogenation. Although they possess certain anti-poisoning ability, its hydrogenation activity is relatively low (Hart et al., 2020; Jing et al., 2020; Ohta et al., 1999). Noble metal catalysts show excellent hydrogenation saturation ability, but the high cost and easily poisoned by heteroatomic compounds containing S and N have restricted their industrial application (Tao et al., 2013; Zeng et al., 2018). In recently researches, transition metal carbides, transition metal nitrides and transition metal phosphides show great potential in the field of aromatic hydrogenation (Dongil, 2019; Prats et al., 2019; Zhang et al., 2019). More importantly, transition metal phosphide catalysts not only possess high hydrogenation activity, but also exhibit anti-toxicity ability (Usman et al., 2015). Among all phosphide catalysts, Ni2P catalysts show high intrinsic hydrogenation activity due to its unique crystal morphology and electronic structure, which is expected to become a new generation of efficient aromatic hydrogenation catalyst (Oyama, 2003; Oyama and Lee, 2008; Zhao et al., 2020). However, the unsupported Ni2P catalysts have the problems of small specific surface area, poor dispersion of active phase, low mechanical strength and poor heat dissipation, which are unfavorable to its hydrogenation activity. Therefore, researchers usually load the Ni2P active components on various supports so as to expose more active sites and enhance hydrogenating activity.As to the supported Ni2P catalysts, the properties of different supports have an important effect on the dispersion of Ni2P particles (Oyama and Lee, 2008). In recent years, compared with the traditional mesoporous silica nanoparticles, wrinkle silica nanospheres (WSNs) has attracted rapid attention due to its unique three-dimensional central-radial pore channels (Polshettiwar et al., 2011; Thankamony et al., 2015). Up to now, WSNs has a wide range of applications, such as catalysis, biotherapy delivery, water treatment, dye-sensitized solar cells, supercapacitors, fluorescent probes, titanium dioxide capture, and biological imaging, photonics, composite materials, etc. Singh and Polshettiwar found that WSNs materials showed better textural stability, thermal stability and CO2 capture capacity than the traditional mesoporous MCM-41 due to its unique fibrous morphology (Singh and Polshettiwar, 2016).In order to improve the activity of the catalyst, scientific researchers focus on the introduction of organic complexing agents in the preparation of hydrogenation catalysts. Thomson R firstly introduced nitrilotriacetic acid (NTA) into NiMo/SiO2 catalyst during impregnation process (Thompson, 1986). The hydrogenation denitrification (HDN) activity of NTA modified NiMo/SiO2 catalyst was found to be 6 times higher than that of the traditional NiMo/SiO2 catalyst without complexing agents. After that, many complexing agents, such as citric acid (CA), ethylene glycol (EG), nitrilotriacetic acid (NTA), ethylenediamine tetraacetic acid (EDTA), cyclohexylenediamine tetraacetic acid (CYDTA) and ethylenediamine (EN), were widely applied to the preparation of hydrogenation catalysts (Ding et al., 2017; Garcia-Ortiz et al., 2020; Jiang et al., 2020; Li et al., 2021; Santolalla-Vargas et al., 2020). Furthermore, Oyama and his groups found that small sizes of Ni2P particles could expose more Ni(2) sites, thus enhancing the hydrogenation activity (Zhao et al., 2015; Shu et al., 2005). However, researches on the synthesis of small Ni2P particles over the supported catalysts, especially less than 5 nm, are still limited. Therefore, the goal of this research is to synthesize small sizes of Ni2P particles supported on the wrinkle silica nanoparticles.In this study, the series of Ni2P/WSNs catalysts were prepared by introducing chelators EDTA or NTA during the impregnation process. And the corresponding effects of chelators NTA and EDTA to Ni2P/WSNs catalysts were also studied. The reaction of naphthalene hydrogenation was used to evaluate the catalytic activity and stability of chelators modified Ni2P catalysts.The WSNs material was synthesized as mentioned in the previous papers (Hu et al., 2019). 10 g CTAB and 6 g urea were dissolved in 300 mL distilled water and 300 mL cyclohexane. After complete dissolution, 40 mL TEOS was added dropwise. Then the mixture was continuously stirred for 24 h under 70 °C. The WSNs products were obtained by centrifugation, desiccation and calcination at 550 °C for 6 h.The supported Ni2P/WSNs catalysts were prepared by impregnation process and temperature-programmed reduction (TPR). The oxidic Ni2P/WSNs catalyst precursors were obtained via a two-step incipient wetness method impregnated with aqueous solutions of nickel nitrate (Ni(NO3)2·6H2O) and ammonium hypophosphite (NH4H2PO4). After each impregnation, the catalyst was dried at 100 °C for 12 h and then calcined at 550 °C kept for another 3 h. The precursors were reduced at continuous H2 flow rate of 150 mL min−1 and temperature of 440 °C kept for 1 h, then rising to 550 °C.The EDTA modified Ni2PE/WSNs catalysts were prepared similar to the above process. Firstly, WSNs sample was impregnated with the aqueous mixed solutions of nickel nitrate (Ni(NO3)2·6H2O) and EDTA. After dried at 100 °C for 12 h and then calcined at 550 °C for 3 h, the oxidic catalyst precursors were obtained by impregnating with ammonium hypophosphite (NH4H2PO4). The molar ratios of Ni:P:EDTA were 1:1:0.5, 1:1:1 and 1:1:1.5, then the corresponding catalysts were named as Ni2PE(0.5)/WSNs, Ni2PE(1.0)/WSNs, and Ni2PE(1.5)/WSNs respectively. Finally, the Ni2PE/WSNs catalysts were collected by the same TPR progress.The NTA modified Ni2PN/WSNs catalysts were prepared similar to the above Ni2PE/WSNs process. The oxidic Ni2PN/WSNs catalyst precursors were obtained by a two-step incipient wetness method with the mixed solution of nickel nitrate (Ni(NO3)2·6H2O) and NTA, and ammonium hypophosphite (NH4H2PO4). After each impregnation, the catalyst was dried at 100 °C for 12 h and then calcined at 550 °C kept for another 3 h. The synthesized catalysts were named as Ni2PN(0.5)/WSNs, Ni2PN(1.0)/WSNs, and Ni2PN(1.5)/WSNs with different molar ratios of Ni:P:NTA (1:1:0.5, 1:1:1, 1:1:2). Finally, the Ni2PN/WSNs catalysts were collected by the same TPR progress.Wide-angle X-ray diffraction patterns were collected on a Japan Shimadzu X-6000 system (Cu Ka radiation, 40 kV, 30 mA, λ = 0.1540598 nm). X-ray photoelectron spectroscopy (XPS) analysis was conducted using a PerkinElmer PHI-1600 ESCA spectrometer. N2 adsorption-desorption experiments were performed at 77 K after degassing samples in flowing N2 at 350 °C for 4 h (using a Q Micromeritics Tristar 3020). The morphological features of the samples were characterized by scanning electron microscopy (SEM, Hitachi SU-8010, 5.0 kV). Transmission electron microscopy (TEM, Philips Tecnai G2 F20 S-TWIN, 300 kV). The size distributions of Ni2P phases were counted by more than 300 Ni2P particles from different images through high-resolution transmission electron microscopy. The average Ni2P particles were calculated according to Eq. (1). (1) D aver = ∑ i = 1 n d i n i where d i represents the size of each Ni2P particle and n i is the total number of Ni2P particles.The naphthalene hydrogenation was carried out with a fixed bed reactor. In general, 1.0 g of Ni2P supported catalysts were loaded in the middle of a stainless steel tube reactor with both ends filled with mesh quartz sands. Before the reaction, the synthesized oxidation precursor was reduced to the active nickel phosphide in H2 (4 MPa) at a flow rate of 150 mL min−1 by heating from room temperature at a heating rate of 5 °C min−1 to a temperature range as maintaining at 120 °C for 1 h, then rising up to 440 °C for 1 h and 550 °C for another 3 h, finally followed by cooling to room temperature in the continued H2 flow. The 5% naphthalene in cyclohexane was used as the model compound to access aromatic hydrogenation activity. The system was pressurized under the condition of 4 MPa, H2/Oil volumetric ratio of 500 (v/v), liquid hourly space velocity (WHSV) of 10 h−1. The catalysts were evaluated at 300–380 °C with an interval of 20 °C. The outlet stream was analyzed using gas chromatography-mass spectrometry (GC-MS).The conversion of naphthalene can be expressed by Eq. (2): (2) Naphthalene ( % ) = N f − N P N P where the N f is the mass fraction of naphthalene in the feedstock and N p is the mass fraction of naphthalene in the products.The peaks at 40.8o, 44.6o, 47.3o, 54.2o and 72.7o are tracked in Fig. 1 , which are attributed to the characteristic responses of Ni2P (PDF 3–953) (Tang et al., 2017). It is found that the peak intensities of Ni2P particles become gradually weak with the addition of chelating agents, which maybe be ascribed that Ni2P particles on the catalyst surface are too small to be detected by XRD characterization (Pullan et al., 2016; Yang et al., 2006).The XRD patterns of the spent catalysts are also characterized to test their stability. As shown in Fig. 1b, there is almost no change of the Ni2P characteristic peaks and no other unrelated peaks are detected in XRD, which confirms the high stability of the series Ni2P/WSNs catalysts after reaction.The N2 adsorption-desorption patterns of the synthesized catalysts after the addition of chelators are shown in Fig. 2 a. It can be seen that all catalysts showed type IV isotherms, indicating the good mesoporous structure. Fig. 2b shows the pore size distribution of the modified Ni2P/WSNs catalysts, and the results are summarized in Table 1 . The surface area and average pore size of the catalysts modified with chelators EDTA and NTA increase to some extent compared to the unmodified catalysts, which indicates that chelators can effectively inhibit the aggregation of Ni2P active phases.It can be seen in TEM images, all the catalysts still retain their original morphologies, which demonstrate that the addition of chelators has little damage to the morphology of the WSNs support. In addition, the morphologies of the spent catalysts after reaction are nearly the same as the fresh catalysts, as displayed in Fig. 3 and Fig. 4 , further confirming the high stabilities of the chelators modified Ni2P/WSNs catalysts.The average particle sizes of Ni2P can be statistically calculated from high resolution TEM images (Figs. S1 and S2), as summarized in Table 2 . It can be found that the average Ni2P particle sizes of Ni2PE/WSNs and Ni2PN/WSNs catalysts are both smaller than those of unmodified Ni2P/WSNs catalysts. Moreover, Ni2PE/WSNs catalysts modified by EDTA possess smaller sizes of Ni2P particles compared with Ni2PN/WSNs catalysts with the addition of NTA, confirming that the chelator EDTA plays a greater role in preventing Ni2P aggregation than the chelator NTA.It can be seen from Fig. 5 that H2-TPR patterns show several H2 consumption peaks in the vicinity of 400 °C and 700 °C. The H2 consumption peak near 400 °C is attributed to the reduction of NiO species (Louis et al., 1993), while the H2 consumption peak near 700 °C is derived from the P species due to the high P—O bond energy (Zuzaniuk and Prins, 2003).The reduction peak of Ni species is shifted to the higher temperature for the samples obtained with the addition of chelators, indicating that the interaction between Ni and support become stronger as the chelator EDTA or NTA being added. Meanwhile, the reduction temperature of P species over EDTA modified catalysts is shifted to slightly lower temperature, leading to the small reduction-peak interval of Ni and P species, thus promoting the formation of Ni2P active phase.XPS was used to investigate the differences in the chemical states of Ni and P on the surface of Ni2P supported catalysts. As can be seen from Fig. S3 and Table 3 , there are two XPS peaks of Ni 2p at around 853.1 eV and 856.9 eV, which are caused by the Niδ+ (0 < δ < 1) species in Ni2P and Ni2+ interacting with PO4 3−, respectively.Three P 2p XPS peaks appeared at the binding energy of about 129.5 eV, 133.5 eV and 134.5 eV (Fig. S4), which are assigned to Pδ− (0 < δ < 1) in Ni2P, P3+ in H2PO3 − and P5+ in PO4 3−, respectively. It can be found from Table 2, the supported Ni2P catalysts modified with the chelators of ETDA and NTA exhibit much higher proportions of Ni2P than the traditional Ni2P/WSNs catalysts without chelator addition.Naphthalene is chosen as the model compound to investigate the aromatic hydrogenation activity of synthesized catalysts, and the reaction pathway of naphthalene hydrogenation is displayed in Fig. S5. As shown in Fig. 6 , the hydrogenation conversion of the EDTA and NTA modified catalysts are higher than that of supported Ni2P catalysts without chelators. The decalin selectivity graph is shown in Fig. 7 . In the experimental temperature range, EDTA-modified catalysts show high hydrogenation ability, among which Ni2PE(1.5)/WSNs exhibits the highest decalin selectivity (almost reaching 100%). In the case of the NTA modified catalysts, high decalin selectivity of naphthalene is shown in temperature of 300 °C, but then decreases a lot when the reaction temperature enhances.The Ni2P supported series catalysts are successfully prepared on wrinkle silica nanoparticles (WSNs) support through temperature programmed reduction, which can be confirmed by the Ni2P characteristic peaks as shown on the XRD pattern. More importantly, the peaks of Ni2P become broader and weaker as the addition of chelators, indicating that the corresponding sizes of Ni2P also become smaller according to the Scherrer formula (Song et al., 2018). These results are in agreement with the TEM results, as shown in Table 1. The average sizes of Ni2P particles decrease as the increasing addition of chelators. Among all the catalysts, Ni2PE(1.5)/WSNs catalyst possesses the smallest Ni2P average particle size of only 2.6 nm. The Ni2P particle size obtained by the addition of EDTA and NTA in this work is smaller than the published works, realizing the formation of utral-small Ni2P particles (Table S1). Previous reports (Rui and Smith, 2010; Zhang et al., 2017) have been reported that the chelators can form a metal complex with the active metals, so as to restrict the aggregation of metal specials on the supports. Therefore, the chelators have a positive effect on the formation of smaller Ni2P particles on the supported catalysts.The addition of chelator can contribute to the formation of Ni2P active phase. As seen in XPS results, the Ni2P proportion of the supported Ni2P catalysts with the addition of EDTA and NTA is higher than that of Ni2P/WSNs catalysts without chelators. More importantly, when the content of chelators is enhanced, the corresponding percentage of Ni2P is also increased. Therefore, it can be speculated that EDTA and NTA chelators are beneficial to the reduction of Ni2P, which can effectively avoid the generation of other impurity phase. Moreover, H2-TPR results showed that the reduction-peak interval between Ni and P species becomes smaller as the addition of chelators, which further confirms the positive effect on the formation of Ni2P active phase. During the H2-TPR process, nickel species were first reduced and then H2 dissociated from metallic Ni sites to form hydrogen spillover effects, which contributes to the reduction of the P species (Chen et al., 2010; Yang et al., 2013). The high dispersion of nickel species influenced by the chelator EDTA and NTA could have a chain effect on the later reduction of the P—O bond. Therefore, the better Ni dispersion of supported Ni2P catalysts resulted from chelator additions are beneficial to the formation of nickel phosphide phases.On close observation of the structure of the chelating agent, it can be seen from Fig. 8 that there are four carboxyl groups and two nitrogen atoms in the case of EDTA, while there are only three carboxyl groups and one nitrogen atom with regard to NTA (Rufus et al., 2004). It is well known that Ni ion has a coordination number of six. When the Ni ion is bound to the chelator EDTA, all six ligand groups of the EDTA fill all the available coordination sites of Ni ion (Wang et al., 2002). However, four of the six coordination sites of Ni ion are occupied by the chelator NTA, leaving two free binding sites on the Ni ion (Lauer and Nolan, 2002). Therefore, Ni-EDTA complex is more stable compared with Ni-NTA complex since all coordination sites are involved in the formation of Ni-EDTA complex. In this case, the addition of EDTA can significantly reduce the interaction between Ni2P and the support, thus forming relatively small Ni2P particles with good dispersion.The result of naphthalene hydrogenation can also reflect the different stabilities of the chelators EDTA and NTA, as shown in Fig. 7. The NTA modified catalysts exhibit high decalin selectivity under low temperature of 300 °C, which may be ascribed to the two free binding sites available on the Ni metal ion. However, the hydrogenation conversion of NTA modified catalyst decreases dramatically when the reaction temperature raises higher, while EDTA tailored catalysts still remain high stability under high temperature owing to all the coordination sites involved in the formation of complexation. The long-period (100 h) naphthalene hydrogenation experiments (Fig. S6) and XPS results of the spent Ni2PE(1.5)/WSNs catalyst (Fig. S7 and Table S2) also shows that Ni2PE(1.5)/WSNs catalyst possesses outstanding catalytic stabilities.A facile strategy was developed for the preparation of the supported Ni2P catalysts with small Ni2P particles through the addition of chelators in this research. The reaction evaluation of naphthalene hydrogenation has shown that Ni2P supported catalysts with the addition of EDTA and NTA chelators display higher catalytic activity than the traditional catalysts, among which Ni2PE(1.5)/WSNs catalyst with the smallest Ni2P particles nearly reaches 100% naphthalene conversion. Moreover, EDTA modified catalysts display high catalytic stability under high temperature due to the chelator EDTA completely bound to all coordination sites of Ni ions. On the contrary, NTA modified catalysts show high decalin selectivity on the temperature of 300 °C, but decrease a lot when the reaction temperature becomes higher due to the two free binding sites on the Ni ion.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This research is financially supported by the National Natural Science Foundation of China (No. 21878330), Key Research and Development Program of Ministry of Science and Technology of China (No. 2019YFC1907602) and Scientific Research and Technology Development Program of China National Petroleum Corporation (2020B-2116).The following is the Supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.petsci.2022.11.017.

Ni2P supported catalysts exhibit high catalytic activities in hydrogenation reaction, of which the particle sizes of Ni2P active phases are the key influential factor. This research focus on the effect of chelators on the size of Ni2P particles over wrinkle silica nanoparticles (WSNs) by introducing chelating agents EDTA and NTA during impregnation process. The characterization results show that chelators modified catalysts possess smaller size of Ni2P particles than the unmodified Ni2P catalysts. Among all the synthesized catalysts, the EDTA modified Ni2PE(1.5)/WSNs catalyst possesses smallest average particle size of Ni2P, only 2.6 nm. Moreover, the Ni2P catalysts with the assistance of EDTA exhibits better catalytic activity than that of NTA under high reaction temperature, which can be ascribed to the strong bonding between EDTA and Ni. And the EDTA modified Ni2PE(1.5)/WSNs catalyst shows highest hydrogenation ability, almost reaching 100% decalin selectivity.

Data will be made available on request.Carbon nanofibers (CNFs) have attracted considerable attention in a wide variety of applications, including biosensing [1], energy storage [2], water purification [3] and catalysis [4]. CNFs are cylindrical nanostructures that consist of stacked graphene sheets of various sizes and orientations. Their length may vary from tens to hundreds of nanometers, while their diameter may reach up to tens of micrometers. CNFs possess a large surface area which can be functionalized with a variety of chemical species. In addition, CNFs have high electrical conductivity, which makes them ideal candidates for electrochemical biosensing [5–7].CNFs can be grown using a variety of techniques, including electrospinning [8], chemical vapor deposition (CVD) [9] and hot filament assisted sputtering [10]. Plasma enhanced CVD (PECVD) is a widely used method for growing vertically aligned CNFs (VACNFs) at relatively low temperatures. A PECVD growth process for CNFs typically involves a carbonaceous gas (e.g. C2H2, C2H4) and an etchant gas (e.g. NH3 or H2), which are activated in a glow discharge. The growth of CNFs occurs through the nucleation of a nanoscale metal catalyst layer (e.g. Ni, Fe, Pd or Pt) [11]. An adhesive layer (e.g. Ti, Cr, W) is often deposited between the substrate and the catalyst layer, in order to prevent the intermixing of the catalyst and substrate materials.The microstructure and macroscale morphology of CNFs depend on the process parameters, e.g. temperature, time, gas flow, reactor pressure and plasma power. Moreover, the composition and thickness of adhesive and catalyst layers also affect these properties. These effects have been reported in various studies. For example, Melechko et al. demonstrated that CNF growth mode changes from tip-type to base-type by changing the flow ratio of carbonaceous and etchant gases in the PECVD process [12]. Röthlisberger et al. documented the effects of Ni layer thickness on bidirectional growth of CNFs (i.e. both tip-type and base-type simultaneously) [13]. Merkulov et al. reported the growth of individual VACNFs by controlling the size of catalyst layer dots, using Ni as the catalyst layer and Ti as the adhesive layer [14].An aspect of the reaction that is often overlooked is the interaction between the substrate, catalyst and adhesive layers. To the best of our knowledge, this phenomenon has not been systematically investigated. Furthermore, many studies on electrochemical biosensing of electrodes do not address the effects of well controlled macroscale geometrical parameters – length, diameter, population density, etc. – on the performance of the electrode.In this paper, we report PECVD growth of CNFs on two types of substrates – 80 nm Cr + 20 nm Ni and 20 nm Ti + 20 nm Ni (hereafter referred to as Cr-Ni and Ti-Ni, respectively). For each substrate type, we prepared four sets of CNF samples by varying the growth time (1, 5, 10 and 30 min), while the other growth parameters were kept constant. We studied the differences in macroscale geometry between the CNFs prepared under these conditions using scanning electron microscopy (SEM). We used phase diagrams and thermodynamics combined with detailed transmission electron microscopy (TEM) study to rationalize the differences between CNFs grown on the two types of substrates. Finally, we investigated basic electrochemical properties of each type of CNF, and rationalized the effects of macroscale morphology on pseudocapacitance and electrochemical windows of the CNFs.p-type Silicon wafers (Siegert Wafers, Germany) were used as substrates for all the samples. First, the wafers were coated with adhesive and catalyst metal layers – 20 nm Ti followed by 20 nm Ni for Ti-Ni-CNFs, and 80 nm Cr followed by 20 nm Ni for Cr-Ni-CNFs. An electron beam evaporator (MASA IM-9912) was used for depositing metal layers. The chamber pressure was approximately 2 × 10−7 mbar during evaporation. Subsequently, the wafers were cleaved into smaller pieces, measuring approximately 7 mm × 7 mm. Finally, CNF growth was carried out using a PECVD reactor (Aixtron Black Magic).The PECVD process was carried out as follows: First, the chamber was pumped down to 0.1 mbar. Then, the chamber was heated to 400 °C with a ramp speed of approximately 250 °C per minute. When the temperature reached 395 °C, the chamber was injected with a 100 sccm NH3 buffer. The ramp rate was then increased to 300 °C per minute and the chamber was heated up to 600 °C. When the temperature reached 575 °C, 230 W DC plasma was ignited. 30 sccm C2H2 was simultaneously injected into the chamber, while the flow rate of NH3 was increased to 125 sccm. These parameters were maintained for 1, 5, 10 or 30 min, in order to prepare CNFs of four different lengths. The chamber pressure was approximately 3 mbar during the growth process.CNF morphology and geometry were studied using SEM (Zeiss Supra 40 and Zeiss Sigma VP). Length, diameter and area analyses were carried out using imageJ. We estimated the average length and diameter by measuring 20 CNFs from the cross-sectional SEM images (Fig. 1 ). Area analysis was carried out as follows: Area covered by Ni was highlighted by applying brightness/contrast and threshold settings to the top-view SEM images of 1 min grown CNF samples (Fig. S2). The percentage of white pixels, which corresponds to Ni, was then estimated using imageJ software. An area of 7000 nm × 4800 nm was used in this analysis. While a more detailed analysis would be required to obtain precise quantitative values, our analysis is sufficient to demonstrate that there is a significant difference between the two substrates.TEM imaging was performed on a Jeol JEM 2200FS TEM equipped with an X-ray energy dispersive spectrometer (EDS). Cross-sectional TEM samples were prepared by EAG Laboratories (USA) using a focused ion beam (FIB). Sputtered carbon was used as the filler material. The sample was coated locally at the cross-section site with two additional layers of carbon.Binary phase diagrams (Figs. 4, S6) were generated using FactSage Education 8.1 Package. FactSage is a fully integrated database computing system in chemical thermodynamics and consists of a variety of information, database, calculation and manipulation modules that access various pure substances and solution databases [15].Cyclic voltammetry was performed with a Gamry Reference 600 potentiostat and Gamry Framework software in a three-electrode setup with a Ag/AgCl as reference electrode and a Pt wire as the counter electrode. The solutions were purged with nitrogen gas for at least 15 min. The potential window measurements were done in 0.15 M H2SO4 and PBS pH 7.4 (NaCl (137 mM), KCl (2.7 mM), Na2HPO4 (10 mM), and KH2PO4 (1.8 mM)). The outer-sphere redox (OSR) probe was 1 mM Ru(NH3)6 2+/3+ prepared from hexaammineruthenium(III) chloride (Sigma-Aldrich) dissolved in 1 M KCl (Merck Suprapur).SEM images presented in Fig. 1 demonstrate the effect of growth duration on the length of CNFs on Cr-Ni and Ti-Ni substrates. Measured values are presented in Fig. 2 and Table 1 . There is a significant difference between the evolution of length in both types of substrates. Samples grown for 1 min do not contain nanofibers – instead, we observe the initial stages of CNF nucleation. There are notable differences between Cr-Ni and Ti-Ni substrates already at this stage (Fig. 3 ), which we elaborate below.At 5 min, CNFs are clearly visible in both the substrates. On Cr-Ni substrates, the average length of CNFs is approximately 221 nm, while on Ti-Ni substrates, the average length is considerably larger - 361 nm. After 10 min of growth, the corresponding values are 415 nm and 774 nm, respectively. After 30 min, however, the CNF lengths on both substrates are similar - 915 nm and 873 nm, respectively. Thus, the rate of CNF growth is lower in Cr-Ni substrates compared to Ti-Ni substrates up to 10 min. Interestingly, the length seems to saturate at <1 μm for both substrates. Moreover, CNFs grown on Cr-Ni substrates have a narrower distribution of length, i.e., they are more uniform in length. The average diameters of CNFs do not differ significantly between the two substrates. The average diameters also remain similar for different growth duration (approx. 70 nm). However, Ti-Ni-CNFs have a wider distribution of diameters.It should be noted that measuring the average dimensions of CNFs is not straightforward. The longer CNFs are curved, and their diameters are not strictly identical along the entire length. Moreover, the smaller CNFs are likely to be obstructed by larger ones in the SEM micrographs. However, we believe that the values presented in this work provide a qualitative comparison between different batches.SEM images presented in Fig. 3 demonstrate the difference between the population density of fibers grown on Cr-Ni and Ti-Ni substrates. The nucleation of Ni film occurs differently in both types of substrates. On the Cr-Ni substrate, we observe that the Ni film breaks down into particles of more uniform size. Moreover, a larger proportion of the surface is covered with Ni nanoparticles. On the other hand, on the Ti-Ni substrate, we see a less even distribution of Ni nanoparticles. Moreover, a smaller proportion of the surface is covered with Ni. In the Cr-Ni substrate, 50.04 % of the visible surface is covered with Ni, whereas the corresponding value for Ti-Ni substrate is only 37.13 % (Fig. S2, Table 1). Fig. 3 and Table 1 support our observations from Fig. 1, that CNFs grown on Cr-Ni substrates have a narrower distribution of length and diameter. A similar analysis could not be carried out for the 5-, 10- and 30-min grown samples because longer CNFs partially obstruct the underlying surface.We can safely assume that the entire Ni film is nucleated in the early stages of the reaction, therefore, the population density of CNFs remains the same across the four growth times (1, 5, 10 and 30 min). Thus, even though lengths of CNFs grown for 30 min are similar for both substrates, we observe a greater population density of CNFs in Cr-Ni substates. This results in a larger surface area, and hence a larger pseudocapacitance as discussed below.In addition to the reaction between the carbon source and the catalyst layer that has been extensively studied [14,16–18], the much less investigated interaction between the catalyst and adhesive layers also has a significant role in the CNF growth process. To rationalize interfacial effects, we use binary and ternary phase diagrams that are available in the literature (see below) and the concept of local equilibrium at the interfaces between the phases. Local equilibrium is defined so that the equilibrium exists only at the interfaces between the different phases present in the system. This means that the thermodynamic functions are continuous across the interface, and the compositions of the phases right at the interface are very close to those indicated by the equilibrium phase diagram. This also indicates that there are activity gradients in the adjoining phases. These gradients, together with the diffusivities, determine the diffusion of components in the various phases of a joint region. This concept is different from the assumption of global equilibrium, where it is assumed that the system's Gibbs free energy (G) function has reached its global minimum value and then, the system is in mechanical, thermal, and chemical equilibrium with its surroundings. Consequently, there are no gradients inside the individual phases, and no changes in the macroscopic properties of the system are to be expected.Binary phase diagrams of the Ni-Ti and Ni-Cr systems (Fig. 4 ) provide us information about the equilibrium phases present in these systems at different temperatures. Phase diagrams do not provide any information about the kinetics of these reactions or the spatial distribution of the phases. However, they provide a useful framework for comparing the two systems, and especially for ruling out thermodynamically impossible phases [19].In the Cr-Ni phase diagram (Fig. 4(a)), there are no intermetallic compounds that are stable at 600 °C. Based on the assumption that local equilibrium is established in the system, we can see that at this temperature, stable phases are BCC_A2 (i.e. BCC Cr), BCC_A2 + FCC_A1, and FCC_A1 (i.e. FCC Ni). In addition, whereas the solubility of BCC Cr is very high to FCC Ni, the solubility of FCC Ni to BCC Cr is orders of magnitude smaller. Therefore we expect that most of the Ni would remain unreacted and available for CNF catalysis, at the growth temperature (600 °C). This is consistent with the observation of the relatively high population density of fibers grown on Cr-Ni substrate compared to Ti-Ni substrate (Fig. 3). However, at lower temperatures, CrNi2 is stable as well. Hence, if the ramp rate is low, it is possible that intermetallic compound CrNi2 would start to form during ramping, which would reduce the amount of Ni that is available for CNF growth. This explains the significantly lower population density of fibers on Cr-Ni substrate at a low ramp rate (Fig. 5 ).TEM micrograph shown in Fig. 6 as well as the associated EDX analyses (Fig. S3) provide strong evidence for the above reasoning. We can see that the interface after the CNF growth consists of Cr with some dissolved graphite and/or chromium carbide, and there is practically no Ni left at the interface. There are also no intermetallic layers visible at the interfacial area consistent with the binary phase diagram.On the other hand, the Ti-Ni system contains several intermediate phases at 600 °C (Fig. 4(b)). Thus, at equilibrium, we expect the formation of Ti2Ni, TiNi3, α-Ti and Ni. Even though our system is not at equilibrium (precursor gases and plasma are injected into the system after a few minutes of ramping), we can reasonably assume that a significant part of Ni reacts with Ti or diffuses through the Ti layer towards the Si substrate. The latter is driven by the fact that Ni has a high affinity for forming silicide [20] (Fig. S6). Thus, the more thermal energy we provide to the system, the less Ni will be available for CNF formation. The following observations support this hypothesis: (1) When we decrease the ramp rate, there is no CNF growth on Ti-Ni substrates. (2) If we add a pre-annealing step in the recipe, we get no CNF growth (some samples still result in growth under these conditions, but the vast majority do not). (3) CNF growth on Ti-Ni substrate is not very repeatable. On the other hand, Cr-Ni substrates resulted in uniform CNF growth across multiple batches. Finally, (4) the TEM cross-section (Fig. 6) clearly shows the formation of an intermediate reaction layer at the Ti/Si interface, which contains significant amounts of Ni. Fig. 2 shows that 5 min and 10 min Cr-Ni-CNFs are considerably shorter than their Ti-Ni counterparts, while they reach similar lengths after 30 min. Can we somehow rationalize this observation based on the thermodynamics and kinetics of the system? One important factor is the dissolution of carbon into the underlying metal layers. Our EDS scans indicate the presence of carbon in both Cr and Ti underlayers (Fig. S3). But what are the differences? Firstly, the 80 nm thick Cr layer is likely to dissolve significantly more carbon than the 20 nm thick Ti layer simply owing to its higher overall volume. Secondly, based on the diffusion coefficients calculated from the diffusion couple experiments [Cr,Ti] the intrinsic diffusion coefficient of carbon in Cr is slightly smaller than that of Ti, indicating that it will likely take considerably more time for carbon to reach the Cr/Si than the Ti/Si interface. In fact, there is a clear carbon peak at approximately the midpoint of the Cr layer (Fig. S3(c)), while there is no carbon peak at the Cr-Si interface (Fig. S3(d)). This clearly indicates that carbon has not yet reached the bottom of the Cr layer. On the other hand, there is a clear carbon peak at Ti layer as well as the Ni-silicide layer below it (Fig. S3(e, f)), which indicates that carbon has diffused throughout the underlayer already after 10 min of growth. Hence, as the Cr layer is not fully saturated with carbon after 10 min, some of the available carbon will continue to dissolve into Cr instead of forming CNFs, and this is reflected in the observed lengths of the growing CNFs.The presence of oxygen is expected to play a significant role in the evolution and distribution of phases at the Si-Ti and Si-Cr interfaces [21]. In our experiments, both Cr-Ni and Ti-Ni substrates were exposed to air between our processing steps. Hence we expect the substrates to be saturated with oxygen at room temperature. It should be noted that Ti has a higher affinity for oxygen than Cr [21]. From the ternary Ti-Si-O and Cr-Si-O phase diagrams, one can readily see that the thermodynamic stability of the interfaces is very different. Based on Fig. 7(a) and (b), local thermodynamic equilibrium is possible between Cr and SiO2, whereas it is not the case with Ti and SiO2 as in the former case there is a tie-line connecting the two phases directly, a feature that is missing in the Ti-Si-O phase diagram. This naturally means that from the thermodynamic point of view, Cr-SiO2 interface is far more stable than Ti-SiO2 where the formation of several additional phases is expected.Richter et al. reported that when Ti starts to form TiSi2, the oxygen that was initially contained in the consumed Ti region accumulates into the Ti film near the Ti-TiSi2 interface. This phenomenon, commonly referred to as the “snowplow effect”, slows down and ultimately prevents the formation of further TiSi2. In experiments done on a pure Si-Ti system (i.e. no Ni overlayer), it has been reported that oxygen accumulation leads to the formation of a Ti5Si3 interlayer between TiSi2 and Ti [24]. This can be readily rationalized by the fact that out of the Ti-silicides only Ti5Si3 exhibits significant ternary oxygen solubility as well as can at the same time exist at local equilibrium with SiO2. As our annealing time is rather short, we do not see the sequential formation predicted by the stable phase diagram in Fig. 7(b) but instead are dealing with local thermodynamic equilibrium [25]. Nevertheless, we can conclude that the inherently unstable Ti-SiO2 interface will undergo interdiffusion and redistribution of species, as also shown in the TEM micrographs. As the solubility of oxygen to Ti is extremely high, and the driving force for the dissolution is high [26], it is likely that SiO2 at the interface will also be mostly reduced and the resulting oxygen incorporated into the Ti-phase.On the other hand, no redistribution of oxygen was reported during Cr-silicide formation under similar experimental conditions [21]. This behavior can again be rationalized based on the available ternary phase diagram (Fig. 7(a)), which shows, as discussed above, that local equilibrium exists between Cr and SiO2. Thus, there is no need, from the thermodynamics of the system, to rearrange the phase at the interphasial area and therefore also the SiO2 at the interface can be expected to stay intact. Note also that the solubility of oxygen in Cr is much smaller than that in Ti.What are the consequences of the above discussion for our system? We can see that in the Cr-Ni layer structure, there is hardly any Ni at the interface (it is located exclusively at the tips of the fibers) whereas in Ti-Ni layer structure, there is a Ni-silicide phase between the Ti and S as shown by the TEM micrographs. It has been reported that the formation of NiSi is suppressed in the presence of a thin native oxide film (which is present in our samples) [27]. On the other hand, Lee et al. reported that the stability of NiSi is enhanced if Ti is incorporated in Ni thin films [28]. In their study, Ni silicidation reaction was observed at a significantly lower temperature due to the incorporation of Ti, and it was proposed that Ti reacts with interfacial SiO2, resulting in the formation of a Ni-permeable diffusion membrane [28]. Again, this behavior is evident based on the phase diagrams in Fig. 7, which shows that because of the absence of local equilibrium between Ti and SiO2 additional phases will form, and oxygen redistribution (partial or complete reduction of SiO2) is inevitable as discussed above. This will then provide Ni feasible access to Si substrate and result in formation of NixSi1-x phase(s).We used cyclic voltammetry (CV) to investigate the electrochemical properties of CNF samples. In this work, we focus on two important electrochemical parameters - analytical potential window and pseudocapacitance. Sulphuric acid is among the most widely used electrolytes in electroanalytical chemistry, while Phosphate Buffered Saline (PBS) is frequently used as an electrolyte in biosensing applications. Therefore, we determined the pseudocapacitance and analytical potential window of the CNFs in these two electrolytes. The effects of the dimensions of CNFs on the electron transfer kinetics were studied using an OSR probe Ru(NH3)6Cl3 (in PBS), which is known to be insensitive to the surface chemistry of the electrodes [29].The solvent window of an electrolyte is defined as the potential range between the oxygen evolution reaction (OER) and hydrogen evolution reaction (HER), which occur at the cathodic and anodic ends, respectively. From the electroanalytical point of view, it is more useful to define an analytical potential window where the analyte signal can be precisely measured, even though the exponential increase in current due to HER and OER is not yet seen. The analytical potential window is defined using a self-chosen threshold current value and is, by definition, narrower than the solvent window. We selected a threshold current density of ±150 μA/cm2 for this purpose.The analytical potential windows (Table 2 ) were determined with cyclic voltammetry in H2SO4 and in PBS. The cyclic voltammograms are presented in Fig. S5. Ti-Ni-CNFs possess larger potential windows than Cr-Ni-CNFs in both electrolytes for all CNF lengths. However, the difference between the potential windows of Ti-Ni-CNFs and Cr-Ni-CNFs was smaller in PBS in comparison to H2SO4, where Cr-Ni-CNFs and Ti-Ni-CNFs with varied lengths showed potential windows ranging from 1.02–1.42 to 1.23–1.54 V respectively. Cr-Ni-CNFs demonstrated a larger potential window in PBS compared to H2SO4, indicating better stability of the Cr-Ni-CNFs in a physiological saline environment. Contrary to Cr-Ni-CNFs, the potential windows of Ti-Ni-CNFs are approximately similar in both PBS and in H2SO4. Moreover, the width of the potential windows for Cr-Ni CNF decreased with an increase in fiber length.Pseudocapacitance (C dl ) is a faradaic property that arises on electrode surfaces during electrochemical reactions, wherein due to thermodynamic reasons, charge q depends on potential, resulting in pseudocapacitance C = d(Δq)/dV [30]. Several electrochemical processes contribute to pseudocapacitance, including adsorption, intercalation and surface redox reactions. We expect the pseudocapacitance to increase with the increase in CNF length as well as population density, since both these parameters increase the available surface area.The pseudocapacitance of the electrical double layers of all electrodes was calculated from cyclic voltammograms recorded in blank PBS (pH 7.4) and in H2SO4 (pH 0.8) at different scan rates (10 mv/s–400 mV/s). The difference between anodic and cathodic current densities (defined from the measured current dividing it by the geometric area of the electrode) at different scan rates and the equation Δi = 2 × C × v was used to determine the numerical value of the pseudocapacitance.Cr-Ni-CNFs showed higher C dl in H2SO4 than in PBS electrolyte, while C dl for Ti-Ni-CNF electrodes is approximately similar in both electrolytes (Table 2). As expected, the C dl of the electrodes increases with the increase in the length in both Cr-Ni and Ti-Ni systems, indicating the increase in the surface area. C dl of 5- and 10-minute grown Ti-Ni-CNFs deviate from the trend. Nonetheless, C dl of the longest CNFs is significantly larger than the shortest CNFs - about two times larger for Ti-Ni-CNFs and three times larger for Cr-Ni-CNFs. However, C dl of the electrodes in Cr-Ni system is higher, indicating that the overall surface area, which is contributed by the population density, diameter and lengths of the fibers, is higher in comparison to Ti-Ni-CNFs. These results correlate with our analysis above, where we demonstrated that Cr-Ni-CNFs have greater population density, and therefore, greater surface area in comparison to Ti-Ni-CNFs. Overall, Cr-Ni-CNFs showed enhanced capacitance with a decrease in the potential window with respect to the increase in the fiber length. While length of the fibers did not influence the potential window of Ti-Ni-CNFs in any systematic way, C dl , however, increased with the increase in fiber length.Peak potential separation (ΔE p ) and the ratio of the oxidation to the reduction peak current (I p,a /I p,c ) values at 1 V/s scan rate are shown in Table 2 for electrodes with the shortest and longest CNF lengths studied for both Cr-Ni and Ti-Ni interfaces. Based on these results, the electron transfer kinetics appear nearly reversible for all the electrodes. However, I p,a /I p,c is closer to 1 for the CNFs with 1-minute growth in comparison to CNF with 30 min growth on both types of substrates. This indicates higher reversibility of the OSR redox reaction on electrodes with shorter fibers. Moreover, according to the ΔE p values, the electron transfer kinetics appear slightly faster with the longest CNF in comparison to the shorter CNF on both interfaces. However, this may be caused by thin liquid layer formation, which will have a similar effect of reducing the ΔE p values than higher heterogeneous electron transfer (HET) rates have. This phenomenon occurs at a certain ratio of the diffusion layer to surface feature thickness and is therefore dependent on both the fiber length and the scan rate. As the nanofiber lengths reach hundreds of nanometers, thin liquid layer formation is a feasible option in this system, especially in the case of the longest CNFs within the scan rates used in this study. A detailed study of the thin-liquid layer formation on these electrodes is presented in a separate work [29].We have demonstrated that the interaction between the catalyst and adhesive layers plays a notable role in the growth behavior, the macroscale morphology of PECVD-grown CNFs and their electrochemical performance. Our results show that (1) Cr-Ni-CNFs have a larger population density than Ti-Ni-CNFs, (2) Ti-Ni-CNFs grow faster for the first 10 min of the growth process, however, both types of CNFs saturate to similar lengths after 30 min, and (3) the macroscale morphology of the CNFs can be used to tune their electrochemical properties. All these features can be rationalized by considering the interfacial interactions in the two systems. Owing to the inherent instability of the Ti-Ni interface at our process temperature, a portion of the Ni in the catalyst layer diffuses through Ti and forms a silicide. It is likely that oxygen redistribution also plays a role in the formation of NixSi1-x phases. As a result, a smaller amount of Ni is available for CNF nucleation, and therefore, Ti-Ni-CNFs have a smaller population density than their Cr-Ni counterparts. The stability of the Cr-Ni interface at our process temperature, on the other hand, results in a higher availability of Ni for CNF nucleation, and therefore, a larger population density of fibers. It is likely that the difference in the rate of growth between the two types of substrates is caused by the gradual dissolution of carbon into the thicker Cr layer. Further, we show that the macroscale geometry of fibers influences, for instance, the pseudocapacitance of CNF electrodes without significantly affecting the electron transfer kinetics. Thus, this study paves the way towards designing application-specific CNF electrodes by precisely controlling their macroscale morphology. Ishan Pande: Conceptualization, Methodology, Writing - Original draft preparation, Writing - Reviewing and Editing, Investigation, Resources, Visualization. Laura Ferrer Pascual: Investigation, Writing - Original draft preparation. Ayesha Kousar: Writing - Original draft preparation. Emilia Peltola: Funding acquisition, Writing - Review & Editing. Hua Jiang: Investigation. Tomi Laurila: Conceptualization, Methodology, Writing - Original draft preparation, Writing - Reviewing and Editing, Resources, Funding acquisition, Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by funding from the Academy of Finland (#321996 and #328854) and Jane and Aatos Erkko Foundation. The authors acknowledge the provision of facilities and technical support by Aalto University at OtaNano - Nanomicroscopy Center (Aalto-NMC) and at Micronova Nanofabrication Centre. I.P. would like to thank Elli Leppänen and Petri Mustonen for discussions regarding the PECVD process, and Dr. Jani Sainio for help with sample characterization. Supplementary material Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.diamond.2022.109566.

The effect of catalyst materials and different process parameters on the growth of carbon nanofibers (CNFs) has been widely investigated. Typically, an adhesion metallization is required together with the catalyst to secure adequate attachment to the surface. The interactions within this multilayer structure and their effect on CNF growth and morphology has, however, not been thoroughly assessed. Thus, this work presents the growth behavior, the macroscale morphology, and the basic electrochemical characteristics of CNFs grown on two types of substrates - (1) Si + 80 nm Cr + 20 nm Ni, and (2) Si + 20 nm Ti + 20 nm Ni. Our results show that the macroscale geometric parameters of CNFs can be readily altered by using different adhesive layers. The inherently unstable Ti-Ni interface results in diffusion of Ni towards the silicon wafer to form silicide, which reduces the amount of available Ni for CNF nucleation, and therefore, the population density of fibers is reduced. On the other hand, the Cr-Ni interface results in a larger population density, but the rate of growth is reduced due to diffusion of carbon into the thicker Cr layer. The results are rationalized by using relevant binary and ternary phase diagrams. Further, cyclic voltammetry experiments show that the pseudocapacitance of CNFs shows a correlation with the length and population density of fibers, while the electron transfer kinetics appear nearly reversible for all the electrodes. This simple approach can be used for tailoring CNFs for specific applications by controlling their macroscale geometrical parameters.

Data will be made available on request.Within the renewable energy portfolio, biomass is a key resource for mitigating climate change and reducing reliance on fossil fuels [1]. Lignin is a major component of biomass and has the potential to be converted into hydrocarbon fuels and value-added compounds [1]. Fast pyrolysis technique may be used to extract desired compounds from biomass, and the liquid produced is referred to as bio-oil [1]. Bio-oils produced from the rapid pyrolysis of lignin often include a high concentration of oxygenated compounds. Due to the high oxygen content of bio-oil, it has significant disadvantages such as high viscosity, low solubility in other hydrocarbons, low volatility, corrosiveness, and low heating value, which prevents its direct use as a transportation fuel [2]. Additionally, the reactive compounds have extremely low stability during the storage process. As a result, in order to create carbon neutral fuels from bio-oil, it must be upgraded to meet these fuel specifications.Catalytic hydrodeoxygenation (HDO) is one of the rapidly developing technologies for upgrading bio-oils to produce transportation fuel or value-added chemicals [1]. This process is generally performed by using a hydrotreating process under high pressure [3]. It is essential to deoxygenate bio-oils to eliminate the high viscosity and chemical instability and also increase heating value [2]. Ideally, oxygen is removed in the presence of catalyst with minimal saturation of the aromatic rings, which also reduces the hydrogen consumption [1].In order to make the HDO process more economically appealing, H donors other than hydrogen gas have been considered as alternatives [4]. Water is a cheap and available source of hydrogen that near its critical point can also be an excellent solvent for liquid phase HDO, making the process sustainable and economically more feasible. In this alternative process, catalysts must be designed to facilitate water dissociation on the surface, which will provide hydrogen to the catalytic HDO reaction. Noble metal and non-noble metal catalyst supported on carbon have been studied by our research group before, demonstrating the feasibility of the viability of “H2-free” HDO reactions using water as reaction media [4–7]. Generally speaking, noble metals present higher activity compared to transition metal catalysts. Ru/C displays the highest activity among studied catalysts (i.e. Au/C, Pd/C, Rh/C and Ru/C). The high activity is attributed to the smaller mental particle size, greater dispersion of metal particles and its intrinsic activity for this reaction [4].Guaiacol has been selected as lignin model compound for the HDO reaction because the molecule is composed of two typical functional groups, i.e. hydroxy and methoxy groups [2]. The targeted products of guaiacol HDO reaction are ideally hydrocarbons or hydrogenated aromatic compounds. Transition metal-based catalysts, noble metal catalysts and zeolites-based catalysts have been widely investigated to produce hydrodeoxygenated aromatic compounds from guaiacol. To assist HDO reactions two functions are required for the catalyst, including the activation of the oxygen containing groups on the reactant (water activation) and the hydrogen donation (hydrodeoxygenation reaction) [4,6,7]. For noble metal catalysts such as Pd/C, Rh/C, Au/C, Ru/C and Pt/NC [4,6,7], the activation of the O-containing groups on the reactant take place on the metal sites or the metal-support- interface, and the hydrodeoxygenation reaction occurs on the surface of noble metals [1]. Despite their high activity and stability in the HDO reaction, noble metals constitute critical raw materials with limited availability, and it is necessary to identify earth abundant, low-cost alternatives which are more sustainable. Ni catalysts have been extensively investigated in HDO process considering the cheap price and similar performance in comparison to the noble metal catalysts [8]. Ni-based catalysts have good capability towards C-C bond rupture, and high activity in hydrogenation. However, Ni catalysts are susceptible to coke deposition, leading to the deactivation of the catalyst and hence poor stability [5]. The selection of support can significantly influence coking resistance of Ni-based catalysts [9]. It is reported that the performance of the conventional catalysts can be improved by using Zr2O as promoter in order to prevent coke deposition [10]. Therefore, Zr2O was explored as a promoter for this research effort.Carbon materials are ideal candidates for catalysts supports for HDO reaction, since they are inert, with limited interactions with the active phase. Activated carbon is also a viable support considering its hydrophobicity, which could decrease the possibility of metal deactivation in water-existing reaction systems [11]. Graphene and its derivates deserve more attention due to their large surface area, and unusual electronic, mechanical, and thermal properties [12]. Graphene consists of a monolayer of carbon atoms arranged in a hexagonal structure. It is reported that graphene supported catalysts present the highest activity in deoxygenation of vegetable oil among other carbon materials including glassy spherical carbon, activated carbon and mesoporous carbon. The superior activity can be attributed to its large pore size, which facilitates the transportation of reactant and fine dispersion of metal particles on the surface of graphene support [12]. To optimise metal-graphene interaction, further actions could be done such as reducing the nanoparticle sizes, improving the homogeneous distributions of the nanoparticles and increasing the number of defect sites on graphene surfaces [13]. Various methods have been proposed to engineer the electronic structure of graphene, including preparing carbon sheets with different layers and graphene with and without defects, chemical functionalities of graphene, and chemical doping [14,15]. Chemical doping is the introduction of a heteroatom substituting a carbon atom in the graphitic structure [16]. Nitrogen substitution are considered magnificent candidates owing to the fact that they have comparable atomic size and strong valence bonds with carbon atoms [15,16]. The increased deoxygenation capacity of the N-doped samples is attributed to increased activity of the N-support and N-metal interfaces. Such interfaces are envisioned as electron-rich regions capable of activating C-O bonds [17].In our work, different Ni-based graphene supported catalysts with/without nitrogen doping have been synthesised, characterised, and studied in guaiacol HDO reaction using water as hydrogen supplier. Consequently, in this study, the effect of synthesis methods, nitrogen doping and the addition of ZrO2 as a promoter were explored.Reduced graphene oxide (Gr) and N-doped reduced graphene oxide (Gr-n) were employed as supports in this investigation. As a precursor, graphite oxide (GO325) was required for the production of both Gr and Gr-n. GO325 was made using commercial natural flake graphite (G, 99.9% purity) obtained from Alfa Aesar and a modified Brodie process [18]. The reduced graphene oxides were created by thermally reducing GO325 in a vertical cylindrical packed bed reactor. The reactor was loaded (350 mg GO325) and purged with N2 for one hour at room temperature at a flow rate of 100 sccm. Following that, the N2 flow was lowered to 87 sccm, and a flow of 3 and 10 sccm H2 and NH3 respectively, were introduced into the reactor for the N-doped sample. Then, several temperature treatment programs were used. The initial ramp was from room temperature to 100 °C at a rate of 5 °C/min. After that, it was raised to 700 °C at a rate of 5 °C/min and held at that temperature for 5 min. After the furnace heating was completed, the reactor was allowed to cool to 400 °C, the H2/NH3 flows were turned off, and the system was allowed to cool in N2 atmosphere. This graphene was given the name Gr-n. The reduction of the undoped sample was carried out in the same manner as stated above, but without the addition of NH3 to the reactor; this graphene was designated as Gr.Four types of catalysts, labelled as Ni/Gr, Ni/Gr-n, NiZrO2/Gr and NiZrO2/Gr-n were synthesised.Wet impregnation synthesis method it is utilised for supported catalysts. Firstly, the necessary amounts of metal precursor (Ni(NO3)2·6 H2O) were dissolved in deionised water and added to the support that was prior synthesised. After that, in order to obtain homogeneity of the suspensions, they were stirred at room temperature. Secondly, the excess water was removed in a rotary evaporator under reduced pressure and the materials were dried in an oven at 80 °C for 12 h. The last step of the method was the calcination at 500 °C (5 °C/min ramp) for 3 h under an inert atmosphere.10 wt% ZrO2 was impregnated on Gr and Gr-n. 0.0989 g Zr(NO)36 H2O was dissolved in 50 mL of acetone, and 350 mg of the appropriate support was added while stirring for 4 h. The solvent was then evaporated, and the resultant sample was dried in an oven at 100 °C overnight. The samples were then calcined at 350 °C for 5 h with a temperature ramp of 1 °C/min. Then, by using Ni(NO3)2·6 H2O as a precursor salt and the same process as described above, impregnation of 15 wt% Ni was carried out. Ni loading was the same for all synthesized catalysts. Sigma-Aldrich supplied all of the reactants.The catalysts have been characterised by means of XRD, H2-TPR, TEM and XPS. XRD. X-ray diffraction (XRD) analysis was conducted on fresh, reduced and used catalysts using an X′Pert Pro Powder Diffractometer by PANalytical. The 2θ angle was increased by 0.05° every 240 s over the range of 10–80 °. Diffraction patterns were recorded at 40 mA and 45 kV, using Cu Kα radiation (λ = 0.154 nm). H 2 -TPR. Temperature programmed reduction with hydrogen (TPR) analysis was carried out on the calcined catalysts in a quartz tube reactor. 50 mg of sample was heated to 900 °C at a rate of 10 °C/min with a total flow of 50 mL min−1 of 5% H2 in N2. A CO2-ethanol trap was used to condense the gaseous products, mostly water, before the on-stream thermal conductivity detector (TCD). The H2 uptake was quantified by comparison with the hydrogen consumption of a CuO reference sample. TEM. Information about the supported metal particles was acquired by TEM (Transmission electron microscopy) in a JEOL 2100 F field emission gun electron microscope operated at 200 kV and equipped with an Energy-Dispersive X-Ray detector, XEDS. The sample was ground until powder and a small amount was suspended in acetone solution using an ultrasonic bath. Some drops were added to the copper grid (Aname, Lacey carbon 200 mesh) and the solvent was evaporated at room temperature before introduction in the microscope. XEDS-mapping analysis was performed in STEM mode with a probe size of 1 nm using the INCA x-sight (Oxford Instruments) detector. XPS. The XPS spectra were obtained by using non-monochromatic Al radiation (200 W, 1486,61 eV) through a SPECS GmbH with UHV system and with an energy analyser (PHOIBOS 150 9MCD). The samples were pre-treated at 500ºC for an hour in H2 and subsequently, for another hour in He at room temperature. After that, the samples were placed in the sample holder using a double-sided copper tape and transferred to the analysis chamber. The survey spectra were obtained with a 50-eV pass energy and region spectra were obtained at 20 eV pass energy. The binding energy (BE) was finally measured taking as a reference the C1s peak at 284.6 eV and the equipment error was considered as less than 0.01 eV for the determination of energies.The HDO reactions were conducted in a batch reactor (Parr Series 5500 HPCL Reactor with a 4848 Reactor Controller) using 300 mL PTFE gaskets. Catalysts were pre-treated ex-situ in a continuous flow quartz reactor, at 550 °C for 1 h in a 100 mL/min gas flow (H2:Ar=1:4) before being used in the HDO reaction. A quantity of 0.5 g of guaiacol, 49.5 g of water and 0.05 g of catalyst were loaded in a glass-lined steel vessel. To avoid any air contamination, N2 was bubbled through the solution for 5 min under a stirring speed of 100 rpm before closing the reaction vessel. Then, the reactor was heated to the desired temperature (250 °C/300 °C) and held at this temperature for 4 h under a stirring speed of 300 rpm. The pressure of the vessel was fixed according to the natural pressure generated by the solvent (water) at 50/100 bar during the reactions respectively. After the reaction, the spent catalyst was recovered from the liquid by filtration, followed by drying treatment. The organic products were dissolved and recovered with ethyl acetate extraction. The organic compounds products were identified by a gas chromatography-mass spectrometry (GC-MS). Quantitative analysis was performed with a gas chromatograph-flame ionisation detector (GC/FID). The GC injector temperature was 280 °C. The GC separation was performed by using a Carboxen Packed Analytical Column (30 m×320 µm×0.25 µm). A split ratio of 8:1 was held. The column was firstly held at 50 °C for 1 min, then increased to 240 °C at a heating rate of 5 °C/min and held at 240 °C for 10 minThe conversion of guaiacol and selectivity (based on C mol) of the products was calculated using Equation 1 and 2, respectively. (1) Conversion of Guaiacol ( mol % ) = ( m Gin − m Gout ) * N G M G m Gin / M G * N G * 100 (2) Carbon weighted selectivity of product x ( mol % ) = m x * Nx M x ( m Gin − m Gout ) * N G M G * 100 mGin: Initial mass of guaiacol [gr]; mGout: Detected mass of guaiacol in the organic phase [gr]; mx: Mass of product x [gr]. MG: Molar mass of guaiacol[mol/gr]; Mx: Molar mass of product x[mol/gr]. NG: Number of Carbon in guaiacol; Nx: Number of Carbon in product x.The catalytic behaviour of the reduced catalysts in the guaiacol HDO process was studied at 250 °C and 300 °C for 4 h. Activity results are displayed in Figs. 1 and 2. Numerical data can be found in the supporting information ( Tables 1 and 2 respectively).The reaction results of our Ni-based catalysts at 250 °C are presented in Fig. 1. Guaiacol conversion of all the catalysts varied between 15% and 20%. Such conversion ranges might look modest however we shall emphasise that in our study the HDO process is conducted using water as hydrogen source and suppressing completely external H2 input our reaction system. Hence these are interesting results given the significant process savings. A clear effect of the N-doping on the catalytic activity was seen since the Ni/Gr-n exhibited the highest guaiacol conversion (20%) compared to 17% of the undoped catalyst. However, no promotion effect on the conversion of guaiacol was observed when ZrO2 was added as a promoter.Three mono-aromatic compounds including phenol, cresol and catechol were detected on the organic phase. Such products are associated with a potential guaiacol HDO reaction pathway proposed in Fig. 3. The formation of catechol is the most preferred process because the C(sp3)-O bond is most likely to be broken owning to its low bond energy [19]. Despite retaining two oxygens, catechol is one of the intermediates that ultimately leads to more advanced (more deoxygenated) products such as phenol (1 oxygen), benzene, or cyclohexane (fully deoxygenated products) [5]. The production of partial deoxygenated compounds phenol and catechol were improved over Ni/Gr catalyst in comparison to all other samples indicating the superior ability of demethoxylation and/or dihydroxylation [5]. When N was present, the results showed slight variations between the catalysts. The N-doped catalysts produced some phenol, showing that the C-O cleavage was preferred in N-doped systems. The latter is consistent with previous studies of palm oil HDO utilising N-doped activated carbons catalysts, in which the increased deoxygenation capacity of the N-doped samples is attributed to increased activity of the N-support and N-metal interfaces. Such interfaces are envisioned as electron-rich regions capable of activating C-O bonds [17]. Overall, all three samples have similar selectivity towards phenol, cresol and catechol being phenol the most advanced deoxygenation product since it represents just the last step prior to benzene, the final product in the deoxygenation route according as depicted in Fig. 3. Indeed phenol presence in our liquid products mixtures is an encouraging results since despite the absence of external hydrogen source our catalysts can trigger the reaction and get very close to full deoxygenation.To enhance conversion levels, the reaction temperature was raised to 300 °C. Catalytic behaviour of Ni-based catalysts and Gr-supports at 300 °C are presented in Fig. 2, where differences in activity can be observed. Guaiacol conversion of all the catalysts varied between 40% and 55%. This guaiacol conversion was almost doubled in comparison with the catalytic activity presented at 250 °C in Fig. 1 showcasing that upon tunning the reaction parameters remarkable catalytic performance boosting can be attained. Although a full reaction parameters optimisation is beyond the scope of this proof-of-concept paper this result suggests there is big room for overall process improvement reinforcing the potential of “H2-free” HDO strategies.The conversion increased at 300 °C compared to that obtained at 250 °C is a general trend for all the studied catalysts. For example, the conversion of guaiacol obtained over calcined Ni/Gr (17%) at 250 °C was 26% lower compared to that obtained over reduced Ni/Gr (43%) catalyst at 300 °C. Also, temperature and pressure had a greatest influence on the activity of Ni-Gr sample, since 20% of guaiacol conversion was obtained at 250 °C and 45% of guaiacol conversion at 300 °C. By comparing both Figs. (1 and 2), we can determine that temperature and pressure had great influence on the catalytic performance. Therefore, we can conclude that during the HDO process an increase of 50 °C in temperature improved notably the conversion of guaiacol. More remarkably, both the addition of the catalyst and the rise of temperature and pressure, had a positive effect on the selectivity of the products and conversion of the guaiacol. Furthermore, promotion effect on the conversion of guaiacol was observed when adding ZrO2 as a promoter at 300 °C, due to the enhanced oxygen mobility provided by ZrO2 which allowed the activation of C-O bonds [10]. The presence of nitrogen modifies the electrical density and acid/base characteristics of carbon, hence influencing its overall reactivity. As previously reported by W. Jin and co-workers, the increased activity of N-doped supports is due to the beneficial effect of nitrogen, which might aid to stabilize metal particles and prevent their re-oxidation [17]. In addition, nitrogen sites inserted into the carbon network are envisioned as electron-rich reaction sites. Regardless of the reaction routes, there is no question that N as a dopant has a favourable influence on the upgrading reaction [17]. To conclude, the NiZr2O/Gr-n catalyst showed the best result according to the objective, since the production of cresol has been enhanced as a reaction product and is the catalyst that shows best catalytic activity. By nitrogen-doping our carbon support, we may possibly increase the activity [17]. In view of these results, Ni-based catalysts are also an advisable choice considering their relatively low price.It is pointed out that the products analysed were the dominant products in the organic liquid phase. The rest of compounds, up to 100% of selectivity were other aromatics hydrocarbons in addition to some secondaries products derivates of reactions like decarboxylation cracking, and hydrocracking can be found. In terms of selectivity results, three partially deoxygenated mono-aromatic compounds, namely phenol, cresol and catechol were detected in the organic phase as per observed also at 250 °C. The production of the partial deoxygenated products phenol and cresol was improved over Ni/Gr catalyst, indicating its superior ability of demethoxylation and/or dihydroxylation at 300 °C.A schematic representation of the potential HDO pathways of guaiacol is proposed in Fig. 3 [20]. The high selectivity of catechol in all the product distribution indicated a preferential HDO pathway. The formation of catechol was the most preferred process because the C(sp3)- O bond was most likely to be broken, since it presented the lowest bond energy [19]. Unfortunately, benzene was not produced in our reaction system. This result should not be considered unfavourable since the main challenge in an “H2-free” HDO process is to incorporate the hydrogen into the organic molecules without an external H2 supplier. It is important to remember that because we are engaging in an HDO process in which there is no addition of external hydrogen, only locally produced hydrogen during the reaction may combine with the oxygenated molecules. However, the process economic viability and safety issues attributed to hydrogen manipulation and transport make this pathway desirable for oxygenated hydrocarbon upgrading despite the generally low conversion values reported.A quantitative analysis of the nitrogen species presents in the samples was obtained by deconvolution of the XPS spectra of the N1s core level. In Table 1, the data of the N1s region XPS analyses are shown for selected catalysts. The main nitrogen component on the fresh and reduced catalyst is pyridinic nitrogen followed by pyrrolic and quaternary which have been created with similar ratio. The similar ratios and binding energies obtained on both the fresh and spent catalysts seem to indicate that the nitrogen species are stable under reaction conditions.The XPS spectra of Ni 2p3/2 in the reduced-passivated catalyst is shown in Fig. 4. The corresponding binding energies (B.E), atomic percentages and relative proportions are provided in Table 2. Ni/Gr and, NiZr2O/Gr catalysts exhibited a peak at 852.8 eV and NiGr -n showed a peak at 853.0 eV, corresponding these peaks to Ni0. In these catalysts, peaks at higher B.E associated to Ni2+ are also observed. Ni/Gr catalyst could be deconvoluted into contributions at 861.2 eV (Ni2+ satellite peak), 858.1 eV and 855.3 eV (Ni2+) and 852.8 eV (Ni0). NiZr2O/Gr, Ni/Gr-n and NiZr2O/Gr-n exhibited similar species. This suggests the presence of metallic Ni and Ni2+ species on the support’s surface of the catalysts, in accordance with the XRD data obtained ( Fig. 5). In Table 2 it is observed that the atomic percentages of Ni0 are higher in the Ni/Gr-n and NiZr2O/Gr-n catalyst than in the Ni/Gr and NiZr2O/Gr catalyst, suggesting a possible stabilising effect of nitrogen on the metallic Ni in good agreement with the H2-TPR data described below. Fig. 5 shows the different XRD pattern of Ni/Gr, Ni/Gr-n, NiZr2O/Gr and NiZr2O/Gr-n samples for fresh, reduced and post reaction (at 250 °C and 300 °C) catalyst forms.A weak and broad diffraction peak at around 2θ = 26.5° can be observed in the XRD patterns of Ni/Gr and NiZrO2/Gr catalysts (Fig. 5 (A) and (C) respectively), assigned to the (002) planes of the graphitic carbon frame (JCPDS 41–1487) [21,22] with more amorphous structure and lower order in crystallinity. In contrast, this peak for n-doped samples was stronger and sharper, indicating a higher order of graphic structure. Moreover, these materials recovered the graphitic structure to a higher extent [23]. A second diffraction peak at 44.0◦ corresponding (100) plane of graphite indicated the reduction of the GO matrix [24,25]. However, this peak overlapped with the Ni metallic peak at 2θ = 44.5° (JCPDS 87–0712).An interesting finding was that metallic Ni (JCPDS 87–0712) [5] was the main phase instead of NiO for fresh catalysts (except Ni/Gr-n), but NiO presence could not be discarded as the main peak could overlapped with the Ni metallic peak. This was probably due to the partial reduction of NiO during calcination process assisted by the support [26]. No diffraction peak of NiO (JCPS 04–0835) appeared in the XRD patterns of all reduced samples, indicating the success of reduction pre-treatment. Moreover, NiO reduction zones have been observed in the TPR profiles, which are discussed below.The characteristic diffraction peaks at 2θ = 30.2°, 34.5° and 50.2° corresponding to the (101), (110) and (200) refection of ZrO2 phase (JCPDS 70–1769) [27,28] can be clearly observed in the XRD patterns of ZrO2 containing catalysts (Fig. 5 C) and D)). The diffraction peaks of tetragonal phase of ZrO2 were stronger and sharper in n-doped catalyst compared to that of non-doped sample. Results indicated that there was a higher extent of crystallinity of ZrO2 in n-doped catalyst. However, the characteristic peaks of monoclinic phase ZrO2 were not observed in our case. It is reported that the t-ZrO2 phase is formed at 400 °C, since the synthesis temperature utilised was below this temperature mixed phase ZrO2 (both t-ZrO2 and m-ZrO2) were not expected consistently with the XRD data [29].Redox properties and information concerning metal-support interactions were studied by H2-temperature-programmed-reduction (TPR) analysis. The H2-TPR profile of Ni/Gr catalyst is shown in Fig. 6. Three reduction zones can be observed, at around 280 °C, 330 °C and 530 °C. They all corresponded to the reduction of finely dispersed NiO on the support. Normally, smaller particle size presents reduction zone at lower temperatures [30].In case of Ni/Gr-n, it is hypothesised that when the Ni atom is placed onto the doped support, there is a repulsion between the Ni and the nitrogen dopant, causing the Ni atom to establish an association [14]. The dopant alters the local surface binding configuration, increasing the binding energy, and hence altering the Ni-carbon interaction. This implies that doping might boost the catalyst's durability [14].In the NiZrO2/Gr sample TPR profile, two reduction regions can be observed at 200 °C-400 °C and 550 °C-850 °C. The first peak centred at 250 °C is formed as a consequence of the NiO reduction on the support [30]. NiZr2O/Gr sample has a lower reduction temperature, indicating it is easier to reduce. This might be due to lower particle size of Ni compared to the other samples due to the lower reduction temperatures in the TPR.The second peak was attributed to the reduction of oxygen superficial groups on the support [30]. In general, this peak is formed due to the reduction of finely dispersed NiO on the support, as when ZrO2 is used in Ni-based catalysts it stabilises the cubic structure at high temperatures and improves oxygen storage capacity [31]. Moreover, it is important to mention that hydrogenation of carbon atoms in graphite [32] was taken into consideration at temperatures higher than 800 °C, since methane production has been observed in this temperature range. Two reduction peaks were present in the H2-TPR profile of NiZrO2/Gr-n sample, one at around 300 °C and around 500 °C. They all corresponded to the reduction of finely dispersed NiO on the support [30].As observed in the TPR profiles, the highest reduction temperature of the metal active phase of the catalyst was around 550 °C. Therefore, this was the temperature selected to reduce the catalyst prior to the reaction, since it has been demonstrated in previous publications that the reduction of the active phase of the samples improve the catalytic performance [14]. The success of the reduction prior to the reaction has been demonstrated by the XRD patterns above.Transmission Electron Microscopy (TEM) was used to study the composition and distribution of the different elements in the synthesised reduced samples. The TEM images of fresh Ni/Gr, Ni/ZrO2Gr and Ni/ZrO2Gr-n are presented in Fig. 7. The micrographs clearly showed the exfoliated graphene layers along with zirconia and some nickel particles.In general, a better metal dispersion in the N-doped sample can be observed in the TEM images, corroborating that the presence of nitrogen in the sample can help to obtain a better dispersion of active phases, as discussed in the TPR analysis. Some areas were further analyzed by the corresponding Fast-Fourier transform (FFT) pattern, and they are shown in Fig. 8 A) and B). In general, a relevant search on several zones of the material, showed that the undoped sample barely displayed zones where Ni and ZrO2 were collocated. In contrast, the catalyst NiZrO2-Gr-n did present this Ni-Zr interaction. Moreover, the analysis of the FFT pattern confirmed that ZrO2 was mainly present in its tetragonal phase.Coking and metal sintering of active phase may happen under liquid phase reactions with high pressure [9]. Hence, the XRD patterns of spent catalysts were analysed. As shown in Fig. 5 (A, B, and C), part or all of the metallic nickel were oxidized into NiO during the HDO reaction for Gr supported catalysts. In comparison to the calcined samples, some new diffractions peaks have been detected for the Ni/Gr, Ni/Gr-n, and NiZrO2/Gr catalyst. Diffractions peaks at 37.3◦, 43.3◦ and 63◦ were observed, corresponding to the (111), (200) and (220) planes of the NiO fcc phase respectively, in agreement with JCPDS no. 00–047–1049. This points out oxidation of the Ni metallic phase, which is partly expected given the selected reaction media (H2O). On the other hand, such oxidation phenomenon was not observed for the NiZrO2/Gr-n samples, hence reflecting higher stability due to the nitrogen doping and the ability of Zr to act as a promoter [10,16]. The different width of this peak is related to the different interlayer distance generated during the thermal reducing treatment. Therefore, it can be appreciated that at a higher the reaction temperature, the particle size increases and becomes more amorphous due to sintering of the metal phase.TEM images of the samples (Ni/Gr, Ni/Gr-n, NiZrO2/Gr and NiZrO2/Gr-n) after the catalytic reaction at 300 °C are presented in Fig. 9. By comparing all figures, few particles agglomeration can be seen over NiZrO2/Gr (Fig. 9 C)), but the particle distribution still remained homogenous for all samples. In comparison to the calcined samples, the sintering of the metal particles marginally increased the particle size. This was a result of the reduction procedure and reaction conditions used. Despite the very demanding reaction conditions, this finding demonstrated the structural and morphological stability of the NiZrO2/Gr sample for the hydrothermal upgrading of lignin model compounds.Based on a novel "H2-free" HDO strategy, this research showcases a promising catalytic route for biomass upgrading. More specifically, we show that hydrothermal deoxygenation of guaiacol as a lignin model compound can be performed without external hydrogen input. For the in-situ production of hydrogen coupled to HDO, we propose multicomponent catalysts capable of activating water and facilitating the subsequent HDO reaction. Our catalysts based on Ni nanoparticles supported on N-doped and non-doped graphene decorated with zirconia particles were able to partially deoxygenate the original feedstock effectively. Furthermore, the samples were stable under the reaction conditions (high pressure, high temperature in a hydrothermal medium), as indicated by post-reaction XRD and TEM examination at 250 °C and 300 °C. Our results indicate that the NiZr2O/Gr-n catalyst led to the best results, as the synthesis of cresol as a reaction product was increased, and it exhibited the highest catalytic activity. Cresol can be used as sources for high-value chemical products [33]. Nitrogen doping of the support was shown to improve conversion in all cases and is an effective strategy for promoting activity. In light of these findings, Ni-based catalysts are also a viable option for the HDO reaction of guaiacol due to their comparatively cheap cost and comparable catalytic performance compared to noble metal catalysts.Overall, the catalytic performance of the designed catalysts may be considered moderate in contrast to existing catalysts in the standard HDO when high pressure hydrogen is supplied. Generally, it is necessary to increase overall deoxygenation efficiency by fine-tuning the catalyst composition for the water assisted HDO process. Also performing HDO of chemical intermediates (such as catechol) will aid to gather knowledge of the hydrogen transfer route and reaction pathways in the water assisted HDO process and further guide the catalysts design. Nonetheless, the distinctive benefit of our approach is the absence of external hydrogen input. Hence, despite still on its early development stages, our concept should drive future research efforts to enhance the catalytic formulation and improve performance. In this approach, we demonstrate the crucial role of heterogeneous catalysis in bio-resources upgrading to aid in the transition to a low-carbon economy. S. Parrilla-Lahoz: Writing – original draft, Conceptualization, Visualization, Investigation. W.Jin: Writing – original draft, Conceptualization, Visualization, Investigation. L. Pastor-Pérez: Funding acquisition, Conceptualization, Project administration, Supervision. M.S. Duyar: Funding acquisition, Conceptualization, Project administration, Supervision. L.Martínez Quintana: Experimental, Data curation. A.B. Dongil: Funding acquisition, Conceptualization, Project administration, Supervision. T.R. Reina: Funding acquisition, Conceptualization, Project administration, Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper: Tomas Ramirez Reina reports financial support was provided by Junta de Andalucía Consejería de Educación. Tomas Ramirez Reina reports financial support was provided by Spain Ministry of Science and Innovation.Financial support for this work has been obtained from Junta de Andalucía project P20-00667, co-funded by the European Union FEDER. This work is also sponsored by the Spanish Ministry of Science and Innovation through the projects PID2019-108502RJ-I00 and grant IJC2019- 040560-I both funded by MCIN/AEI/10.13039/501100011033 as well as RYC2018-024387-I funded by MCIN/AEI/10.13039/501100011033 and by ESF Investing in your future. Financial support from the European Commission through the H2020-MSCA-RISE-2020 BIOALL project (Grant Agreement: 101008058) and RYC2020-030626-I (MCIN) and project 20228AT002 (CSIC) is also acknowledged.

Catalytic hydrodeoxygenation (HDO) is a critical technique for upgrading biomass derivatives to deoxygenated fuels or other high-value compounds. Phenol, guaiacol, anisole, p-cresol, m-cresol and vanillin are all monomeric phenolics produced from lignin. Guaiacol is often utilised as a model lignin compound to deduce mechanistic information about the bio-oil upgrading process. Typically, a source of H2 is supplied as reactant for the HDO reaction. However, the H2 supply, due to the high cost of production and additional safety precautions needed for storage and transportation, imposes significant economic infeasibilities on the HDO process's scaling up. We investigated a novel H2-free hydrodeoxygenation (HDO) reaction of guaiacol at low temperatures and pressures, using water as both a reaction medium and hydrogen source. A variety of Ni catalysts supported on zirconia/graphene/with/without nitrogen doping were synthesised and evaluated at 250 °C and 300 °C in a batch reactor, with the goal of performing a multi-step tandem reaction including water splitting followed by HDO. The catalysts were characterised using H2-TPR, XRD, TEM and XPS to better understand the physicochemical properties and their correlation with catalytic performance of the samples in the HDO process. Indeed, our NiZr2O/Gr-n present the best activity/selectivity balance and it is deemed as a promising catalyst to conduct the H2-free HDO reaction. The catalyst reached commendable conversion levels and selectivity to mono-oxygenated compounds considering the very challenging reaction conditions. This innovative HDO approach provides a new avenue for cost-effective biomass upgrading.

The utilization of natural gas has become more challenging as many resources contain high amounts of CO2. The Natuna Sea is the largest natural gas resource in Indonesia, with 46 TSCF proven reserves consisting of 71 mol% CO2 and 28 mol% of CH4 [1]. To date, the natural gas reserve in the Natuna Sea has not been utilized because of its higher CO2 content compared to other natural gas reserves [2,3]. Dry reforming of methane (DRM) is considered a promising technology for converting natural gas reserves into synthesis gas (syngas), as it utilizes methane and CO2 [4]. In the chemical industry, syngas is an important intermediate because it is a source of hydrogen and raw material required to produce various chemical compounds [5–10]. Syngas produced from the DRM is supposed to have a H2/CO ratio close to 1, which is suitable for producing oxygenated chemicals and hydrocarbons through the Fischer-Tropsch process [11–14]. The syngas can then be converted into various products such as wax, olefin, alcohol, and dimethyl ether [15,16]. The main reaction in the DRM is as follows: (1) CH 4 + CO 2 ⇌ 2 CO + 2 H 2 Δ H o 298 = 247 k J m o l As shown in equation (1), DRM is an endothermic reaction; thus, a high operating temperature is required to achieve high equilibrium conversion [17]. According to Wang et al. (1996), 870–1,040 °C is the optimum temperature range to minimize catalyst deactivation caused by carbon formation [18].However, one of the major challenges in commercializing the DRM is severe catalyst deactivation due to sintering and carbon deposition [19]. Sintering occurs in the active phase of the catalyst because of the high operating temperature required to achieve high conversion in the DRM [20]. Meanwhile, solid carbon deposition on the catalyst surface is caused by methane cracking (Eq. (5)) and the Boudouard reaction (Eq. (6)) [21,22]. Thus, it is important to develop a catalyst formulation with good activity and less carbon formation. (2) CH 4 →  C + 2 H 2 Δ H o 298 = 75 k J m o l (3) 2 CO →  C + CO 2 Δ H o 298 = − 172,4 k J m o l Other side reactions that affect the H2/CO ratio and reactant conversion are steam reforming (4), reverse water gas shift (5), and carbon gasification (6) [22,23]. (4) CH 4 + H 2 O ⇌ 2 CO + 3 H 2 Δ H o 298 = 228 k J m o l (5) CO 2 + H 2 ⇌  CO + H 2 O Δ H o 298 = 41 k J m o l (6) C + H 2 O ⇌  CO + H 2 Δ H o 298 = 131 k J m o l The reverse water gas shift reaction, which simultaneously occurs with the DRM, causes the H2/CO ratio to decrease, and the CO2 conversion becomes higher than the CH4 conversion [24].Previous researchers have investigated the utilization of several noble and non-noble metals such as Pd, Pt, Ru, Co, and Ni for use as active phase in catalysts [25–31]. However, the use of noble metals as the active phase is not economically attractive because of their scarcity and high price [32]. Nickel-based catalysts are more appealing because nickel is more abundant in nature and is cheaper [33]. However, nickel is rapidly deactivated because of the coking and sintering phenomena [34,35].The support material also plays a vital role in catalyst activity and carbon resistance [36]. Gamma-alumina (γ-Al2O3) is commonly examined as a catalyst support, but it can deteriorate at high temperatures and can undergo a phase change to α-Al2O3 [37]. Zeolite and other mesoporous materials such as KIT-6, SBA-15, and MCM-41 have also been developed as catalyst supports because of their high surface area and good thermal stability [38–41]. MCM-41 is a mesoporous material with a pore diameter of 2–4 nm, uniform 2-dimensional hexagonal structure, and surface area of ∼1000 m2/g to allow for high metal dispersion [27,42]. Several studies have shown that a nickel-based catalyst supported by MCM-41 needs further improvement to enhance its activity and suppress coke formation. Fakeeha et al. (2019) reported that an MCM-41-supported catalyst yielded a CH4 conversion of 48%. Furthermore, CO2-Temperature Programmed Desorption (CO2-TPD) showed that the number of base sites in the Ni/MCM-41 catalyst was smaller than that in other materials (Al2O3, SiO2, and SBA-15). Thus, it can be concluded that the adsorption of CO2 during the reaction is inhibited because of the fewer base sites. Hence, modification is needed to add some base sites to the Ni/MCM-41 catalyst so that its activity and stability could be enhanced. The number of base sites of the catalyst can be increased by adding a promoter [43].Ibrahim et al. (2018) compared the effects of adding Cs, Ce, Gd, Sc, and Ga promoters to the Ni/MCM-41 catalyst. The experiment showed that catalysts with Ga, Gd, and Ce promoters yielded higher CH4 and CO2 conversions due to the additional metal active sites from these promoters. Meanwhile, Thermogravimetric Analysis (TGA) revealed that the Ni/MCM-41 catalyst with a promoter also showed less carbon formation than did the catalyst without any promoter [30]. Finally, Al-fatesh et al. (2019) examined the effect of adding a Gd promoter to the Ni/MCM-41 catalyst. Research has shown that Gd-promoted catalysts yield 80.6% CH4 conversion and 87.9% CO2 conversion, with negligible carbon formation [44].Another potential material that can be used to enhance catalyst activity and stability is the base promoter [45]. According to Nikolaos and Thessaloniki (2018), the base promoter can enhance the adsorption of CO2 and suppress the sintering of the catalyst [46]. Jeong et al. (2006) examined the effect of adding Mg, Mn, K, and Ca as promoters to Ni/HY catalysts. The Ni–Mg/HY catalyst activity test resulted in more than 85% CH4 conversion without catalyst deactivation for 72 h. This was because MgOx generated from Mg covered the nickel surface to prevent the agglomeration of nickel [47]. Horiuchi et al. (1996) found that the addition of Mg, K, Ca, and Na to a Ni/γ-Al2O3 catalyst enhanced the adsorption of CO2, thereby reducing the amount of carbon formation [48]. The novelty of this research is the combination of MCM-41 as catalyst support and Mg, Ca, Na, and K as catalyst promoter for conducting DRM. This study aims to examine the effect of several base promoters (Mg, Ca, Na, and K) on the activity of MCM-41-supported nickel catalysts for the DRM and on the amount of carbon deposition.MCM-41 powder was purchased from XFNano Material, China (surface area: 1014.9 m2/g). Ni(NO3)2.6H2O (Merck, ∼99%) was dissolved in water and then mixed with either Mg(NO3)2·6H2O (Merck, 99.9%), Ca(NO3)2·4H2O (Merck, 99.9%), or KNO3 (Merck, 99.9%), or NaNO3 (Merck, 99.9%) depending on the type of catalyst promoter. The amount of metal promoter loading was 1 wt%. For instance, to prepare 3 g of 5 wt% Ni-1 wt% Mg/MCM-41 catalyst, 0.74 g Ni(NO3)2·6H2O was dissolved in 2.93 mL of water and mixed with 0.32 g Mg(NO3)2·6H2O. After impregnation, the catalysts were dried at room temperature and calcined in air at 700 °C (heating rate 5  °C/min) for 4 h.X-ray diffraction (XRD) patterns were acquired using a Bruker D8 Advance with Cu Kα1 (λ = 1.5406 Å) radiation at 40 kV and 35 mA, meanwhile low angle XRD patterns were obtained using X-Ray Scattering Shimadzu SAG-6 (λ = 1.54 Å). N2 physisorption was conducted using a Micromeritics Tristar II instrument. Before the measurement, the samples were degassed at 250 °C for 3 h. A Micromeritics Chemisorb 2750 instrument was used to carry out CO2-TPD analysis. For TPD measurements, 150 mg of the sample was heated to 200 °C and then flushed with helium for 1 h. CO2 adsorption was conducted at 50 °C for 30 min using a 40 mL/min mixture gas of 5% CO2/95% He. H2-Temperature Programmed Reduction (H2-TPR) was conducted using Micromeritics Chemisorb 1750. Prior to the measurement, the samples were heated to 150 °C for 1 h using helium at a flow rate of 40 mL/min. Afterward, the samples were cooled to room temperature. The line system purging was performed for 30 min, and then the samples were reduced at 800 °C (heating rate 10  °C/min) with 40 mL/min of 5% H2/95% Ar. TGA analysis was carried out using a TA-60WS Shimadzu instrument to measure the amount of carbon deposition. For this measurement, 15 mg of the sample was heated to 900 °C at a heating rate of 20  °C/min.The activity test was conducted in a fixed-bed reactor at atmospheric pressure and 700 °C. First, 60 mg of the catalyst was loaded into the reactor, and the system was purged with nitrogen at 400 °C for 1 h. Afterward, the catalyst was reduced in situ at 700 °C and at atmospheric pressure for 1 h, and then, the system was flushed with nitrogen. Feed consisting of mixed gas was introduced at a flow rate of 60 mL/min (Gas Hourly Space Velocity = 60,000 mL/g-cat.h) and at a CH4: CO2: N2 ratio of 1:1:1. The product from the reaction was analyzed by Shimadzu Gas Chromatography with PQ and MS columns. The catalyst activity and stability were investigated at 700 °C, atmospheric pressure, and time on stream (TOS) of 240 min. A schematic of the equipment used for activity testing is shown in Fig. 1 .The conversion of CH4 and CO2, as well as selectivity of H2 and CO, and H2/CO ratio were calculated based on the following formulas. (7) CH 4  Conversion = F CH 4 , in − F CH 4 ,  out F CH 4 , in x  100 % (8) CO 2  Conversion = F CO 2 , in − F CO 2 ,  out F CO 2 , in x  100 % (9) Yield H 2 = F H 2 ,  out 2 F CH 4 , in x  100 % (10) Yield CO = F CO ,  out F CH 4 , in + F CO 2 , in x  100 % (11) H 2 CO  Ratio = F H 2 ,  out F CO ,  out The synthesized catalysts were characterized by XRD analysis to determine their crystalline phase. Fig. 2 shows the XRD pattern of the catalysts.It can be seen from Fig. 2a that diffraction peaks appear at 2θ of 37.3°, 43.3°, 62.8°, 75.4°, and 79.4°. According to JCPDS no. 47-1049, those peaks correspond to the crystalline NiO phase. The NiO phases in these catalysts are (1 1 1), (2 0 0), (2 2 0), (3 1 1), and (2 2 2) [49]. The crystallite size of NiO was determined using the Scherrer equation. The crystallite sizes for 5 wt% Ni/MCM-41, 5 wt% Ni–Mg/MCM-41, 5 wt% Ni–Ca/MCM-41, 5 wt% Ni–Na/MCM- 41, and 5 wt% Ni–K/MCM-41 were 5.507, 6.288, 6.645, 4.190, and 4.901 nm, respectively. Furthermore, peaks at 2θ = 23° showed the amorphous phase of MCM-41 [50,51]. The addition of a base promoter other than Na and K does not result in a significant difference in the crystalline structure, which means the promoter oxide is well-dispersed in the catalyst [51]. However, as shown in Fig. 2, there was peak broadening in the case of Ni–Na/MCM-41 and Ni–K/MCM-41 because of the formation of small crystalline particles in the catalysts, as described by Contreras and Fuentes (2012) [52].Moreover, it can be inferred from Fig. 2b that in calcined MCM-41 sample, there is peak ranging from 2θ = 2–3° that indicates the diffraction patterns of typical MCM-41 material [57]. However, in other catalysts, mainly in Ni–Na/MCM-41 and Ni–K/MCM-41, the diffraction peak is not clearly seen. Since the diffraction peak of MCM-41 is not observed, it is indicated that the structure of MCM-41 is collapsed.N2 physisorption was performed to determine surface area and pore size. As shown in Fig. 3 , the catalysts showed type IV isotherms, and capillary condensation occurred in the high P/P0 region [13]. In addition, the catalysts displayed an H1-type hysteresis loop and exhibited cylindrical geometry with uniform pore size [44]. As shown in Fig. 3, there was a significant amount of nitrogen uptake in the relative pressure range of 0.3–0.4 in MCM-41 support. This presented an indication of micropores and mesopores in MCM-41 [53]. The surface area and pore volume of the catalysts are shown in Table 1 . Additionally, Table 1 shows there was a significant decrease in the surface area, particularly for the Ni–Na/MCM-41 and Ni–K/MCM-41 catalysts. This phenomenon was mainly caused by sintering, which causes the catalyst support to collapse, thereby causing a dramatic decrease in the surface area [54]. Sintering can be caused by the presence of water inside the pores of the support [54].Furthermore, pore size distribution of the MCM-41-based catalysts is shown in Fig. 4 . It can be seen that Ni/MCM-41 has pore diameter between 1.5 and 8.6 nm, with average pore diameter of 2.64 nm. Moreover, Ni–Mg/MCM-41 has pore diameter between 1.5 and 58.76 nm average pore diameter: 5.98 nm and Ni–Ca/MCM-41 has pore diameter between 2.7 and 74 nm (average pore diameter: 9.65 nm). On the other hand, Ni–Na/MCM-41 has pore diameter between 14.76 and 147.6 nm (average pore diameter: 13.4 nm) and Ni–K/MCM-41 has pore diameter between 12.6 and 79.9 nm (average pore diameter: 6.69 nm).H2-TPR analysis was conducted to observe the reducibility of the catalyst. As shown in Fig. 5 , all catalysts show peaks in the high-temperature regions. Peaks that appeared at 350–500 °C were related to the reduction of Ni2+ to Ni0, while peaks at 500–800 °C, indicate a strong interaction between NiO and the catalyst support [13]. The addition of the base promoter affected the reducibility of the catalyst. Base promoter replenishment increases total H2 consumption, as shown in Table 2 . The addition of the base promoter affected the reducibility of the catalyst. Base promoter replenishment increases total H2 consumption, as shown in Table 2. The base promoter formed metal oxide on the catalyst surface, such as MgOx, that made the active phase better dispersed on the catalyst support. Thus, it made the amount of H2 consumption to reduce the catalysts is higher than the catalyst without promoter.The CO2-TPD profiles depicting the basicity measurements of the catalysts are shown in Fig. 6 . From CO2-TPD analysis, the capacity of CO2 adsorption at the catalyst surface can be measured. Strong basic site is indicated by peaks at 50oC–128 °C, while medium basic site is showed by peaks at 220oC–360 °C, and peaks strong basic site is declared at 580.7–780 °C [51]. As can be seen from Fig. 6, all catalysts showed peaks in the temperature range 100–150 °C, which indicating weak base sites on the MCM-41-based catalyst. Furthermore, the addition of Mg and Ca promoters produced higher peaks in regions 100–150 °C and 250–400 °C; thus, it is indicated that the addition of Mg and Ca promoters leads to higher CO2 adsorption capacity of the catalysts. The higher CO2 conversion of Ni–Mg/MCM-41 and Ni–Ca/MCM-41 during activity testing is due to the basicity aspect. Furthermore, it also can be seen that Ni–Na/MCM-41 catalyst does not show any peak in temperature range 350–800 °C. This finding is similar to that reported by Lovell et al. (2014), for incorporating Na into Ni/MCM-41 catalyst does not promote significant increase of basicity [49]. On the other hand, Ni–K/MCM-41 showed the same trend of CO2-TPD to other catalyst, but with different intensity.An activity test was conducted to observe the impact of catalyst type on conversion, selectivity, and yield of dry methane reforming. The activity test was conducted using a fixed bed reactor, with a time on stream of 240 min and at a temperature of 700 °C. The feed was a gas mixture consisting of CH4, CO2, and N2 at a ratio of 1:1:1. In this experiment, nitrogen was used as the internal standard and diluent. The activities of the synthesized catalysts were compared to those of commercial catalysts (methanation and steam reforming catalysts from fertilizer plants) in the DRM. The methanation catalyst consisted of 27.95 wt% nickel, 58.86 wt% alumina, 12.10 wt% calcium, 0.99 wt% silica, and 0.10 wt% ferrous. Meanwhile, the steam reforming catalyst is consisted of nickel and alumina. The results of the activity tests of the catalysts are shown in Fig. 7 .The methanation catalyst showed conversion decline during the first 30 min, and then there was no conversion of CH4 and CO2 in the 150th min. Therefore, catalyst deactivation, as shown by the methanation catalyst, is due to methane decomposition and Boudouard reaction. Fig. 8 and Table 3 shows the TGA results for the amount of carbon deposition that leads to catalyst deactivation. Based on Fig. 8, the use of a methanation catalyst for the DRM resulted in carbon deposition as much as 41 wt%. Furthermore, it can be concluded that the methanation catalyst leads to the reaction towards carbon formation. On the other hand, the MCM-41-based catalysts had less carbon deposition than the other catalysts.Moreover, steam reforming catalysts yielded CH4 and CO2 conversions of as much as 85% and 64%, respectively. Although the catalyst generated high reactant conversion, it also yields 30 wt% carbon deposition, and as can be seen from Fig. 7, the H2 yield constantly decreased. Therefore, it is indicated that the steam reforming catalyst also directs the reaction toward carbon formation.From Fig. 7, it can also be seen that all synthesized catalysts generated good stability for 240 min. No significant decrease in reactant conversion was observed during the reaction. Based on the thermodynamic analysis, the maximum conversion that can be obtained at 700 °C and atmospheric pressure for CH4 is 91.5%; meanwhile, the maximum conversion of CO2 is 66.3% [55].The catalyst that showed the best performance in converting CH4 was 5 wt% Ni–Mg/MCM-41, which yielded 72% conversion, followed by 5 wt% Ni–Ca/MCM-41, which yielded 69% conversion. These catalysts gave higher CH4 conversions than the unpromoted catalyst (62%). The good activity of the 5 wt% Ni–Mg/MCM-41 catalyst can be explained by H2-TPR characterization results, where the 5 wt% Ni–Mg/MCM-41 catalyst generates 3.146 mmol/g H2 consumption. According to Ibrahim et al. (2018), high total hydrogen consumption is related to the number of active sites in the catalyst for the DRM. The more active sites in the catalyst, the higher is the conversion [51].The order of the synthesized catalysts that yielded the highest conversion of CO2 was 5 wt% Ni–Ca/MCM-41 (55%) > 5 wt% Ni–Mg/MCM-41 (54%) > 5 wt% Ni/MCM-41 (52%) > 5 wt% Ni–K/MCM-41 (44%) > 5 wt% Ni–Na/MCM-41 (35%). The CO2 conversion capability of the catalyst is related to the basicity of the catalyst, as a catalyst with stronger basicity can adsorb more CO2, resulting in higher CO2 conversion. Based on Fig. 6, both Ni–K/MCM-41 and Ni–Na/MCM-41 has weak and medium base sites, and thus both catalysts yield lower CO2 conversion than the other synthesized catalysts. Furthermore, sintering phenomena on the catalysts that is indicated from N2 Physisorption analysis (mainly Ni–K/MCM-41 and Ni–Na/MCM-41) reduces the number of active sites in the catalysts and makes the support structure to collapse; hence, the activity of the catalyst is low. It can also be concluded that catalysts with Mg and Ca addition as promoters give a higher conversion of CH4 and CO2 than the 5 wt% Ni/MCM-41 catalyst due to its CO2 adsorption capacity that is indicated from CO2-TPD and also due to its well-dispersed active phase that is indicated from H2-TPR analysis. Moreover, in this study, the conversion of CO2 is higher than CH4 conversion. It is predicted that the presence of water that is produced from reverse water gas shift reaction (CO2+H2 ⇌ CO + H2O) inside MCM-41 pores shifts the reaction towards steam reforming reaction (CH4+H2O ⇌ 2CO+3H2), so that the CH4 conversion is higher than the CO2 conversion. Furthermore, CH4 decomposition reaction occurs more dominantly at higher temperature, so that the amount of CH4 conversion is getting higher [30]. Fig. 9 displays the product yield and H2/CO ratio. From Fig. 9a, it can be seen that 5 wt% Ni–Mg/MCM-41 and 5 wt% Ni–Ca/MCM-41 generated H2 yields of 45% and 40%, respectively. Those results are related to CH4 conversion obtained by both catalysts, as the higher CH4 conversion, the more H2 production. Furthermore, from Fig. 9b it can be concluded that 5 wt% Ni–Mg/MCM-41 and 5 wt% Ni–Ca/MCM-41 also produced high, stable CO yield. Combination of H2 and CO yield resulting in H2/CO ratio, which can be seen in Fig. 9c. From Fig. 9c, 5 wt% Ni–Mg/MCM-41 and 5 wt% Ni–Ca/MCM-41 generated H2/CO ratios of 0.83 and 0.78, respectively [56]. On the other hand, the unpromoted 5 wt% Ni/MCM-41 catalyst generated an H2/CO ratio of 0.88. The result is almost the same as that obtained by Amin et al. (2013), where the H2/CO value obtained was 0.83. Thus, all synthesized catalysts had a ratio of H2/CO less than 1. This indicates that the reverse water gas shift reaction (CO2+H2⇌CO + H2O) occurred dominantly [56]. If the reverse water gas shift reaction is dominant, H2 is consumed as a reactant; meanwhile, CO is generated as a product; therefore, CO is higher than H2, causing an H2/CO ratio of less than 1.The activity testing result in this study showed almost the same trend with Jeong et al. (2006) that investigating the effect of Mg, Mn, K, and Ca as promoter to Ni/HY catalyst. The result was Ni–Mg/HY gave the best performance (with CH4 conversion more than 85%, H2/CO ratio 0.97, and carbon deposition 18 wt%), followed by Ni–Mn/HY, Ni–Ca/HY, Ni/HY, and the last was Ni–K/HY. The superior performance of Ni–Mg/HY was caused by the presence of MgOx species covering the nickel surface, so that it prevented the nickel to agglomerate and the conversion of CH4 could be maximized [47].In this experiment, 5 wt% Ni–Mg/MCM-41 was tested at two different values of WHSV: 60,000 mL/g-hour and 72,000 mL/g-hour. The results of the catalyst testing are shown in Fig. 10 . The catalyst tested at WHSV = 60.000 mL/g-h yielded higher values for both CH4 and CO2 conversions. Because of the lower WHSV, the contact time between the reactants and catalyst increases, so more reactants are converted. The experiment conducted by Ibrahim et al. (2018) also generated the same result, as 5 wt% Ni+1 wt% Ga/MCM-41 tested at WHSV = 39,000 mL g−1 h−1 yields higher conversion than the catalyst tested at WHSV = 78,000 mL g−1 h−1 [51].Ni/MCM-41, Ni–Mg/MCM-41, Ni–Ca/MCM-41, Ni–Na/MCM-41, and Ni–K/MCM-41 catalysts were successfully synthesized. XRD characterization showed that the size of NiO crystallite is 4,19 nm–6,65 nm. N2 physisorption results showed a decrease in the surface area and pore volume due to pore blockage. The sharp drop that occurred in Ni–Na/MCM-41 and Ni–K/MCM-41 was caused by severe sintering, so that the support was collapsed. H2-TPR results showed that the addition of base promoters strengthened the interaction between NiO and MCM-41. CO2-TPD results showed that the addition of Mg and Ca promoters increased the CO2 adsorption capacity of the catalyst. Furthermore, activity tests on Ni/MCM-41, Ni–Mg/MCM-41, Ni–Ca/MCM-41, Ni–Na/MCM-41, and Ni–K/MCM-41 yielded CH4 conversions of 62%, 72%, 69%, 36%, and 46%, respectively, with corresponding CO2 conversion rates of 52%, 54%, 55%, 35%, and 44%. Ni–Mg/MCM-41 and Ni–Ca/MCM-41 produced the highest H2 and CO yield. All catalysts yielded good stability for 240 min. On the other hand, the commercial catalyst for methanation showed activity for only 140 min, with carbon deposition as much as 41 wt%. This phenomenon indicates that the methanation catalyst shifts the reaction toward carbon formation. Steam reforming commercial catalysts yield CH4 and CO2 conversions of as much as 85% and 65%, respectively; however, there was a decrease in the H2 yield with time, and the carbon deposition is 31 wt%. This indicates that the steam reforming catalyst also shifts the reaction toward carbon formation. In this study, the effect of WHSV on the activity of the 5% Ni–Mg/MCM-41 catalyst was investigated. The catalyst tested at WHSV = 60.000 mL g−1 h−1 showed higher activity than the catalyst tested at WHSV = 72.000 mL g−1. h−1. A catalyst with a lower WHSV has more space time, and thus, it leads to higher conversion.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the Ministry of Education, Culture, Research, and Technology of the Republic of Indonesia through the research grant of Konsorsium Riset Unggulan Perguruan Tinggi 2021 [grant number 304/IT1.B07.1/SPP-LPPM/VII/2021]; and partially supported by Indonesia endowment fund for education (LPDP), the Ministry of Finance of Indonesia [grant number RISPRO/KI/BI/KOM/II/16507/I/2020].

Dry reforming of methane (DRM) is considered a promising reforming technology that converts natural gas in the Natuna Sea into synthesis gas, which can be further utilized to produce beneficial chemicals such as olefins, alcohols, and liquid hydrocarbons. However, the challenges in commercializing the DRM process are carbon deposition and sintering of the catalyst at high temperatures, because of which the catalyst is easily deactivated. This study aimed to test the activity and stability of MCM-41-based catalysts for the DRM; determine the effect of promoter type on the activity and stability of MCM-41-based catalysts; and determine the effect of base promoter addition on the amount of carbon deposition. MCM-41-based catalysts were synthesized using incipient wetness impregnation method. XRD, N2 Physisorption, H2-TPR, CO2-TPD, and TGA analysis were conducted to determine the physicochemical properties of the catalysts. The catalysts activity was tested in a fixed-bed reactor, under atmospheric pressure at 700 °C. Overall, all catalysts exhibited good stability for 240 min. Moreover, catalysts with Mg and Ca promoters showed the highest CH4 and CO2 conversion among all catalysts. Ni–Mg/MCM-41 catalyst yielded 72% CH4 conversion and 54% CO2 conversion, meanwhile Ni–Ca/MCM-41 yielded 69% CH4 conversion and 55% CO2 conversion. Furthermore, MCM-41-based catalysts with base promoter produced small amount of carbon deposition.

Reducing greenhouse gas emissions and the consumption of fossil fuels are necessary to limit the ongoing climate change. Hence, renewable energy systems such as wind or solar power plants are a suitable solution to provide sustainable energy. One drawback of these technologies is the weather dependency. To overcome the weather dependency, energy storages and high efficient on demand power supply is needed. Hydrogen seems to be a promising energy storage but higher volumetric energy densities are often advantageous. Therefore, hydrogen based energy storage media with high volumetric energy density such as ammonia and hydrocarbons are used. These fuels can be later reformed into hydrogen rich gas compositions. Today, most reforming reactors are used to gain hydrogen from different fuels or hydrogen carriers. For example, the reforming of carbon-based fuels now accounts for more than three quarters of global H2 production [1]. However, the reformed hydrogen rich gases have to be converted into heat or electric energy. One possibility for high efficient on demand heat and electric energy supply are solid oxide fuel cell (SOFC) systems [2]. In previous research it was found, that SOFC systems are able to use a variety of hydrogen based storage media such as (i) ammonia [3], (ii) methane [4,5], (iii) carbon monoxide [6], (iv) gasified biomass [7–9] or (v) synthetic liquid hydrocarbons [10–12]. These fuels can be produced by supplying solid oxide electrolysis cells and fuel post processing methods with green energy, water and CO2 [13,14]. Further, the reversible operation of solid oxide cells, operation in electrolysis and fuel cell mode, were successfully applied [15,16]. However, if SOFC-systems are fuelled with carbonaceous fuels, a fuel reformer upstream of the SOFC can be useful. Reforming of methane upstream of the SOFC for example, reduces thermal stresses within SOFCs if compared to direct internal reformed methane [17]. Direct internal reforming refers to steam reforming taking place at the SOFCs anode. The reduction of thermal stresses along SOFCs can prevent leakages and breaking of the few hundred μ m thick cells. Thermal stresses are induced by endothermic steam reforming of hydrocarbons at the SOFC gas inlet and by exothermic oxidation reactions along the SOFC [17]. Further, direct internal reforming of other fuels, especially long chain hydrocarbons can lead to rapid degradation and failure of the SOFC [12,18]. Due to thermal stresses and rapid degradation, external reforming of hydrocarbons upstream of the SOFC is often necessary. Most used external reforming processes are, (i) steam reforming [19], (ii) partial oxidation [20], (iii) auto-thermal reforming [19,21] and (iv) dry reforming [22]. However, inadequate operating conditions or local effects caused by reaction kinetics, non-uniform temperature distribution and non-uniform fuel mixing can lead to degradation of catalysts even at safe operating conditions [23–25]. The degradation is caused by reducing chemical reactive surfaces. Latter leads to a reduction of the reactivity of the catalyst and lower fuel conversion rates. Degradation can even lead to a total system failure and catalyst destruction. Degradation mechanisms of Ni catalysts are coking [26,27], oxidation of Ni [26], poisoning due to sulphur [28–30], chlorine and other impurities [9,29,30]. If such degradation effects are detected at an early stage, countermeasures can be applied. These countermeasures can limit further damage of the catalyst or even reverse degradation. Still, no direct degradation online monitoring of Ni-based catalysts based on electrochemical impedance spectroscopy (EIS) measurements is known in the field of reforming carbonaceous gases. One study used a radio frequency-based-method for in situ coke detection [31]. However, established practices to detect degradation are measuring temperature profiles and gas compositions along fixed bed reforming reactors [28,32]. Several studies are available that address the field of sensor development to detect coking of reforming catalysts. The sensors which are based on the impedance measurement method consist of a catalytic active material which changes its conductivity due to the amount of carbon loaded on them. The catalytic active material is either especially manufactured as sensor [33,34] or commercial catalysts are used as active sensor material [35,36]. Both types of sensors, however, measure coking of the catalytic active sensor material instead coking of the used catalysts. Using sensors instead of direct monitoring of the catalyst does not allow to detect changes within the catalyst directly. In addition, the placement of the sensor is very important and often difficult. Especially for reforming processes like auto-thermal reforming, where different reactions (partial oxidation, steam reforming and water gas shift reaction) and degradation mechanisms can occur within short ranges. In this study, we applied EIS based online monitoring on operating commercial Ni catalysts. The EIS measurements are applied to monitor the catalysts during heat up, reduction and operation. Within this study, the catalysts are contacted in three different ways. This is done to determine influences and characteristics of the contacting method on the measurement results. To validate the EIS measurement results, scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), Raman spectroscopy and X-ray diffraction (XRD) measurements are conducted post mortem. Considering this, the main goal of this study is to find an additional online monitoring method which expands the knowledge of the behaviour of catalysts. Results achieved, such as observation of NiO reduction or carbon deposition, are first steps towards EIS based online monitoring of the state of health of catalysts. Furthermore, additional information about the processes occurring in the microstructure of the catalyst could be obtained in situ. Using EIS based measurements to monitor catalysts is also advantageous for commercial SOFC-reformer applications since SOFCs are already often monitored using EIS measurements [16,37–39]. As a result, only one device would be necessary to monitor the chemical active parts of SOFC-reformer systems. The results of the impedance measurements could be implemented in diagnostic algorithms as proposed in [40].In this section, the experimental setup, test preparation and the testing procedure are explained. The first part of this section is about preparing the commercial Ni catalysts to apply electrical measurement methods. Within the second part, the testing procedure is described and the third part is about the used test rig.Commercial available Ni based catalysts for steam reforming are used within this work. The catalyst consists of K promoted NiO on a Calcium-Aluminate support and is commercially available. To conduct electrical measurements, the catalysts have to be contacted to wires. We used silver wires with a diameter of 1 mm to connect the measurement device and the contacted catalyst. To contact the catalyst with silver wires, three different methods are used to investigate the influence of contacting on the measurement results, see Fig. 1.The first contacting method, contacting with bare silver wires, is based on application of silver wires with a diameter of 3 mm. The ends of the silver wires are sanded to fit tight in the catalyst holes. After sanding, the prepared silver wires are stuck in opposite holes of the catalyst, see Fig. 1(a). It was expected, that the different thermal expansion coefficients result in a sufficient contacting at reforming temperature.For the second contacting method, silver wires with silver ink, silver wires with a diameter of 2 . 5 mm are used. In addition, 50 mg of silver ink is applied to contact each silver wire with the catalyst. The silver ink is applied inside the holes, where the silver wires are placed, see Fig. 1(b). After assembling the catalyst contacting, the silver ink is dried for 1 h at  200 ∘ C in air. The silver ink is purchased as Silver SOFC Ink from FuelCellStore© and consists of 70 wt% Silver and 30 wt% Diethyl Glycol Ether Acetate.Connecting silver wires with Ni mesh and silver ink to the catalyst is the third contacting method. The silver ink is applied in the central hole of the catalyst, on its outer cylindrical surface and between the silver wires and Ni meshes, see Fig. 1(c). The silver wire, which is fixed on the outer Ni mesh has a diameter of 1 mm . One end of this wire is flattened and laid between the catalyst and the Ni mesh. The silver wire placed inside the central hole has a diameter of 3 mm . The contacted catalysts are dried for 1 h at  200 ∘ C in air.In this work, the measurements of four test specimens (same Ni catalysts but different electrical contacting) are presented. They are chosen to give an impression of the impact on the measurement results and the scattering within a contacting method. The four test specimens include one specimen contacted with bare silver wires, two specimens contacted with silver wires and silver ink and one test specimen contacted with silver wires with Ni mesh and silver ink, see Table 1. After preparing the catalysts as described in Section 2.1, the catalysts are built into the reformer test rig. The catalysts are then heated up to target temperature (700 °C or 750 °C) under a constant N2 flow rate of 2 slpm (standard litre per minute). When the target temperature is reached, the reduction process is started. The catalysts are reduced in 0 . 5 slpm H2 and 2 slpm N2 for 12 h . After the reduction, the H2 flow is turned off. The catalysts are then loaded with coke by applying an automated 0 . 4 slpm CH4 pulse for 20 s , see Fig. 2. The described CH4 pulse is repeated several times and between each CH4 pulse, an EIS measurement is conducted in N2 atmosphere. To avoid oxidation, the catalysts are supplied with 2 slpm N2 during the whole testing phase. The used catalyst test rig for testing EIS measurements of Ni based catalysts consists of five main parts: a gas mixing unit, a tube furnace, a catalyst within a reformer, an impedance measurement device and a continuous off-gas analyser, see Fig. 3. The gas mixing unit is used to supply the catalyst with the required fuel mixture. For the tests within this work, only methane, nitrogen and hydrogen are used. The volume flow rates of each gas are controlled by mass flow controllers (MFCs), which are purchased by Voegtlin Instruments GmbH© [41].The reformer is placed downstream of the gas mixing unit within a tube furnace. The tube furnace is purchased from Carbolite Gero GmbH & Co. KG© (Carbolite Gero CTF 12/75/700) [42]. The reformer consists of a stainless steel tube with a flange at the outlet. An alumina tube is placed within the reformer to isolate the catalyst and wires against the metal parts and prevent short circuits. The silver wires to contact the catalyst are led out radial between two flanges at the outlet of the reformer. To avoid an electrical contact between the silver wires and the flange, isolation material is used. The wires are isolated using glass fabric adhesive tape purchased from HORST©. In addition, isolating and slightly flexible sealing made out of mica or aramid is placed between the flanges. Upstream of the catalyst, a thermocouple type K (TCat) is placed to monitor the gas temperature.A gas analyser from ABB© is located downstream of the reformer to monitor the off-gas composition. The gas analyser consists of a sample gas cooling unit (SCC-C), a sample gas feed unit (SCC-F) and a gas analyser module (AO2020) [43]. The impedance was measured using either an impedance analyser from BioLogic© (SP-150) [44] or Gamry© (Reference 3000) [45]. The impedance measurements were carried out in potentiostatic mode with the parameters listed in Table 2. The parameters were adapted for each of the three phases of the test procedure according to the expected rate of change. Due to fast changes during the heat up phase, the impedance is only measured at one frequency.All microscopic investigations were conducted by the Austrian Centre for Electron Microscopy and Nanoanalysis using a Zeiss Ultra 55.Within this section, the results observed during the catalysts monitoring, as well as post mortem analysis results are shown. Since the (i) heat up, (ii) reduction and (iii) carbon deposition of catalysts might have an impact on the catalyst performance, all of these processes were investigated. The post mortem analysis is shown in the first part to interpret the EIS measurement results. The results of the EIS measurements during the heat up process are discussed in the second part of this section. The reduction process is analysed in the third part and online monitoring of catalyst degradation due to carbon loading is presented in the fourth part.To gain further information and be able to interpret the EIS measurement results in a correct manner, scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), Raman spectroscopy and X-ray diffraction (XRD) is used. SEM in combination with EDS is used to investigate microstructural changes in the cross section of the catalyst. Raman spectroscopy identified the structure and relative quantity of carbon. XRD was used to gain information about the relative quantity of NiO and Ni on the catalysts surface.The applied test procedure which is described in Section 2.2 leads to microstructural changes of the used Ni based catalysts. These changes are proofed by post mortem analysis and are used to interpret the EIS measurement results in the following sections. A sketch of the microstructural changes which are observed during the heat up, reduction and carbon deposition process is shown in Fig. 4. The catalyst, in its original state, consists of NiO on an alumina carrier. Additional, nano crystalline carbon and calcium aluminate is found, see Fig. 5(a). In Fig. 5, peaks P1 and P2 indicate calcium aluminate. Peaks P3, P4, P5 and P6 indicate graphite and peak P7 indicates O–H groups. The Raman shift measurements revealed, that carbon is present on the original catalyst and the present carbon has a nano crystalline appearance. This nano crystalline carbon is not present on catalysts which were heated up in N2 atmosphere. Hence, it is assumed that nano crystalline carbon from the original state mainly reacts with NiO to Ni, CO and CO2, see Fig. 4(a). To determine the reduction state of the catalyst after the heat up and reduction process, see Figs. 4(a) and 4(b), the relative surface quantities of NiO and Ni are measured with XRD. The XRD results of four different reduction states are shown in Fig. 6. In Fig. 6, peaks P1, P2, P5 and P6 indicate NiO and peaks P3, P4 and P7 indicate Ni. Fig. 6a shows the original catalyst where all Ni is mainly present in form of NiO. Fig. 6b shows a mixture of Ni and NiO although no reducing medium was used. The catalyst was only heated up and cooled down in N2 atmosphere. The presence of Ni on this catalyst indicates the self reducing effect caused by the reaction of nano crystalline carbon with NiO. Fig. 6c shows a completely reduced catalyst without the presence of NiO at its surface. The reduction procedure used for the catalyst in Fig. 6c is the standard reduction procedure within this work as described in Section 2.2. Another reduction procedure with a lower H2 flow rate ( 0 . 2 slpm H2) was tested and is shown in Fig. 6d. This reduction procedure shows a lower ratio of NiO to Ni than in Fig. 6b but a higher NiO to Ni ratio than in Fig. 6c. It is visible, that the reduction process is not completed using 0 . 2 slpm H2 in 2 slpm N2 for 12 h , see Fig. 6d. The last step of the test procedure is the carbon deposition process, as described in Section 2.2. The presence of carbon after the carbon deposition process and the absence of carbon after the reduction process are proven by Raman shift spectra, see Fig. 5. In Fig. 7, SEM and EDS of the original state of the catalyst and the state after the carbon deposition process are shown. The brighter parts of the EDS image (Fig. 7(d)) correlate with the brighter parts of the SEM image (Fig. 7(c)). The brighter parts of the SEM image are Ni and the brighter parts of the EDS image are carbon depositions, compare Figs. 7(c) and 7(d). Hence, carbon seems to form at the catalytic active parts of the catalyst. Formation of solid carbon on the catalytic active parts of the catalyst leads to performance deterioration due to less catalytic active surface area. The knowledge of the microstructure of catalysts completes the post mortem analysis. Hence, the cross-sections of catalysts at different states are compared in Fig. 8. The original state of the catalyst shows a layer of NiO (bright part in Fig. 8(a)) at the surface of the catalyst. The occurrence of NiO is confirmed by EDS and XRD. NiO is also found as a thin layer on the surface of the alumina support throughout the catalyst cross section (bright edges in Fig. 8(a)). The appearance of NiO on the surface and the grey colour of the original catalyst leads to the assumption, that non-stoichiometric NiO with excess oxygen is present. This type of NiO has a higher conductivity than stoichiometric NiO. Although both types of NiO have a semiconductive behaviour [46,47]. The cross section of a catalyst which was held at 750 ∘ C for 60 h is shown in Fig. 8(b). The comparison of the SEM image in Figs. 8(b) and 8(a) reveals a shrinking of Ni/NiO grains which could be explained by a transition of NiO grains to Ni grains. This reduction is observed in XRD, see Fig. 6.The SEM of the catalyst, which is reduced for 12 h at 0 . 5 slpm H2, Fig. 8(c), shows no clear differences compared to Fig. 8(b). Only the XRD revealed, that nearly all NiO is reduced to Ni, see Fig. 6. Fig. 8(d) shows clear differences compared to Figs. 8(a)–8(c). The catalyst shown in this figure was loaded with carbon by applying 16 CH4 pulses as described in Section 2.2. Fig. 8(d) shows that the Ni surface layer of the catalyst diffused in the alumina carrier structure or eroded since there is no Ni layer visible on the surface as it is in Fig. 8(a) to 8(c).EIS based online monitoring of catalysts depends on knowledge of the initial or not degraded state. Different absolute impedance values were measured after reducing the catalyst and before starting the carbon deposition tests. This made it difficult to compare the state of degradation of the catalysts. To overcome this problem and get a better understanding of factors influencing the EIS results, measurements are also conducted during the heat up and the reduction phase. Within this section, we show that contacting methods have a significant impact on the EIS results after and during the heat up process. Fig. 9 shows the absolute impedances and phase shifts at an AC frequency of 1 0 5 Hz for the three contacting methods introduced in Section 2.1. Contacting using bare silver wires is shown in Fig. 9(a). A decreasing absolute impedance and increasing phase shift can be observed up to 600 ∘ C . A further increase of the temperature led to an increase of the absolute impedance and a decrease of the phase shift. This behaviour is only observed for contacting with bare silver wires. Hence it is assumed, that the resistance of the contacting is responsible for the shown characteristic. Figs. 9(b) and 9(c) show the absolute impedances and phase shifts for contacting with silver wires and silver ink, applied to two catalysts. A steady decrease of the impedance and a steady increase of the phase shift is observed except the temperature range between 400 ∘ C and 500 ∘ C . Comparing both measurements with each other reveals relatively noisy and high absolute impedances in Fig. 9(c). We assume, that although the contacting procedure was the same in Figs. 9(b) and 9(c), slight differences due to the manual contacting can be seen, resulting in a higher contacting resistance in Fig. 9(c). Finally, the measurement results for contacting with silver wires, Ni mesh and silver ink are shown in Fig. 9(d). This contacting method shows the lowest absolute impedance values compared to other contacting methods. However, the tendencies of the absolute impedance and phase shift are similar to these of Fig. 9(b). This leads to the conclusion, that trends can be observed even with higher contacting resistances. Nonetheless, the heat up procedure should be monitored to identify the initial state of the catalyst independent of the used contacting method. The lower absolute impedance observed with the silver wires, Ni mesh and silver ink combination could be attributed to a reduction of the measured distance through the catalyst and an increase in contact area (nickel mesh). At the beginning of the heat up procedure, all Ni of the catalyst is mainly present in form of NiO, see Section 3.1. These NiO grains have a semiconductive behaviour as long as the molar fraction of Ni in Ni–NiO grains is lower 20 mol% . For higher yields of Ni in Ni–NiO grains, a metallic conductive behaviour of the Ni–NiO grains was observed by Tare et al. [47]. This change of semiconductive to metallic conductive behaviour during the reduction of NiO grains to Ni grains seems to explain the absolute impedance drop in Figs. 9(b) and 9(d). Moreover, as mentioned above the catalyst used are partially reduced during the heat up process. This was proven by post mortem analysis carried out at both, the initial stage and after the heat up process, see Section 3.1.The reduction process is the last step to prepare catalysts for the reforming process. During the reduction process, H2 is added to the N2 flow to reduce NiO to Ni. The reduction should cause a decrease in the catalysts resistance since Ni has a lower specific resistance than NiO. However, morphological changes of the catalyst can occur if temperature and gas composition changes. Hence, effects such as the shrinkage of the Ni–NiO grains when they are reduced to Ni grains have to be taken into account. The shrinkage causes wider gaps between the Ni grains, see Fig. 4. Due to this gap, the Ni grains act as capacitors which increases the overall absolute impedance of the catalyst. Further, a reduction process, which has already lead to a reduced resistance is observed during the heat up phase.During the reduction process of the catalysts, impedance spectra between 1 0 − 1 Hz and 1 0 6 Hz are measured, see Figs. 10(b), 10(d), 10(f) and 10(h). The absolute impedances and phase shifts at 1 0 5 Hz over time are shown in Figs. 10(a), 10(c), 10(e) and 10(g). All test specimens show a slight decrease of the absolute impedance within the first one to two hours of the reduction process. The highest reduction of the impedance is observed for contacting with bare silver wires. After the initial reduction of the absolute impedance, a slight increase of the absolute impedance is observed. This increase seems to be related with the absolute impedance value at the beginning of the reduction process ( 0 h ). Smaller absolute impedance values show a relatively higher increase of the absolute impedance. In Figs. 10(a), 10(c), 10(e) and 10(g) it is also observed, that the absolute impedance does not seem to be settled at a certain value, although no NiO is detected by the XRD after the reduction process. The steady increase of the impedance is especially visible at low absolute impedance values, see Figs. 10(c) and 10(g). Hence, other hydration or reduction processes or Ni agglomeration might be observed in Figs. 10(a), 10(c), 10(e) and 10(g), as it is for SOFC anodes [48]. Summarizing this part shows, that lower contacting resistances seems to increase the observability of microstructural changes within the catalyst or the electrical contacting of the catalyst.The measured impedance spectra between 1 0 − 1 Hz and 1 0 6 Hz are shown in Figs. 10(b), 10(d), 10(f) and 10(h). Within these figures, Bode plots of the EIS measurement results are shown for 0 h , 6 h and 12 h after the start of the reduction process. Catalysts contacted with bare silver wires, Fig. 10(b), show a capacitive behaviour in the high frequency range with phase shifts up to − 90 ° . Catalysts contacted with silver wires and silver ink show different characteristics in the high frequency range comparing Figs. 10(d) and 10(f). Fig. 10(d) shows a characteristic close to an ohmic resistance but with a slight capacitive behaviour. The capacitive behaviour is especially visible for the measurement after 6 h , whereas a capacitive characteristic is visible in Fig. 10(f) in the high frequency range. The catalyst contacted with silver wires, Ni mesh and silver ink is shown in Fig. 10(h). The characteristic of this contacting method is clearly different to the others. Instead of a capacitive behaviour at high frequency ranges, an inductive behaviour is visible. Contacting with silver wires, Ni mesh and silver ink also shows a decreasing inductive behaviour along the reduction time. The summarized results from Figs. 10(b), 10(d), 10(f) and 10(h) show, that test specimen with a lower absolute impedances tend to a less capacitive behaviour. This leads to the conclusion, that the contacting resistance might not only reflect an ohmic resistance but a combination of a resistor and a capacitor. The influence of different contacting methods is visible within all measurements done during the heat up and reduction process. Therefore, this influence has to be taken into account for further measurements and monitoring during carbon deposition.The main objective of online monitoring is to detect degradation effects at early stages. Carbon deposition is one major degradation effect for catalysts used in syngas production. The performance deterioration of Ni catalysts through carbon deposition is caused by carbon covering catalytic active surfaces, blocking gas channels and causing Ni dusting. Carbon deposition on and between Ni grains should be detectable through decreasing electrical resistances, since carbon has a higher conductivity than Ni and is bridging gaps between the Ni grains.During the carbon deposition process, the impedance spectra are measured after each CH4 pulse, see Section 2.2. This is done to measure the changes of the impedance caused by possible carbon deposition. Fig. 11(a) shows the absolute impedance and phase shift at 1 0 5 Hz over time for the test specimen “Cat. silver wires with silver ink, low resistance”. It is observed, that the impedance decreases and the phase shift increases after every CH4 pulse. Such behaviour is linked with carbon deposition on the catalyst since carbon was found on the test specimen after the carbon deposition process, see Section 3.1. The degradation of the catalyst is identified as a decrease of the methane cracking reaction (Eq. (1)). Latter is caused by deposited carbon which leads to less catalytic active surface area of the catalyst. Due to less catalytic active surface area, a lower amount of CH4 is converted and the measured volumetric percentage of methane increases. (1) CH 4 → 2 H 2 + C In Fig. 11(a) the decrease in methane cracking is observed through decreasing H2 and increasing CH4 peaks over time. The first methane pulse shows nearly the same volume fraction of CH4 and H2 in the dry off-gas. However, the following CH4 pulses show a decreasing volume fraction of H2 and an increasing volume fraction of CH4. The decrease of CH4 conversion behaves like a degressive function, which is indicated by the decrease of H2 peaks measured in the off-gas in Fig. 11(a). The decreasing conversion rate of CH4 is caused by coking of the catalyst. The same degressive function of CH4 conversion caused by coking of the catalyst is observed in other studies. Franz et al. and Li et al. observed a decrease of CH4 conversion due to coking under dry-reforming conditions ( CO 2 / CH 4 = 1 ) [22,49]. Their observed decrease of CH4 conversion also behaves like a degressive function. The same behaviour for the decrease of CH4 conversion is found under bi-reforming conditions and summarized by Mohanty et al. [50]. In Fig. 11(a), the absolute impedance at 1 0 5 Hz shows a decreasing rate of change for an increasing number of CH4 pulses which are interpreted as an increasing amount of deposited carbon. The absolute impedance value changed by 80 % after the first CH4 pulse, whereas the absolute impedance after the second CH4 pulse decreased by 90 % compared to the initial state or 47 % compared to the absolute impedance value after the first CH4 pulse. The measurements reveal, that even slight carbon deposition (the first CH4 pulses) has a significant impact on the impedance values as seen by the decrease of the absolute impedance value in Fig. 11(a). The decrease of the absolute impedance is explained by higher conductivity of carbon compared to Ni and carbon or graphite particles bridging gaps between Ni-grains, see Fig. 4(c). If carbon deposition is detected at early stages, countermeasures such as increasing the operating temperature, the amount of steam or oxygen could be applied. These countermeasures could lead to a prolonged lifetime of the catalyst.The impedance spectra of the test specimen “Cat. silver wires with silver ink, low resistance” are shown as Bode plots in Fig. 11(b). The Bode plots show a decreasing phase shift for frequencies above 1 0 4 Hz at increasing carbon loading. Below 1 0 4 Hz , the phase shift does not show changes for an increased carbon load. The absolute impedance decreases with increasing carbon load along the whole frequency spectra. A change of the absolute impedance characteristic is also observed at frequencies above 1 0 5 Hz for increasing carbon loading. However, the decrease of the absolute impedance is the dominating aspect for monitoring carbon depositions. The same trends for the absolute impedance values and phase shift caused by carbon deposition on a Ni based impedance sensor were found by Müller et al. [51].The same evaluation as in Fig. 11 is done in Fig. 12 for the test specimen “Cat. silver wires with silver ink, high resistance” to identify the influence of the contacting resistance. The absolute impedance at 1 0 5 Hz over time shows higher changes for the test specimen “Cat. silver wires with silver ink, high resistance” compared to “Cat. silver wires with silver ink, low resistance”. The change of the absolute impedance after the first CH4 pulse is 94 % in Fig. 12(a) and 80 % in Fig. 11(a) compared to the respective initial state. Hence, it seems that higher contacting resistances are favourable for online monitoring of carbon deposition on Ni catalysts. Though, the higher sensitivity could be caused by local phenomenon such as graphite deposition between the silver wires and the catalyst. Thus, the measurement might not be representative for the whole catalyst and its state of health. However, the Bode plots of the measured absolute impedances show the same trends and characteristics for both contacting resistances, compare Figs. 11(b) and 12(b). Even the absolute impedance after the 16th CH4 pulse is nearly identical for the test specimen in Figs. 11 and 12. During the measurements conducted to test specimen “Cat. silver wires with silver ink, high resistance” the 7th and 8th CH4 pulse were skipped, see Fig. 12(a). The impedance spectra, however, were measured at the same time steps as previous measurements. This was done to prove that a steady state is reached between CH4 pulses. These two additional measurements between two CH4 pulses show nearly the same absolute impedances and phase shifts. The maximum deviation of the absolute impedance value of these three measurements is within a range of ± 1 . 3 % of their mean value. This suggests, that carbon deposition and no other phenomenon is observed since carbon deposition increases the conductivity of the catalyst due to its high conductivity. Further, the deposited carbon creates electrical connections of Ni-grains which also increases the conductivity of the catalyst. It should be mentioned, that the first CH4 pulse is not visible in the dry off gas composition of Fig. 12(a) since an error at the gas pump occurred.This study demonstrated applicability of EIS-based online monitoring tools to identify degradation mechanisms that occur in commercial Ni-based reforming catalysts, which was presented for the first time. We investigated this measurement method to expand the knowledge of online monitoring methods for commercial Ni-based catalysts since no EIS based online monitoring method for commercial Ni based catalysts were available in literature.By applying EIS measurements during the heat up and reduction of the catalyst, it was possible to determine microstructural changes such as NiO reduction by dropping absolute impedance values and set the impedance value for the not degraded condition of each test specimen. Further, three different contacting methods were tested and their influence on the measurement results is presented. Differences of the ohmic resistance of more than 1 0 2 Ω between the contacting methods were observed after the heat up procedure. Nevertheless, it was possible to observe catalyst degradation due to carbon deposition even with higher contacting resistances. Even after a short period of carbon loading, changes in the absolute impedance values up to 94 % of the initial values were measured. However, to detect other microstructural changes such as NiO reduction, it might be necessary to choose a contacting method with a lower contacting resistance, e.g. contacting with silver wires with Ni mesh and silver ink.These initial tests and findings lay the foundation for online monitoring of catalysts, especially but not only for SOFC-reformer systems. This measurement methodology could further be used to gain more insights in the reforming processes at laboratory scale and help finding safe and high efficient operating conditions. Within our future research, we are going to apply the measurement methodology to real small scale reformer applications and even to small scale SOFC-reformer systems. The following abbreviations are used in this manuscript: AC Alternating current DC Direct current EDS Energy dispersive X-ray spectroscopy EIS Electrochemical impedance spectroscopy MFC Mass flow controller SEM Scanning electron microscopy slpm standard litre per minute SOFC Solid oxide fuel cell XRD X-ray diffraction Z Impedance Michael Höber: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Visualization, Project administration. Philipp Wachter: Methodology, Investigation, Resources. Benjamin Königshofer: Investigation, Resources. Felix Mütter: Investigation, Visualization. Hartmuth Schröttner: Investigation, Data curation. Christoph Hochenauer: Supervision, Project administration, Funding acquisition. Vanja Subotić: Conceptualization, Methodology, Resources, Writing – review & editing, Supervision, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This project has been funded by partners of the ERA-Net SES 2018 joint call RegSys (www.eranet-smartenergysystems.eu) - a network of 30 national and regional RTD funding agencies of 23 European countries. As such, this project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 775970. The authors gratefully acknowledge the funding of this project entitled “AGRO-SOFC” (Grant No. 872299) by The Austrian Research Promotion Agency (FFG) . We also want to mention that this work is done within the research initiative “Nachhaltige Personen- und Gütermobilität”.

More than 75 % of today’s H2 production is based on reforming processes using heterogeneous catalysts. In addition, catalysts are needed for hydrogen generation from renewable resources such as biomass or biogas. However, no direct online monitoring of commercial Ni based catalysts is established. Catalysts are only monitored indirectly by measuring gas compositions, temperature profiles or using coke sensors, although direct online monitoring could detect degradation mechanisms at early stages. We demonstrate the methodology for electrochemical impedance spectroscopy based online monitoring of commercial Ni catalysts. Furthermore, we studied the impact of three different contacting methods of Ni catalysts with ohmic resistances between 10 Ω and 1 0 5 Ω after the heat up procedure on the measurement results. Monitoring of the heat up phase revealed, that choosing the right contacting method is essential to observe processes such as NiO reduction, whereas monitoring of degradation due to carbon loading was observed with every tested contacting method. The demonstrated online monitoring of catalysts could be used to find and maintain more efficient and stable reforming conditions. In addition, the gained knowledge could even be used to prolong the lifetime of catalysts by in situ adapting of operating conditions.

Data will be made available on request.The increase in crude oil prices and its negative impact on the environment such as air pollution, ozone depletion, and climate change has led to the growing interest in the use of renewable and less-pollutant resources [1,2]. Synthetic gas (or syngas), a mixture of CO and H2 is recognized as an environmentally friendly alternative energy source in recent years and it can be directly used as a fuel source for electricity generation and transport fuel [3,4]. Commonly, syngas is produced through partial oxidation of methane [5], methane steam or dry reforming [6,7], oxidative methane steam or dry reforming [8,9], ethanol steam or dry reforming [10,11] and oxidative ethanol steam reforming [12]. However, methane from natural gas is not a renewable source and thus its availability is limited. There is a growing interest in the use of ethanol among biomass-derived feedstocks [13]. Compared to other feedstocks such as, glycerol, ethanol offers low toxicity, ease of production in large quantities, relatively high hydrogen content and it is free from sulfur-containing compounds [14]. Ethanol can be produced either by fermenting sugar or starch (first generation) or hydrolysing lignocellulose and fermenting it (second generation) [15]. There have been many studies conducted on reforming processes using both non-noble (Ni-based catalysts) and noble metal (Pt and Rh) catalysts to produce syngas. Osaze et al. studied the effect of temperature from 923 to 1023 K over 10 %Ni/SBA-15 catalyst on the performance of methane dry reforming and found that when temperature increased both CH4 and CO2 conversions raised about 83.4 % and 59 %, respectively due to endothermic nature of methane dry reforming [16]. However, Ni catalysts are currently faced with the challenge of early deactivation caused by the coke formation at lower temperatures [17]. In addition, cobalt-based catalysts are also used to produce syngas from oxidative ethanol steam reforming due to their high activity, stability, and low-cost alternative to noble metals [18,19]. Pereira et al. investigated the catalytic behavior and regeneration processes of oxidative ethanol steam reforming over Co/SiO2, Co–Rh/SiO2, and Co–Ru/SiO2 catalysts. By using oxidative treatment, CoRh/SiO2 and CoRu/SiO2 catalysts were activated, resulting in higher ethanol conversion and hydrogen selectivity after regeneration [20]. Sukri et al. also studied the effect of cobalt loading (Co=10 %, 15 %, 20 % and 25 %) over Co/MgO catalysts in methane dry reforming and found that the 10 %Co catalyst showed good activity, stability, the highest CH4 and CO2 conversions, and the lowest rate of carbon deposition at 750 °C [21]. Thus, a new, and environmentally more positive approach is oxidative ethanol dry reforming (OEDR) (cf. Eq. (1)), which converts CO2 greenhouse gas and produces value-added synthesis gas. (1) 3 C 2 H 5 OH + C O 2 + O 2 → 7 CO + 9 H 2 Δ H 298 K 0 = 325.3 kJ mo l − 1 To the best of our knowledge, none of the available studies have explored oxidative ethanol dry reforming over Co/Al2O3 catalyst. Therefore, the objective of this was the chemical and physical characteristics of 10 %Co/Al2O3 catalyst in addition to investigating the effect of reaction temperature on the activity and selectivity of OEDR reaction.The wet impregnation method was used to impregnate 10 % (by weight, metallic) cobalt on alumina [21]. To ensure thermal stability, an adequate amount of puralox alumina (SCCa‐150/200 procured from Sasol, Hamburg, Germany) was calcined for 5 h at 1023 K in a Carbolite (Bemaford, Sheffield, UK) furnace with air and a heating rate of 5 K min−1. An aqueous solution of Co(NO3)3.6 H2O was supplied and magnetically stirred for 3 h with pretreated γ-Al2O3 support in am ambient environment (Sigma‐ Aldrich, St. Louis, Missouri). The mixture was dried at 383 K for 24 h. Moreover, it was calcined in air with a heating rate of 5 K min−1 and kept at constant temperature of 773 K for 5 h. Post crushing and sieving, the catalyst was introduced into a fixed-bed reactor with a particle size between 125 and 160 µm.Micromeritics ASAP-2020 (Norcross, Georgia) at 77 K was used to measure Brunauer-Emmett-Teller (BET) surface areas for 10 %Co/Al2O3 catalyst and γ-Al2O3 support. During BET measurement, the example was degassed for 1 h at 573 K in N2 flow to remove moisture and volatile contaminants. Rigaku Miniflex II (Akishima‐shi, Tokyo, Japan) X-ray diffraction system was utilized to study the crystal structure of γ-Al2O3 support and 10 %Co/Al2O3 catalyst at 30 kV and 15 mA and Cu target was used as a source of radiation (wavelength, λ of 1.5418 Å). Diffraction patterns were scanned from 3° to 80° with an imaging speed of 1° min−1 and a step size of 0.02° to obtain high-resolution X-ray diffractograms. A software tool (Match! version 2.3.3) was used to measure all X-ray patterns. A micromeritics AutoChem II-2920 apparatus was used for both alumina and 10 %Co/Al2O3 catalyst to conduct the H2-TPR experiment. The U-tube of quartz was loaded with 0.1 g of sample and sandwiched with quartz wool. As an initial treatment, the sample was heated to 373 K under 50 ml min−1 in He flow for 30 min to remove volatile compounds from the sample. Following this, the temperature of the sample was increased to 1173 K and kept at the constant temperature for 30 min under 50 ml min−1 10 %H2/Ar mixture. The amount of carbon accumulated on the spent specimen surface after OEDR, temperature-programmed oxidation (TPO) was measured using a thermogravimetric analyzer (TGA Q500, TA Instruments, New Castle, Delaware). During TPO, the catalyst was preheated to 373 K (heating rate 10 K min−1) for 30 min under N2 (100 ml min−1) atmosphere. Thereafter, the temperature was increased from 373 to 1023 K (10 K min−1 ramping rate) under 3 N2:1 O2 flow. Under N2 atmosphere, the sample was cooled to ambient temperature and was isothermally heated. Isothermal heating of the sample was carried out for 30 min and the sample had to be cooled with N2 to reach ambient temperature. Micromeritics AutoChem II-2920 chemisorption system was utilized to determine both catalyst and support acidic properties. Before each measurement, approximately 0.1 g of the sample was pretreated at 773 K for 1 h at 50 ml min−1 under He flow to eliminate moisture and physisorbed compounds. The sample was cooled to 423 K under inert atmosphere after reduction in situ. Thereafter, adsorption was performed for 30 min at the same temperature in 50 ml min−1 of 10 %H2/Ar. The NH3 molecules in the gas phase were removed by purging with He gas for 30 min at 423 K after 1 h of adsorption using 5 % NH3 in He balance. As part of the purging process at the same temperature with He gas for 30 min, NH3 molecules were removed from the gas phase by heating at 1073 K (heating rate 10 K min−1) for 10 min. Thermal conductivity detectors (TCD) were used to measure the quantity of desorbed NH3 gas entering the U-tube from the outlet.A quartz tube reactor having an outer diameter of 3/8 in. and length of 17 in. was used to conduct OEDR experiments. This reactor was placed vertically within a split tubular furnace (LT furnace) during the experiments with stoichiometrically set to 3:1:1 for C2H5OH: CO2:O2 and temperatures between 773 and 973 K under atmospheric pressure. OEDR was performed on the catalyst by reducing it to 973 K with 50 % H2/N2 (60 ml min−1) with heating at a rate of 10 Kmin−1 for 2 h before the reaction. The quartz tube reactor was filled with approximately 0.1 gcat of the catalyst surrounded by a layer of quartz wool. In this experiment, KellyMed KL602 syringe pump (Beijing, China) and Alicat mass flow controller (Tucson, Arizona) were employed to ensure that ethanol and gas (viz, CO2, O2 reactant and N2 diluent) were accurately fed to the top of the reactor. The gas hourly space velocity (GHSV) was calculated as 42 L gcat −1 h−1 for each reaction. To obtain the intrinsic catalytic activity, high GHSV, small catalyst loadings, and tiny particle sizes were selected in order to ensure negligible mass and heat transfer resistances. The detailed calculation is included in the supplementary information for avoiding the mass and heat transfer intrusions. To maintain the 70 ml min−1 flow rate, N2 was used as a tie component. As part of the analysis, a gas chromatograph (GC) from the Agilent 6890 Series (Agilent, Santa Clara, California) fitted with FID and TCD detectors to determine the composition of the gaseous effluent. The carbon balance is calculated by dividing the total moles of carbon in the products with the total moles of carbon reacted. The carbon mass balance was carried out for each run of the reaction, and it was greater than 91.3 %−98.8 %, confirming their remarkable resilience toward coke deposition during the OEDR.The γ-Al2O3 support and 10 %Co/Al2O3 catalyst were examined for their textural characteristics, such as BET surface area, average pore volume, and pore diameter. It was observed that the γ-Al2O3 support had a relatively BET area of 175.2 m2 g−1, an average pore volume of 0.46 cm3 g−1, and a pore diameter of 10.7 nm. However, the surface area, pore-volume, and pore size of the 10 %Co/Al2O3 catalyst were smaller having values of 143.1 m2 g−1, 0.36 cm3 g−1 and 10.6 nm, respectively. This could possibly be due to the introduction of Co oxides onto the γ-Al2O3 support surface. Fig. 1 displays the comparison of fresh and spent XRD profiles of 10 %Co/Al2O3 catalyst and the calcined γ-Al2O3 support. The Joint Committee on Powder Diffraction Standards database was utilized to obtain a qualitative interpretation of the crystalline phase present in all specimens [22]. The γ-Al2O3 phase peaks at 2θ of 18.92º, 32.88º, 37.10º, 45.61º, and 67.17º was detected on fresh 10 %Co/Al2O3 catalyst (JCPDS card number: 04–0858) see Fig. 1(a). Furthermore, the spinel CoAl2O4 phase was observed at 2θ of 59.51º and 65.38º (JCPDS card number: 82–2246) over 10 %Co/Al2O3 catalyst. This was due to strong metal support interaction between Al2O3 and CoO, resulting in the formation of CoAl2O4 (see Fig. 1(b) and (c)) [23]. However, CoAl2O4 form was also observed on spent specimens (see Fig. 1(c)). As a result, it would be expected that the low peak intensity and absence of 2θ = 65.38° would indicate that the lower amount of CoAl2O4 phase on the spent catalyst than the fresh catalyst could be due to the reduction of H2 to Co0 during activation. The XRD patterns of spent 10 %Co/Al2O3 catalyst after the OEDR at P C O 2 = P O 2 = 5 kPa, P C 2 H 5 O H = 15, and 973 K is shown in Fig. 1(c). In both fresh and spent samples, Co3O4 phase was detected at 2θ of 31.45º, 37.10º, and 44.79º (JCPDS card number: 74–2120) see Fig. 1(b) and (c). However, the presence of the Co3O4 phase on the spent catalyst indicates that the Co0 metallic phase was unavoidably re-oxidized during the OEDR process due to the catalyst being sufficiently reduced in H2. Based on a diffractogram of the spent catalyst, the first broad peak centered around 2θ of 26.38º can be attributed to graphitic carbon (JCPDS card number: 75–0444) that is likely to have formed during the decomposition of ethanol and cracking of CH4 intermediate at a high temperature [24]. Additionally, a new peak was observed on spent catalyst at 2θ of 51.50º (JCPDS card number: 15–0806) can be attributed to the Co phase [25,26]. Consequently, the stability of the catalytic performance can be attributed to the maintenance of the active metal phase after the OEDR process.The H2-TPR method was performed to investigate the reducibility of catalyst and support. According to Fig. 2(a), the H2-TPR analysis of calcined γ-Al2O3 did not indicate any reduction peaks and it was stable and did not reduce in response to H2. Furthermore, three significant peaks (P1, P2, and P3) were observed on 10 %Co/Al2O3 catalyst surface (Fig. 2(b)). P1 at temperatures between 458 and 720 K was due to the reduction of Co3O4 into intermediate CoO (cf. Eq. 2), while P2 at temperatures between 743 and 765 K corresponds to the reduction of CoO into metallic Co0 (cf. Eq. 3) [27]. Moreover, another shoulder peak (P3) was observed at temperatures between 766 and 1014 K. This is attributed to the reduction of the spinel CoAl2O4 phase into the metallic Co0 phase [28] (see Eq. 4). (2) Co 3 O 4 + H 2 → 3 CoO + H 2 O (3) CoO + H 2 → Co + H 2 O (4) Co Al 2 O 4 + H 2 → Co + Al 2 O 3 + H 2 O In addition, Papageridis et al. [29] have also revealed that, due to high calcination temperatures, Co2+ ions migrate into the lattice of Al2O3 support and persist in tetrahedral positions in spinel CoAl2O4. As a result, CoO and Al2O3 interact strongly in CoAl2O4 species, which can produce a strong resistance to H2 reduction. Fig. 3 shows a measurement of the NH3-TPD over γ-Al2O3 support and 10 %Co/Al2O3 catalyst. The γ-Al2O3 support and 10 %Co/Al2O3 catalyst exhibit weak, medium, and strong acid sites for different desorption temperatures ranging from 423 to 570 K, 571–710 K, and 721–1026 K, respectively [30,31]. Consequently, the strong acid sites possess a higher NH3 desorption temperature than 713 K and is likely that they correspond to Brønsted acid sites. However, while the weak and medium acid sites possess a lower NH3 desorption temperature, indicating the presence of Lewis and/or Brønsted acids sites [32]. According to Fig. 3, the γ-Al2O3 support contains three different acid centres, resulting in an overall NH3 uptake of 4.77 mmol NH3 gcat −1. Adding Co metal to γ-Al2O3 significantly improved the NH3 uptake from 4.77 to 6.89 mmol NH3 gcat −1 (about 44.4 %). Based on this observation, it is possible that an extra acid site is formed at the interface between the Co metal and γ-Al2O3 support. Cheng et al. [33] reported that the adding Co to the calcined support increased acid site concentration and increased strong acid site concentration. According to this observation, some weak acid sites were replaced during thermal activation by impregnating Co species, resulting in strong acid sites. Thus, the catalytically active site may be protonated and likely located at the interface between the metal and alumina support.In terms of carbon formation on a surface, it is well known that the acidity of the surface is a significant factor, whether the surface is the catalyst or the support. The formation of carbon is accelerated by positively charged acidic sites on a surface due to acidic sites catalyzing the cracking reaction. Gamma alumina is generally used as a support material during the reforming process, and its acidic properties facilitate carbon formation [34,35].This study examined the effect of reaction temperature over 10 %Co/Al2O3 catalyst with stoichiometric amounts of P C O 2 = P O 2 = 5 kPa, and P C 2 H 5 O H = 15 kPa. The study was conducted within a temperature range of 773 and 973 K under atmospheric pressure. As illustrated in Fig. 4, temperature increase from 773 to 973 K resulted in increased conversions of C2H5OH and CO2 by 22.5–93.6 % and 16.9–52.8 %, respectively. This observation can be attributed to the ethanol decomposition reaction (see Eq. (5)) [36]. (5) C 2 H 5 OH → CO + H 2 + CH 4 ( Δ H 298K 0 = 50.1 kJmol - 1 ) The reason for the enhanced performance of C2H5OH conversion rather than CO2 conversion is the presence of side reactions with reasonable decomposition of ethanol and dehydrogenation. The significant conversion of C2H5OH over CO2 was due to the numerous dehydrogenation and ethanol decomposition side reactions [37]. Furthermore, the addition of O2 during the reforming reaction suppresses carbon formation and decreases the required heat, resulting in an exothermic reaction [38]. Fig. 5 illustrates the yields of CO, H2 and CH4 as a function of temperature at P C O 2 =  P O 2 = 5 kPa, and P C 2 H 5 O H = 15 kPa. With an increase in temperature from 773 K to 973 K, the yield of both products (H2 and CO) increased from 16.0 % to 68.1 % and 13.5–58.3 %, respectively. Increasing the temperature resulted in an increase in both H2 and CO, which is consistent with the endothermic nature of Eq. (1). On the other hand, CH4 yield also increased with rising reaction temperature (see Fig. 5). This indicates that during the C2H5OH decomposition (see Eq. (5)), CH4 production rate was higher than the CH4 reforming rate (reforming of CH4 by CO2 to produce syngas). Besides, this may indicate the successful conversion of ethanol into syngas [39]. As Bartholomew previously reported, the increase in CH4 yield with reaction temperature may be due to lower carbon deposition (methane dehydrogenation) [40]. Moreover, O2 as a reactant decreased the amount of carbon deposition during the OEDR reaction while improving the stability of the catalytic reaction for a long period of time.The CH4/CO and H2/CO ratios are determined by varying the reaction temperature at P C O 2 =  P O 2 = 5 kPa and = 15 kPa in Fig. 6. Increasing reaction temperature resulted in a linear increase of H2/CO ratio from 1.2 to 1.5, indicating an improved C2H5OH dehydrogenation reaction [41]. As the reaction temperature increased, CH4/CO ratio improved. It indicates that the rate of dry reforming of CH4 was lower than the rate of C2H5OH decomposition. Alongside, the preferred CO/H2 ratio is less than 2 and can be used as feedstocks in Fischer-Tropsch synthesis to produce green fuels [42]. Table 1 shows the summary of the evaluation of the 10 %Co/Al2O3 catalyst for OEDR, as well as other catalysts recently used in the oxidative steam reforming (OSR) reaction. Based on the results shown in Table 1, the 10 %Co/Al2O3 catalyst exhibited relatively comparable conversion of C2H5OH and H2 selectivity during the OEDR runs when compared with other Co-based and noble-based catalysts in the literature. Even though the 10 %Co/Al2O3 catalyst in this study has a slightly lower activity than noble metal catalysts, from a practical and economic standpoint, it would be a useful catalyst for large-scale syngas production via OEDR.TPO measurements were used to determine the amount of carbon deposition on the surface of the spent 10 %Co/Al2O3 catalyst. Fig. 7 shows the TPO results for the weight percentage of the spent sample. The spent 10 %Co/Al2O3 catalyst deposited the least amount of carbon (28,92 %) at 973 K. Nevertheless, the reaction temperature decreased from 973 to 773 K, and the amount of carbon deposition improved by 41.48 %. This demonstrates quicker deposition of carbon on the catalyst surface. As shown in Figs. 4 and 7, the trend of carbon weight vs. temperature curve is opposite to that of CO2 and C2H5OH conversions, further indicating that the catalytic activity improved via the oxidization of carbonaceous deposition. On the other hand, XRD analysis also showed that graphitic carbon was present on the surface of the spent catalyst (see Fig. 1(c)). Ruckenstein and Wang also reported that the stability of Co/γ-Al2O3 catalysts with several Co loadings and calcination temperature (6 wt. % for Tc =500 °C and 9 wt. % for Tc =1000 °C) exhibited stable activity. However, catalysts with high Co loadings (above 12 wt. %) accumulated significant amounts of carbon during reforming and demonstrated deactivation [51]. Thus, the reduction of carbon deposited on the catalyst surface resulted in a higher conversion of C2H5OH and CO2.The present study describes the OEDR for syngas production over Co/Al2O3 catalyst at various reaction temperatures. The catalyst design consists of 10 wt % Co and γ-Al2O3 support with a high specific surface area, which can prevent the sintering impact. OEDR allows the active metal phase of the catalyst to be maintained during the catalytic process, which contributes to a stable catalytic performance. The interaction between CoO and Al2O3 can produce CoAl2O4 species, and these compounds exhibit strong resistance to H2 reduction. The level of NH3 uptake was increased significantly from 4.77 to 6.89 mmol NH3 gcat −1, resulting in the formation of extra acid sites at the interface of the Co metal and γ-Al2O3 support. The catalyst displays high performance for oxidative ethanol dry reforming to generate synthesis gas. Thus, it is a suitable candidate to be used as a fuel for internal combustion engines and as a chemical feedstock for the production of ammonia and methanol. According to tests conducted under various reaction temperatures, the conversion of C2H5OH and CO2 increased with an increase in reaction temperature and decreased with a decrease in reaction temperature. Further, the addition of oxygen to the feed gas enhances the production of H2, CO, and CH4 while at the same time limiting the accumulation of carbon.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors acknowledge the financial support from the Universiti Malaysia Pahang (UMP) Research Grant Scheme (RDU130376). Fahim Fayaz would like to thank Dr. Dai-Viet N. Vo for his invaluable guidance and support during the research work. Fahim Fayaz is also grateful for the funds received from the Institute of International Education’s Scholar Rescue Fund (IIE-SRF) and the Finnish National Agency for Education (EDUFI) for supporting his postdoctoral fellowship at Tampere University.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.mtcomm.2023.105671. Supplementary material .

Till date, oxidative ethanol steam reforming use Ni-based catalysts to produce syngas. However, Ni catalysts suffer from easy deactivation due to the coke formation at low temperatures. Therefore, oxidative ethanol dry reforming is a promising method and was investigated over 10 %Co/Al2O3 catalyst due to their high activity and stability to produce high-quality syngas. More importantly, the syngas can be upgraded to produce liquid biofuels and chemicals. The catalyst was evaluated in a quartz fixed-bed reactor under atmospheric pressure at P C O 2 = P O 2 = 5 kPa, P C 2 H 5 O H = 15 kPa, with reaction temperature ranging between 773 and 973 K. The γ-Al2O3 support and 10 %Co/Al2O3 catalyst had BET surface areas of 175.2 m2 g−1 and 143.1 m2 g−1, respectively. Co3O4 and spinel CoAl2O4 phases were detected through X-ray diffraction measurements on the 10 %Co/Al2O3 catalyst surface. H2-TPR measurements indicate that the 10 %Co/Al2O3 catalyst was completely reduced at a temperature beyond 1000 K. NH3-TPD measurements indicated the presence of the weak, medium, and strong acid sites on the γ-Al2O3 support and 10 %Co/Al2O3 catalyst. Due to increased reaction temperature from 773 to 973 K, C2H5OH and CO2 conversions improved from 22.5 % to 93.6 % and 16.9–52.8 %, respectively. Additionally, the optimal yield of H2 and CO obtained at 68.1 % and 58.3 %, respectively. Temperature-programmed oxidation experiments indicated that the amount of carbon deposition was the lowest (28,92 %) at 973 K and increased by 41.48 % at 773 K.

Brunauer–Emmett–TellerBarret–Joyner–Halendaaverage pore diameterFourier transform infraredspecific surface area obtained by BET methodspecific surface area of microporesThermal conductivity detectorThermogravimetric AnalysisTemperature programmed desorptionpore volumeMicropore volumeBiomass as a renewable energy source provides decreasing of the emitted CO2 amount. Based on statics, globally, 170 billion metric tons of biomass source is available from which significant waste is left behind [1]. The products resulted in pyrolysis and gasification of different biomass can be used as an alternative fuel, blending component, besides its value-added conversation the products can reach the appropriate quality. In 2020, the emitted carbon dioxide was 31.5 Gt, which - based on the statics - will reach 37 Gt in 2023. It is important to note, that the predicted value can be reduced to 28.5–30 Gt by the use of sustainable and renewable materials [2–5].The synthesis gas, sourced by biomass gasification is one of the most important intermediary components in the chemical industry. Based on the origin of the biomass, the gas product contain different concentrations of hydrogen, carbon monoxide, carbon dioxide, methane, hydrocarbons with low molecular weight, sulphur, nitrogen and chlorine-containing compounds. Henceforward, purifying of the synthesis gas is an important issue, mainly for CO2 capture [6–8].During biomass pyrolysis-gasification, numerous reactions take place, therefore the composition of the gaseous product are mainly affected by them. E.g. the equilibrium reactions can be shifted to the appropriate composition with the use of steam and catalysts. Nickel-containing catalysts are commonly used owing to their high efficiency in cracking of the C–H, C–C and C–O bonds, their low price and their regeneration availability. Besides nickel, the ruthenium and the rhenium-containing catalysts can be listed, however, their price cannot be considered economic friendly, especially in scale-up. Regarding the catalyst support, the ZSM-5, HZSM-5, Al2O3, Al2O3–SiO2 are the most commonly used. Henceforward, the CaO and the natural zeolites should be perspective materials, owing to their particle size and efficiency in not only gasification processes, but in-situ carbon capture. Based on the literature, the CaO, as well as the natural zeolites, can be impregnated promoting the carbon capture processes and increasing the yield of the gaseous product [9–12].Wang et al. investigated the effects of the Ni/Al2O3 and Ni/CaO during high-temperature biomass gasification [9]. They demonstrated, that the catalyst with CaO carrier can increased the gas yield, the hydrogen yield, however, decreased the amount of hydrocarbons with low molecular weight [9]. W.X. Peng et al. concluded, that the residue of high-temperature gasification of wood residue was significantly decreased in the presence of Ni/Al2O3 and Ni/Ce/Al2O3 catalysts, but higher gas yield was found using Ni/Ce/Al2O3 catalyst [13]. It is important to mention that the H2/CO ratio was not significantly affected by the different amounts of the catalysts during the measurement [13]. Higher gas and hydrogen yield was also reported by Yue Chai et al. with the nickel content of Ni–CaO–C catalyst during biomass and plastic waste co-pyrolysis [14]. It is important to note that there was no significant change in the quality of the gaseous products over nickel content of 10%. Afterwards, the regenerated catalysts can be reused, but with decreased efficiency [15]. Shang et al. investigated the regeneration and cycle stability of Ni modified Zr-MOF catalyst during biomass gasification, where the agglomeration of nickel was indicated after several reuse. It was observed, that the catalytic performance of regenerated catalyst was slightly increased [16]. It was also concluded, that the zeolite catalysts had partially lost their activity due to several regeneration cycles using pinewood sample [17].This work aims to investigate the pyrolysis-gasification of biomass with maximizing of the gaseous product and syngas, and simultaneously the carbon dioxide capture. The main goal of the experiments was to compare the effect of the CaO with zeolites in biomass gasification and CO2 adsorption. Since the regeneration and the cycling reuse of the catalyst are proved to be a research gap, the efficiency of the catalysts during ten regeneration cycles was also investigated from an economic point of view.Crashed (<3 mm) and dried form of maize biomass waste (roots, leaves, stems, corn stalk) was used as raw material. Biomass sample has 36.5% carbon, 5.2% hydrogen, 0.9% nitrogen and 57.4% oxygen content.During the experiments four different types of catalysts were used, namely are Ni/ZSM-5, Ni/Al2O3, Ni/CaO, Ni/Clinoptilolite. The support had been chosen based on their acidity, the efficiency in the cracking reactions and the in-situ carbon capture effect in the relevant cases. The preparation method was the same in all cases. The catalyst supporters had been impregnated by the using of Ni(NO3)2·6H2O solution at 80 °C till 3 h, then the slurry was filtered and dried at 110 °C till 10 h. Finally the treated catalysts were calcined at 600 °C till 3 h.The main surface properties of the catalysts were determined by a Micromeritics 3Flex 3500 instrument using the BET (Brunauer–Emmett–Teller) method. The pore-size distribution and pore volumes were calculated using the BJH (Barret–Joyner–Halenda) model. The temperature programmed desorption of ammonia (NH3-TPD) was used to measure the number and acid strengths of sites found on solid catalysts using an analyser Micromeritics AutoChem-2920 precision chemisorption analyzer equipped with a heat-conductivity detector. The prepared modified catalysts were also investigated by an Apreo S LoVac instrument (FEI/ThermoFischer) coupled with an energy-dispersive X-ray spectrometer (AMETEK, Octane Elect Plus), operated at 2.0 for secondary electron imaging, and 25.0 kV for elemental analysis. The main properties of neat catalysts are summered in Table 1 .The Ni/ZSM-5, Ni/Al2O3 and Ni/CaO catalysts were in the form of fine powder with average diameter under 20 μm. Contrary, the Ni/Clinoptilolite catalyst has higher drain size. Another significant difference was found in the BET surface, because the Ni/ZSM-5 catalyst has 335 m2/g surface area, while the others less than 60 m2/g. Similar notable difference was reported regarding the “Smicro“ value. It is important to mentioned, that Ni/ZSM-5 catalyst has the smallest average pore diameter, while Ni/Al2O3 has the highest. Regarding the C–C chemical bond scission effect of the catalysts, not only their surface area and Si/Al ratio, but also the acidity should be also a crucial property. The acidity follows the order of Ni/Al2O3 < Ni/CaO < Ni/ZSM-5 < Ni/Clinoptilolite. EDAX result well demonstrated, that due to the natural source of the clinoptilolite, it has 8.0% other elements (Fe, Ca, K, Na) in addition to those listed. It is also worth to mentioned, that the nickel content of the catalysts changes significantly, which could be explained by the difference surface properties of the applied supporters.The desorption of ammonia is often used in order to determine the strength and amount of acid centres. The strength of the centres correlates with the desorption temperature. Since strongly bound probe molecules have high binding energies, increases temperatures are necessary to desorb these adsorbates [18]. The NH3-TPD was carried out on four samples (fresh catalysts) under same condition. Tmax values of TPD plots of ammonia desorbed from fresh catalysts decreases the in the following order: Ni/CaO > Ni/Al2O3 > Ni/Clinoptilolite > Ni/ZSM-5. However, the number of acid sites of fresh catalysts decreases as follows: Ni/Clinoptilolite > Ni/CaO ≈ Ni/ZSM-5 > Ni/Al2O3. However, it is important to be mentioned, that in case of CaO the ammonia TPD results are not concrete, due to the fact that CaO is a basic oxide. The high acidity of Ni/CaO can be explained by the measurement conditions. During the ammonia TPD, the number of the acidic sites are calculated by the consumed ammonia, which in case of CaO, resulted in the following chemical reaction: (1) 3CaO + 2NH3 → 3Ca + N2 + H2O Basically, the mentioned reaction requires higher temperatures (∼500–700 °C), but owing to the nickel content, lower temperature was adequate (308 °C). Fig. 1 shows the morphology of prepared catalysts impregnated with nickel. Based on their structure, rings and channels of the catalysts, their nickel content evolved differently. The CaO had the highest nickel content (9.2%) owing to its wide channel openings and active sites, while clinoptilolite had the smallest (1.9%) due to its lower surface area and pore volume. Besides, as it can be seen, the distribution of nickel is not unified, which also can be explained by the different pore sizes, average pore diameter of catalysts and the wet impregnation method. Afterwards, it is important to be mentioned, that the surface area of Ni/Al2O3 is significantly lower comparing the Al2O3, due to the impregnation, which could clogged its pores and covered the surface of catalyst.The experiments were carried out in a two zone tubular reactor (Fig. 2 ), aiming the in-situ carbon dioxide capture with increased synthesis gas yield. In the first reactor zone 5g of the raw material was placed, while 2.5g catalyst (Ni/ZSM-5, Ni/Al2O3, Ni/CaO, Ni/Clinoptilolite) was used in the second zone. Firstly, the temperature of the 1st reactor zone was determined, then that of the 2nd was investigated between 500 and 700 °C. Regarding the final temperature of the 2nd reactor zone, it was chosen with the consideration of the regeneration temperature. It is important to mention, that higher temperature than 800 °C for regeneration could cause structure deformation in case of catalysts. Afterwards the determination of the temperatures in the reactor zones, the regeneration cycles (during 10 cycles) of each catalyst was investigated for economic reasons. In case of preserving the structure of catalysts and its decoking, the regeneration was carried out at 800 °C for 1 h. The measurements were performed under inert conditions (nitrogen flow, 42 ml/min), for 20 min. In order to capture the moisture of the gas silica gel, while for the gaseous product a Tedlar type gas bag was used, with only one sampling at the end of the measurement. At the end of the measurements, the product yield was calculated based on their weight balance (2). (2) 100 − R e s i d u e ( g ) + G e n e r a t e d w a t e r ( g ) R a w m a t e r i a l ( g ) ∗ 100 = P r o d u c e d g a s ( % ) The composition of the gas products was investigated using a DANI type gas chromatograph equipped with a programed injector and a flame ionization detector. Rtx-1 PONA type 100 m long column with an internal diameter of 0.25 mm and film thickness of 0.5 μm was placed in the chromatograph. The analysis was performed at 35 °C isothermal condition. The detector and injector temperatures were 230 °C. The chromatograms were evaluated using Clarity software.The hydrogen content of the gaseous products was also determined by gas chromatography using a DANI type gas chromatograph (with TCD detector) equipped with a CarboxenTM 1006 PLOT (30 m × 0.53 mm) column. During the experiments the following temperature program was used, the column space temperature at 30 °C for 18 min was kept, then it was raised to 120 °C with a heating rate of 15 °C/min, then the temperature was maintained at 120 °C for 2 min.To investigate the weight loss of raw material, a thermogravimetric analysis was performed by a Netzsch thermogravimetric analyser. During the measurement nitrogen atmosphere was used, with 20 °C/min heating rate until 800 °C. The arisen gases were analysed by a Bruker type FTIR connected to the TGA. The weight loss and dm/dt result of biomass raw material is shown in Fig. 3 .The decomposition of the raw material took place in several stages, resulting in a residue of 45.73%. As Fig. 1 shows, the degradation of the biomass took place in three main steps. Up to 135 °C, the physically bound moisture of the sample was removed (4.06%). The first decomposition step took place between 135 °C and 245 °C with a maximum of 225 °C. The weight loss rate was 6.36%. The second step was observed between 245 °C and 340 °C with a maximum of 325 °C, while the third decomposition step was observed between 325 °C and 405 °C. The last decomposition step was observed at a maximum of 405 °C. The weight loss of the sample was 19.62% in the second decomposition step and 15.11% in the third. In the first degradation step, mainly the lighter molecular weight volatiles was removed, while in case of the third the hemicellulose units started to degrade, while that of the second degradation step both the hemicellulose and cellulose units. At the last step, mainly the lignin units started to decompose, due to its phenolic polymer structure.The TG-FTIR provides information about the functional groups of volatiles including non-condensable gases, (CO, CO2, CH4) and condensable volatiles (H2O, methanol, acids, phenols) as a function of temperature. As it can be seen in Fig. 3, the raw material degraded as it was found during pyrolysis of lignocellulose materials [15]. The carbon monoxide mainly arises at low temperature (∼350–450 °C) from the cracked ether and carbonyl groups, while at higher temperature (>600 °C), it can mainly be originated from the secondary reactions [15]. At low temperature carbon dioxide arises from hemicellulose, while that of at higher temperature from lignin units. The mentioned CO2 peaks can be observed at 2356 cm−1 and between 668 and 500 cm−1. The different stretching and bonds were revealed in the first 1000 s, except the characteristic peaks of CO2. Regarding the H2O, the peaks are caused by the evaporation of moisture content and dehydroxylation of carbohydrates [15]. O–H stretching of H2O between 4000 and 3468 cm−1, while at 1645 cm−1 H–O–H bending can be detected. The same conclusion can be noted regarding the C=O aldehyde and ketone stretching and C–O–C stretching which appeared at 1830-1650 cm−1 and at 1100 cm−1, respectively. Methane tetrahedral υ4 vibration appeared at 1306 cm−1 with low absorbance, while asymmetrical C–H bending and stretching can be observed at 1500 cm−1, which can be originated from the lignin content. Over time, the mentioned groups cannot be detected excepting the characteristic peaks of CO2.In order to get the appropriate temperatures during the experiments, preliminary experiments are needed. At first, the temperature of the 1st reactor zone should be chosen, considering the yields (Fig. 2). As Fig. 3 shows, with the increase of the temperature the yield of gas was escalated while that of the residue was decreased. Besides, in the range of 400–700 °C the yields was not significantly changed. That phenomenon can be explained by the degradation of the lignocellulosic units. As it is widely known, the biomass is built up from cellulose (40–60%), hemicellulose (15–30%) and lignin (10–25%), having different degradation temperature ranges regarding its component [19]. A significant difference cannot be remarked between 500 and 700 °C, however, at higher temperatures, the yield of gas was increased by 10.6% due to the decomposition of lignin units [19]. Nevertheless, during biomass degradation, water formation takes place, which in case of the mentioned experiments was under 1%. Based on the mentioned low value, the amount of water is not depicted on Fig. 4 .Regarding the gas products, as Fig. 5 depicts, with increasing temperature, the yield of hydrogen, carbon monoxide and methane was escalated mainly owing to the following reactions [20]: (3) CnHm→(m/4)CH4+(n-m/4)C (4) C + CO2⇄2CO (5) CO + H2O⇄CO2+H2 (6) CO+3H2⇄CH4+H2O (7) CnHm + nH2O→(m/2+n)H2+nCO2 (8) CnHm + nCO2→2nCO2+(m/2)H2 Besides, the amount of carbon dioxide and lighter hydrocarbons was decreased, due to the thermal cracking (3), the steam reforming (water has arisen from the moisture of biomass) (7) and dry reforming reactions (8) (carbon dioxide surplus from biomass, even at higher temperatures) [20]. Regarding the amount of carbon monoxide and lighter hydrocarbons, the raise of the temperature was not resulted in significant change, while that of hydrogen, methane and carbon dioxide was remarkable with 1.1–19%, 2.2–8.5% and 5.8–28.8%, respectively. Since our work is mainly focusing on the regeneration cycles of the used catalyst and its properties in carbon dioxide conversation, in the 1st reactor zone lower temperature, 400 °C was used in the following measurements, where the carbon dioxide yield is ∼55%. Besides, the process has lower energy requirement, also, it can provide a possibility for further investigations on the effect of the temperature.After the determination of the temperature in the 1st reactor zone, the temperature of the 2nd was also investigated with and without catalysts between 500 and 700 °C (Fig. 2). As Fig. 6 depicts, the gas yield was increased by the enhanced temperature and with the presence of catalysts. However, it should be mentioned that the differences between the yields are lower than 5%, which can occurred by the diverse particle size of raw material (<3 mm).Also, it is important to note, that remarkable difference could not be observed among catalyst free and thermos-catalytic cases (only 1.2–7.5% of change). Besides the yield, the composition of the gas (Fig. 7 ) was also investigated in the presence of the four different catalysts. In the absence of catalyst the yield of hydrogen and carbon monoxide were increased by the escalated temperature in the 2nd reaction zone, while that of lighter hydrocarbons was decreased by 2.6%. Regarding the results, mainly the same tendency can be observed at all temperatures, while significantly the amount of the components changed. Fig. 7 shows the change in the amounts of components compared to the non-catalytic measurement (Fig. 7(a)) at 500 °C (Fig. 7(b)), 600 °C (Fig. 7(c)) and 700 °C (Fig. 7(d)). In case of hydrogen, carbon monoxide and carbon dioxide yield remarkable changes were observed at both temperatures. The hydrogen yield was increased by 0.3–6.9 mmol/g due to the higher temperature and the nickel content on the catalysts surface. In the presence of catalysts the carbon monoxide yields were enhanced by 3.8–13.8 mmol/g, which can be explained by the shifting in Boudouard reaction (4), water gas reaction (5) and reforming reaction (7). Latter can be observed with the lower methane and higher hydrogen yields in almost each case. Nevertheless, the mentioned Boudouard reaction (3) was only observed in the presence of catalysts, where coke deposition was appeared on its surface promoting the solid-gas reaction.The effect of carbon dioxide capturing and decreasing was detected in case of Ni/Al2O3, Ni/CaO and Ni/Clinoptilolite with 0.3–23.7 mmol/g values. Based on the literature, in case of Ni/Al2O3 the reduction in the yield of CO2 can be described with its specific surface area and crystallite size [21,22]. However, as it was mentioned before, during nickel impregnation, the specific surface area of the Al2O3 has decreased significantly, therefore, the CO2 reduction besides the properties of alumina-oxide, was mainly occurred by the higher temperature and its Ni content which can also enhance the CO2 adsorption capacity, as well as the adsorption activity. Besides, as it also can be mentioned in case of Ni/CaO, chemical adsorption can occur due to the hydroxyl and oxide groups on the surface resulting in carbonate or bicarbonate species [21,22]. Regarding the Ni/Clinoptilolite, the high acidity can be mentioned which can evolve the CO2 capturing effect, therefore CO2 can be easily adsorbed owing to its beneficial properties (high quadruple moment) next to the hydrogen, methane and lighter hydrocarbons [23]. As Fig. 7 depicts the amount of the lighter hydrocarbons was decreased by 1.1 and 10.9 mmol/g, from which the highest value belongs to the Ni/ZSM-5 (8.1–10.9 mmol/g) owing to its high Si/Al ratio, its pore structure (narrow zig-zag pattern, limited diffusion) and better cracking function.In terms of the generated syngas the presence of different catalysts were also investigated. As it can be seen in Fig. 8 , with increasing temperature, the sum of hydrogen and carbon monoxide also increased. Besides, not only the nickel content on the catalyst surface but the carbon capturing/conversion had a positive effect as well as temperature on syngas yield. As it was mentioned before, the highest hydrogen yield (8.9 and 11.1 mmol/g respectively) was generated in the presence of Ni/CaO and Ni/Clinoptilolite, while the highest carbon monoxide was obtained in the presence of Ni/Clinoptilolite. Referred to the results, Ni/Clinoptilolite provided the highest syngas yield at all temperatures (37.4, 38.6 and 48.3 mmol/g). Furthermore, it should be mentioned, that in syngas yield, compared to the non-catalytic points, a significant difference was observed (4.1–20.2 mmol/g). Based on the results, hereinafter 700 °C was used in the 2nd reactor zone, with 400 °C in the 1st reactor zone. Fig. 9 shows the difference in the amount of each component compared to the first measurement point. As it is depicted, the amount of the lighter molecular weight hydrocarbons was increased by each catalyst in a wide range (0.7–11.4 mmol/g) from which Ni/ZSM-5 had the strongest influence in each regeneration cycle. This phenomena can be explained by coking and continuous pore clogging despite the regeneration cycles. Nevertheless, it should be mentioned that Ni/ZSM-5 and Ni/Al2O3 decreased the carbon monoxide and carbon dioxide amount with a significant value (0.5–8.1 and 4.5–11.8 mmol/g, respectively) as the cycles were progressed.Regarding the Ni/CaO, the amount of carbon dioxide until the third regeneration cycle was decreased by 0.2–1 mmol/g, thereafter it was increased by 1.4–4.9 mmol/g. On the basis of earlier work, it can be explained by escalated specific surface area owing to the catalyst sintering effect caused by the severally used high temperature [26]. Besides, the amount of hydrogen and carbon monoxide was reduced by 2.1–7.1 mmol/g and 3.8–8.3 mmol/g, respectively. However, with the decreasing of hydrogen, the methane was increased by 0.1–2.3 mmol/g, which can be explained by the methanation reaction (6). In the presence of Ni/Clinoptilolite the hydrogen and carbon dioxide content was decreased at each regeneration cycle by 1.0–7.1 mmol/g and 0.3–4.4 mmol/g, while carbon monoxide and methane content was increased (1.4–3.7 and 0.3–2.2 mmol/g). Concerning the amount of carbon monoxide, its value from the seventh regeneration cycle was significantly reduced by 4.5–9.2 mmol/g. The introduced phenomena can be explained by the favourable acidic properties of Clinoptilolite in carbon capturing, while the remarkable changes in composition from the seventh regeneration cycle may occurred by the coke deposition in the tetrahedral structure and on the surface of catalyst [23,24]. Fig. 10 shows the yield of the generated syngas at each regeneration cycle, where its yield was changed between the range 24.9–50.5 mmol/g. In the presence of Ni/ZSM-5, Ni/Al2O3 and Ni/CaO a slight decrease can be observed along with the regeneration cycles. However, in case of Ni/Clinoptilolite the synthesis gas yield was escalated until the sixth cycle (43–50.5 mmol/g), despite that it had the highest value from 7th-10th regeneration cycle compared with the used catalysts. Based on the results, it can be concluded, that the Ni/ZSM-5 and Ni/Al2O3 decreased the amount of carbon dioxide until the 5th regeneration cycle, while in case of Ni/CaO its value was reduced until the 3rd cycle. In the presence of Ni/Clinoptilolite the carbon dioxide content showed decreasing even at the last regeneration cycle. Due to the represented data, the Ni/Clinoptilolite resulted in the highest synthesis gas and the lowest carbon dioxide yield, while in case of Ni/Al2O3 and Ni/CaO the generated carbon dioxide has remained under 30 mmol/g for ten regeneration cycles.3.5 Morphology of used catalysts.Afterwards the regeneration cycles, the main properties of the catalysts was investigated, which are summarized in Table 2 . During catalytic pyrolysis-gasification, the deactivation of catalysts strongly depends on the secondary reactions, as well as coke deposition. It should be mentioned, that through the thermochemical degradation of biomass oxygenates such as phenols, alcohols, ketones and aldehydes could formed, which are at higher temperatures and in the presence of catalysts start to decompose with the formation of coke. The generated coke clogs the pores and covers the specific surface area, causing activity decreasing. Therefore, the regeneration of catalysts is necessary.As it can be seen the surface areas (except Ni/Al2O3) was decreased by the coke deposition, which caused clogging in the micropores of Ni/ZSM-5 and Ni/Clinoptilolite and pores at about 4 nm of Ni/CaO. Comparing the fresh catalysts (Section 2.2, Table 1), with the coked and regenerated ones, it can be said, that the highest and lowest decrease of SBET values were observed for Ni/CaO coked (54%) and Ni/ZSM-5 coked (17%) respectively. Also, slight reduction was observed in the Si/Al ratio and Vmicro in case of Ni/ZSM-5 and Ni/Clinoptilolite which can be explained by the partly blocked and inaccessible acidic centres. Besides, in case of Ni/CaO it can be mentioned, that the fresh catalyst did not have micropores, while that of the coked and regenerated were occurred, which also can be clarified with the increased average pore diameter values. This phenomenon can be explained with sintering effect which was caused by the numerous use at 700 °C, as well as with the several regeneration cycles at 800 °C. However, the mentioned effect was occurred until the 6th-7th regeneration cycles. In addition, the regeneration was considered effective due to the lower amount of coke deposition, which was removed almost totally in case of Ni/Al2O3, Ni/ZSM-5 and Ni/Clinoptilolite. Meanwhile no significant regeneration effect was observed after 10th cycle at 800 °C on surface area, pore volume, pore size distribution. The morphological parameters showed almost the same values for “coked” and “regenerated” samples.In this work, agricultural biomass was pyrolysed-gasified in a two-zone tubular reactor. As a catalyst, nickel impregnated ZSM-5, Al2O3, CaO and Clinoptilolite were used. The main aim was the reduction of carbon dioxide with the increase of syngas yield. During the measurements, not only the effect of the temperature and catalysts but the regeneration cycles of catalysts was investigated. At first, the temperature of the 1st and the 2nd reactor zone was determined, where 400 °C and 700 °C, while for the temperature of regeneration 800 °C was chosen. Regarding the 1st zone, in case of great amount of carbon dioxide production, as well as to investigate the CO2 reduction and conversion effect of catalysts, low temperature was chosen. Besides, with the use of low temperature, the process is more energy efficient.In case of all catalysts, the amount of lighter hydrocarbons was increased especially in the presence of Ni/ZSM-5 which was caused by coking and pore-clogging. In the presence of Ni/Al2O3 mostly the temperature had an effect on the product yield owing to the low SBET surface of the catalyst which was mostly caused by the nickel impregnation. Regarding the Ni/CaO it can be mentioned, that due to the 10 regeneration cycles a sintering effect was noticed where a small amount of micropores was generated increasing the SBET of Ni/CaO. Concerning the results obtained in the presence of Ni/Clinoptilolite, it can be mentioned that the highest synthesis gas yield was observed until the 6th regeneration cycle, due to its great quadrupole moment which helps in carbon dioxide capturing, with the increasing of carbon monoxide. Based on the results, it can be noted that the amount of carbon dioxide was decreased until the 5th and 3rd regeneration cycle (in case of Ni/ZSM-5, Ni/CaO respectively), while with the presence of Ni/Clinoptilolite the decreasing can be observed even at the last regeneration cycle. Therefore, the Ni/ZSM-5, Ni/CaO can be used for 10 regeneration cycles with low activity change, however, not for syngas yield increasing in these measurement series.Afterwards, it was stated, that the Ni/Clinoptilolite is suitable for CO2 capturing with syngas yield enhancing for 10 regeneration cycles. However, in case of Ni/Clinoptilolite, longer regeneration cycles should be investigated in case of more information of activity changing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This project has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 823745.

In this work, agricultural biomass waste (maize) was used as raw material in a pyrolysis-gasification process with the presence of nickel-loaded zeolites, CaO and Al2O3. Henceforward, the reusability and the deteriorating effect of the catalysts are important to be analysed, therefore the catalysts were reused for ten regeneration cycles. During the measurements, the effect of temperature, catalysts and the regeneration cycles of catalysts were also investigated, as well as the components of the gaseous product. At first, the proper temperature was determined in the 1st (200–800 °C) and the 2nd (500–700°) reactor zone, where 400 °C and 700 °C, while for the regeneration 800 °C was chosen. Throughout the regeneration cycles in case of Ni/ZSM-5 lighter hydrocarbons was increased while that of the carbon monoxide and carbon dioxide was decreased. In the presence of Ni/CaO the amount of carbon dioxide decreased until the 3rd regeneration cycle, while carbon monoxide and hydrogen was decreased till the 10th cycle. Regarding the Ni/Al2O3, the amount of CO2 and CO was significantly decreased at each cycles, while in case of Ni/Clinoptilolite reduction was observed in hydrogen and carbon dioxide, however not only the amount of carbon monoxide but the syngas yield increased remarkably. Due to the mentioned results, the Ni/Clinoptilolite has an effective syngas enhancing and carbon dioxide capturing effect, even with low pyrolysis temperature.

Data will be made available on request. Data will be made available on request.There are several serious disadvantages to use fossil fuels. These are especially worrying for the environment (global warming) and for the energy independence of nonproducing countries. The transport sector is one of the main consumers of fossil fuels and a key point for the decarbonization strategies of economies. Thus, there is great interest in the production of alternative fuels based on renewable sources and residues.The hydroconversion of vegetable oils, fats and used vegetable oils and Fischer-Tropsch synthesis are two routes for the synthesis of clean fuels. However, both processes yield a mix of linear hydrocarbons that cannot be used directly as fuels; in both cases, a hydroisomerization step is necessary to produce green fuel with physicochemical properties to be used as fuels [1]. Hydroisomerization is a process for converting linear paraffins (n-alkanes) into branched isomers [2]. This process is needed to produce high-quality gasoline (clean gasoline) with improved octane number and jet/diesel fuels as well as lubricant oils with improved low-temperature performance to comply with the cold flow properties of diesel fuel specifications such as the pour point (low-temperature freezing) and viscosity index [2–4]. In addition, an increasing proportion of isomerized gasoline in the gasoline pool contributes to environmental protection and product upgrades [5].The hydroisomerization process is generally performed on bifunctional catalysts with metallic sites for hydrogenation/dehydrogenation and acidic sites for isomerization through carbenium ions [3,6,7]. Among several options, WO 3 supported in Al2O3 or ZrO 2 is a good option for the acidic function, including noble metals in the metal components (platinum, palladium), with Pt being the most active [3,5]. Based on the good previous results we obtained on similar systems, we selected Al2O3 as a support and WO3 as an acidic functional group [1,8]. The proper acid/metal content is decisive in achieving high activity, stability and product selectivity of these catalysts and in obtaining ideal hydroisomerization-cracking behavior of hydrocarbons [4,6,7]. The high cost of noble metals and their limited availability limit their use and applications [3,5]. The hydroisomerization performance (catalyst activity and isomerization selectivity) of bifunctional catalysts is influenced by another important factor: the acidity and pore structure of the support. Hydrocracking is promoted by a strong acidity, whereas high isomerization selectivity is obtained with medium strength acidity [3]. In addition, the hydrocracking of monobranched isomers is more difficult than that of multibranched isomers.Nickel-supported catalysts are used in the oil refining industry to produce high-quality gasoline and diesel fuels and lubricating oils [4] because among the nonnoble metals, nickel has shown the best catalytic activity for hydroisomerization of n-alkanes [2]. These catalysts cost less than noble metal catalysts, are less likely to poison, and have more availability, but the high tendency for hydrocracking reduces the hydroisomerization yield [3,4]. Other disadvantages are the tendency to deactivate by coking and the excess cracking producing a large amount of gasses [4]. The preparation of nonnoble metal catalysts (for instance, nickel) is important to avoid large metal particle formation, however, the high metal loadings are necessary due to the low specific activity of these metals with respect to the noble metals. This problem restricts the optimization of the existing nonnoble catalysts [5]. This means that there is a need for high nickel loading to be active. An increase in the metal particle size is produced by this loading, and pore opening could be restricted [4]. Therefore, it is an appealing task to look for a novel strategy to prepare nonnoble metal-based catalysts with reduced particle size for the hydroisomerization reaction [5]. This means that novel bifunctional catalysts with improved stability and diffusion properties can improve the capacity in the conversion of heavy paraffin fractions [4]. To our knowledge, there have been a few reports about the influence of nickel loading on the hydroisomerization of n-dodecane. Some groups have studied higher nickel content [9,10] than we reported here. In a previous work [8], we studied n-dodecane hydroisomerization with zirconia- and alumina-supported Pt.The present work introduces Ni as a metal to replace Pt in alumina-supported WO3 catalysts for the hydroisomerization of n-dodecane due to the high cost and limited availability of noble metals previously mentioned. However, the low activity of Ni compared to that of Pt, makes that nickel still cannot compete with Pt when the general economy of the process is considered. It seems that the problem of the low catalytic activity could be easily resolved by increasing the Ni loading on the support since nickel is approximately 1000 times cheaper than Pt [2].The employed tungsten precursor was ammonium metatungstate hydrate ((NH4)6(H2W12O40)·xH2O) (99%) purchased from Honeywell, and the nickel precursor was nickel(II) nitrate hexahydrate (Ni(NO3)2·6H2O) (99%) from Sigma-Aldrich. Alumina (γ-Al2O3) was purchased from Saint Gobain-NORPRO (1.5 × 4 mm trilobes, SA6975) and used as support.The catalysts were prepared by applying the wetness impregnation method on the supports to introduce W and Ni. In order to remove the moisture of the support, it was treated overnight at 120 °C before any impregnation.The following procedure was used to incorporate W in the γ-Al2O3 pellets by wet impregnation: a round flask was used to place the pellets in contact with an aqueous solution of ammonium metatungstate hydrate ((NH4)6(H2W12O40)·xH2O) (20 ml for 6 g of Al2O3). The mixture was stirred for 1 h, followed by the evaporation of the solvent under reduced pressure for 20 min. Finally, the solid was calcined in an oven at 500 °C for 2 h [8].The following protocol was used to incorporate Ni by wet impregnation: a round flask was used to place the pellets in contact with an aqueous solution of nickel (II) nitrate hexahydrate (Ni(NO3)2·6H2O) (20 ml for 6 g Al2O3). Pellets and solution were stirred for 1 h after that time, the solvent was eliminated by evaporation at reduced pressure for 20 min. Finally, the solid was calcined at 400 °C for 2 h in an over under static air conditions.Following these procedures, the catalysts were prepared and labeled as Al2O3, WO3-Al2O3, and Ni/WO3/Al2O3= NiW/Al-[wt%] catalysts, where [wt%] indicates the wt% of Ni on the catalysts. The W loading was 15 wt% in all catalysts and the WO3-Al2O3 sample.The textural properties were determined with a Micromeritics ASAP 2420 from the adsorption–desorption experiments [1,8]. The X-ray diffraction patterns were recorded with a X'Pert Pro PANalytical to identify the crystalline phases in the catalysts. The equipment have a CuKα radiation source (λ = 0.15418 nm) and X'Celerator detector based on RTMS (Real Time Multiple Strip) [1,11]. Temperature-programmed desorption and the acidity of the catalysts were performed on a Micromeritics TPD/TPR 2900 apparatus with a thermal conductivity detector (TCD) [8]. The nature of the acidity (Brønsted and Lewis acid sites) was determined by in situ IR spectroscopy of chemisorbed pyridine. The measurements of the IR spectra have been done in a DRIFT cell installed in a 6700 Nicolet FTIR spectrometer [1]. XPS spectra were obtained with a SPECS GmbH electron spectroscopy system with a PHOIBOS 150 9MCD energy analyzer, and Mg X-ray source [8,12].The metal contents of the catalysts were determined by elemental analysis via ICP–OES [8].The hydroisomerization performance of n-dodecane was tested with the prepared catalysts. A trickled-bed mode reactor worked in parallel flow and at high pressure to ensure close contact between the gas, the liquid, and the solid. The calcined catalyst pellets (1 g) diluted in 8 g of inert support were reduced in the reactor at 425 °C and atmospheric pressure. After the reduction, the reaction conditions were set to pressure of 2.0 MPa, reaction temperature of 300 °C, and the reactor was fed by a liquid flow of 0.1 mL·min−1 of n-dodecane and hydrogen flow of 340 mLN·min−1. The gas outlet was measured online with a µ-GC and the liquid samples were collected and analyzed offline. The time-on-stream was 24 h.More detailed information about the catalyst characterization methods and reaction conditions is included in the supplementary information.The chemical compositions of the catalysts were confirmed by ICP analysis, as shown in Table 1 . The measured chemical composition is very close to the nominal values.The textural properties were studied by N2 adsorption-desorption. Table 2 shows the BET surface area, average pore volume and average pore diameter of alumina, support, and catalysts. The surface areas decrease firstly when WO3 is incorporated by impregnation onto the support (Al2O3) to obtain WO3-Al2O3. Then surface areas decrease more when nickel is impregnated onto WO3-Al2O3. The support surface is partially covered by the incorporation of WO3 and NiO as particles which block the support pores and reduce nitrogen access and adsorption capacity. But the variation with respect to WO3/Al2O3 is small, indicating that all catalyst samples have similar textural properties.The porosity of the samples was studied with the N2 adsorption-desorption isotherms of the support and catalysts. According to the IUPAC 2015 classification or Brunauer, Deming, Deming, and Teller (BDDT) classification for mesoporous materials all isotherms can be classified mainly by type IV (Fig. 1S) with hysteresis H1 due to mesoporous aggregates and in the high P/Po there is a small contribution of type II character due to unfilled macroporosity, since the isotherms do not level off entirely. Thus, the materials are formed by mesoporous and somewhat macroporous aggregates [13–18]. Isotherms are very similar in the support and catalysts indicating a high dispersion of the tungsten and nickel active phases on the support. The pore size distributions (Fig. 2S) were located in the mesopore range from 2 to 50 nm for the support and catalysts.The crystalline phases present in the catalysts were studied by X-ray diffraction. All PDF cards belong to ICDD database. The XRD profile of Al2O3 presents three main peaks located at 37.1° (with a shoulder at 39.4°), 46.0° and 66.7°, corresponding to the (110), (111) and (211) planes of γ-Al2O3, respectively, of the cubic crystal system (PDF card 00-001-1303). The WO3/Al2O3 sample profile shows the same peak structure as the Al2O3 sample. Nonobvious diffraction peaks were found for WO3 (PDF card 00-005-0388) or other tungsten oxides which indicates a high dispersion of tungsten oxide in the support and/or a domain size smaller than ∼2 nm for WO3 crystallites [19]. The structure of tungsten species dispersed on oxide supports depends on both the nature of the support and the concentration of tungsten [20].The X-ray profiles of the Ni catalysts show the same diffraction lines from the support except for the NiW/Al-10 catalysts (Figs. 1 and 3S), which shows a new peak at 43.3° due to NiO associated with the (012) plane of the rhombohedral crystal system (PDF card 00-044-1159). It can be deduced that there is a high dispersion of NiO on the surface forming particles that are too small to diffract except for the catalyst with the highest loading (10% wt Ni).The TPR profiles are compiled on Fig. 2 (A and B). Fig. 2B is an enlargement of Fig. 2A up to 500 °C. As the nickel content increases, the reduction peak due to NiO (approximately 400 °C) is shifted to lower temperatures. This is because with low Ni concentrations, NiO particles are smaller, meaning that more species are in close contact with the support (greater interaction) and therefore are more difficult to be reduced. This is also reflected in the XRD diffractograms of the catalysts, which indicate that catalysts with higher Ni loadings present a lower dispersion. In the reduction profile of the NiW/Al-10 catalyst, there is a second peak at approximately 425 °C (Fig. 2B) that is due to NiO reduction to metallic Ni when nickel oxide interacts with WO3 sites. There is a second peak at high temperatures (750 °C) due to partial reduction of tungstate oxide. In the WO3-Al2O3 support, the tungsten oxide band at high temperatures is not visible. This is because metallic nickel activates hydrogen, and through the spill-over process, it facilitates tungsten oxide reduction. Considering these results, 425 °C is the reduction temperature for the activation of the catalysts before the reaction because all NiO will be reduced to metallic Ni.The acidity is an important factor that determines the reactivity of the catalysts used in the hydroisomerization reaction. NH3-TPD was applied to study the acid properties of the catalyst surface, evaluating the total acid site amount, and the distribution of acid sites by strength. Ammonia is a suitable probe molecule for acidity because its small size and basicity allow the interaction with most of the acid sites. The temperature of desorption is indicative of the acid site strength, while the amount of adsorbed ammonia is proportional to the number of acid centers [21].The NH3-TPD profiles for all catalysts are shown in Fig. 3 . There are three different desorption peaks corresponding to different acid strengths: (a) desorption a lower temperature than 250 °C: weak acid sites, (b) desorption temperature from 250 to 400 °C: intermediate strength and (c) desorption at higher temperature than 400 °C: strong. WO3-Al2O3 exhibits two main desorption peaks at 155 and 555 °C corresponding to weak and strong acid sites. WOx species appear to interact strongly with sites on the surface of γ-Al2O3 [22]. At low nickel loadings (0.5 wt.% Ni), the catalysts have features similar to those of WO3-Al2O3. When nickel loading increases, the low-acid site peaks are moved to higher temperatures, and medium-acid sites appear [9]. All catalysts show a peak (520–550 °C) of strong acid sites, which is divided into two peaks at high nickel loadings (6–10 wt%), appearing as a new peak at approximately 470 °C. The acidity was introduced through WOx, so it was expected that all samples show a similar value of total acidity because all samples had a comparable W loading (ca.15% wt). This value was measured and was approximately 1.6 mmol NH3/g.The nature of the surface acid sites was studied of IR of adsorbed pyridine. IR spectroscopy of adsorbed pyridine facilitates the distinction of different acid sites. The catalysts were reduced at 425 °C and then the FTIR pyridine (DRIFT) adsorption spectra (Fig. 4 ) were measured at room temperature.The FTIR spectra show adsorption bands centered at approximately 1610, 1575 and 1448 cm−1, which correspond to adsorbed pyridine on Lewis acid sites (L) [23,24]. The alumina support has three possible Al3+ coordination with Lewis acidity: five, four and three [23]. A small band at 1540 cm−1, which is not visible in some catalysts, can be attributed to vibration modes of pyridinium ions on Brønsted acid sites (B) [24]. The nonobvious band at 1540 cm−1 for some catalysts is ascribed to the coverage of Brønsted acid sites with NiO at high content [3]. The absorption band at approximately 1490 cm−1 is due to a combination of signals of pyridine absorbed on Lewis or Brønsted acid sites. Brønsted acid sites are due to the reducible domains that act as redox sites required for the formation of H+species from H2 [25]. A shoulder at 1622 cm−1 is observed in the samples without nickel or with low nickel content, whose intensity decreases with increasing nickel loading. This signal corresponds to adsorbed Py on very strong Lewis acid sites, which are considered tetrahedral sites with cationic vacancies in close proximity. These species are covered by nickel species [26].The nature and dispersion of nickel and tungsten species on the surface of the catalysts were studied by XPS analysis. Fig. 5 shows the Ni 2p XPS spectra for the NiW/Al-2 to 10 catalysts, Fig. 4S shows the W 4f XPS spectra for the NiW/Al-0.5 to 10 catalysts and Table 3 lists the binding energies (eV) (W 4f7/2 and Ni2p) and atomic surface ratios of the present elements. Al 2p signal at 74.5 eV [8] was the reference for the binding energies to correct for charging effects.In Fig. 5, the signal at approximately 856.7 eV is attributed to Ni2+species (NiO) and are accompanied by a satellite signal at approximately 862.9 eV, characteristic of this oxidation state [4,27,28]. The peak due to metallic nickel is not found because metallic nickel is formed in the reduction process before the reaction. Prior to this reduction, nickel is in the oxide form (NiO).The W 4f signal presents two contributions of the typical doublet corresponding to spin-orbital splitting (W4f7/2 and W4f5/2). One component is attributed to the WO3 species at approximately 35 eV for W4f7/2 [8,25] and 38.2 eV for W4f5/2. A second component at approximately 36 eV for W4f7/2 and 37.3 eV for W4f5/2 can be assigned to Al2(WO4)3, aluminum tungstate [8,29].The surface atomic ratios of Ni/W or Ni/Al generally increase with Ni loading (Table 3), but this increase is lower than the corresponding increase in the nickel bulk concentration. This implies a decrease in the nickel dispersion as the nickel loading is higher. The most evident case is the absence of change in the Ni/Al surface atomic ratio between NiW/Al-8 and NiW/Al-10 catalysts, a clear indication of the formation of bulky crystal structures of nickel species as was detected by XRD. The Ni/W surface atomic ratio is higher than the Ni/W bulk atomic ratio. This effect can be due to the preferential deposition of nickel on tungsten moieties, which reduces the W signal with respect to the expected chemical composition. The lower (Ni/W) xps/(Ni/W) bulk ratio is for the NiW/Al-10 catalyst in which the nickel species start to form bulky structures, implying a lower Ni coverture of W with respect to the other catalysts without the development of bulky structures of nickel species.In conclusion, the characterization techniques applied for studying the catalysts, i.e., tN2 adsorption-desorption, XRD, H2-TPR, FTIR, NH3-TPD and XPS, indicated that there is a high dispersion of NiO and WO3 on the support surface revealed by the absence of NiO and WO3 diffraction peaks except for the NiW/Al-10 catalyst, which shows NiO peaks seen in XRD analysis. The XPS data also showed a decrease in the nickel dispersion with increasing nickel loading, indicating the formation of bulky crystal structures of nickel species, as was detected by XRD. Nickel deposits preferentially on tungsten moieties as the surface Ni/W atomic surface ratio is higher than the Ni/W atomic bulk ratio. The similarity between the support and catalyst textural properties in combination with XRD and XPS results indicates a high dispersion of the tungsten and nickel active phases on the samples. Medium acid sites (moderate acidity) are the main sites responsible for the reaction performance [30–32]. FTIR spectra show that more Lewis acid sites are present than Brønsted acid sites. It is worth noting the shoulder at 1622 cm−1 observed in the samples without nickel or low nickel loading corresponds to adsorbed Py on very strong Lewis acid sites. These species disappear because they are covered by the presence of nickel species.The hydroisomerization of n-dodecane was tested with the series of NiW/Al-x catalysts (x = 0.5, 1, 2, 4, 6, 8, 10). The time on stream was 24 h and the displayed results (conversion and selectivity) are the average of these 24 h. The samples were first reduced at 425 °C, this reduction temperature was deduced from TPR results since at 425 °C NiO will be reduced to metallic Ni. The experimental results show that all catalysts are active for the hydroisomerization of n-dodecane (Treaction = 300 °C, P = 2.0 MPa). The GC analyses show that the products contain iso-dodecane (i-C12), branched C12 and cracked products (C6-10) under the applied reaction conditions. Fig. 6 shows the n-C12 conversions and (i-C12+branched C12) yields for all catalysts studied.The conversion of n-dodecane and selectivity to branched-C12 tends to increase with Ni loading until the catalyst contains 6 wt% Ni. This catalyst shows the highest conversion (28%) obtained, and the XPS analysis showed that this catalyst has a higher Ni loading, and that the Ni/Al surface ratio increases before stabilizing (Table 3). Above this Ni loading, the conversion and selectivity to branched-C12 started to decrease. The enhanced activity with the presence of nickel is related to the location of Ni centers near the acid centers, which favors a fast desorption rate of the intermediates involved in the isomerization reactions [9] due to the higher amount of surface Ni species. It can also be correlated with the acidity of the catalysts. In the TPD measurements, it can be seen that with an increase in the nickel percentage in the catalysts, the intermediate acidity centers grow, which are directly related to the activity [33–35]. Such a catalytic behavior is typical for bifunctional catalysts. The higher accessibility of Ni sites in the proximity of acid sites enables that carbenium ions and olefins dehydrogenate rapidly and desorbs them as alkanes before they undergo cracking reactions. This effect is due to a lower metallic site/acid site ratio than the optimal for isomerization. If the catalysts metal function (in this case nickel) is increased, the formation of isomers via a decrease in the diffusion path between two metallic sites, will be promoted. This is because the proximity of metallic sites reduces the probability of the intermediate species interacting with acid sites and in consequence the crack of these species. For all that, there is an initial increase in isomerization selectivity (isomerized/cracked ratio) with increasing Ni loading from 0.5 to 6% [9,10]. In contrast, in the range of 8–10% Ni, metallic sites are enough to form olefins covering all the acid sites, and the increase in metal function (nickel loading) avoids the isomerization reaction and increases the cracking reactions [9,10] due to its low dispersion. The generation of the acid sites and the elimination of ionic intermediates from the surface before β-scission reactions occur depends on the accessibility of the dissociated hydrogen. Metallic nickel ensures the formation of this dissociated hydrogen needed to generate active acid sites. This avoids polymerization and cracking reactions and increases the isomerization selectivity [9]. The selectivity (Fig. 7 ) also changes as a function of the nickel loading in the catalysts. The maximum selectivity toward iso-C12 and toward branched-C12 was obtained with the NiW/Al-2 catalyst. Higher loads of nickel revealed a change in the selectivity toward cracking products. This trend increases with the quantity of nickel, showing a similar tendency to that observed in the conversion results. Although the NiW/Al-2 catalyst offers the highest selectivity toward the desired products, the NiW/Al-6 catalyst presents the highest yield of the reaction because of the high conversion and selectivity to iso-C12 and to branched-C12 (Fig. 6). This change in the selectivity, observed when a high amount of nickel (> 6 wt%) is employed, can be related to a mismatch between metal/acid centers that contain the catalysts (hydrogenation ability and acidity) [36]. This can be associated with the growth of strong acidity centers for high metal loading, as can be observed in the TPD analysis, which can favor cracking reactions and at the same time reduce the yield of the reaction owing to a slower desorption rate of intermediates [33,34]. Additionally, this can be explained by the metal distribution. A high nickel content can promote cracking reactions due to the worse metal dispersion caused by sintering or the presence of larger nickel particles, as seen in the XRD analysis for the NiW/Al-10 catalyst [9,10,35,37]. The obtained data show a compromise between conversion and selectivity, which is a characteristic behavior of bifunctional catalysts [9,35,37], with NiW/Al-6 being the most active catalyst attributable to an optimum balance between metal and acid centers. XPS data showed a good dispersion, and FTIR of adsorbed pyridine showed that this is the first sample without the 1622 cm−1 signal, a clear indication of a good balance between metal and acid centers. Table 4 compares the results we have obtained with similar reactions published in the bibliography. Hydroisomerization reactions varied in the temperature range of 250–350 °C. In general, zeolites are clearly more active in hydroisomerization reactions but with lower selectivity with respect to the catalytic system presented in this work; see Refs. [2,21]. Several reactions with catalysts supported on SiO2-WO3 [9,10,38] were carried out at 250 °C, and the results show that these catalysts are less selective than our catalytic system. Our catalytic system reaches a very high selectivity to long chain hydrocarbons with respect to the state of the art (Table 4), with a moderate conversion level, similar to previous works with nickel on amorphous supports. The acid/metal balance of the catalytic system presented in this work seems very efficient for the hydroisomerization of long chain hydrocarbons as a balance between conversion and selectivity results is derired and obtained.The conversion of n-dodecane and selectivity to branched C12 trends to increase with Ni loading in the nickel containing catalysts prepared by the wetness impregnation method. The highest conversion (28%) was achieved with the NiW/Al-6 catalyst (6% wt. Ni). Above 6 wt% Ni loading, a decrease in conversion and increase in selectivity toward cracking products were observed. This trend is more pronounced when the nickel loading rises. By increasing the nickel content to 6 wt%, the intermediate acidity was enlarged, which is directly correlated with the activity performance, increasing the yield toward branched-C12. However, above 6 wt% Ni, the formation of larger metal particles and the strong acidity increased, both factors promoting cracking reactions instead of isomerization reactions. At high nickel loading (10% wt. Ni), XPS and XRD data showed a decrease in nickel dispersion which was deposited preferentially on tungsten moieties forming bulky crystal structures. The obtained data confirm the role of an optimum balance between metal and medium-strength acid centers in the catalysts, which is the bottleneck for maximizing the yield in this kind of reaction. D. García-Pérez: Investigation, Formal analysis, Writing – original draft. A. Lopez-Garcia: Investigation, Formal analysis. P. Reñones: Investigation, Formal analysis, Writing – original draft. M.C. Alvarez-Galvan: Supervision, Writing – review & editing, Funding acquisition. J.M. Campos-Martin: Conceptualization, Supervision, Writing – review & editing, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The support of MICIN/AEI (Spain) through project ENE2016-74889-C4-3-R is acknowledged. DGP acknowledges MICIN/AEI for her contract (BES-2017-079679) (Spain). This research is part of the CSIC program for the Spanish Recovery, Transformation, and Resilience Plan funded by the Recovery and Resilience Facility of the European Union, established by Regulation (EU) 2020/2094 (TRE2021-03-012). We acknowledge the support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.mcat.2022.112556. Image, application 1

Heterogeneous catalysts based on alumina-supported tungsten oxides (15 wt% W) with different loadings of nickel (0.5, 1, 2, 4, 6, 8 and 10 wt% Ni) were selected to study the influence of Ni loading on the hydroisomerization of n-dodecane. The catalysts were prepared by applying the wetness impregnation method on the supports to introduce W and Ni. The characterization techniques applied for determining physicochemical properties of the catalysts were N2 adsorption-desorption at 77 K (textural properties), X-ray diffraction (structure and crystalline phases), H2-TPR (redox properties), FTIR, NH3-TPD (acid sites analyses) and XPS (chemical surface analysis). The catalytic properties of such catalysts were found to be crucial in the n-dodecane conversion. The NH3-TPD profiles indicate that the medium acid sites are the main sites responsible for the reaction performance. The formation of bulky crystal structures of nickel species in the high nickel loading catalyst (10 wt% Ni) was confirmed by XRD and XPS results, resulting in the largest cracking activity. The conversion of n-dodecane and selectivity to i-C12+branched C12 tend to increase with Ni loading until the catalyst contains 6 wt% Ni (28% n-C12 conversion and 94% of branched C12 selectivity). The lower selectivity at high nickel loading is due to metal-based cracking reaction. An optimum balance between metal and acid centers is needed to achieve a compromise between conversion and selectivity, avoiding or minimizing cracking reactions.

In recent years, hydrogen has gained increasing attention as an alternative to fossil fuels, enabling net zero targets to be realized. As stated in the 2021 UK Hydrogen Strategy [1], the UK is aiming for a total of 10 GW of low-carbon hydrogen production capacity by 2030 to decarbonize vital industries and provide clean energy across the heat, power, and transport sectors. This requires considerable effort in scaling up and optimizing carbon capture and storage systems as well as new hydrogen production processes, such as sorption-enhanced reforming.Hydrogen can be produced from a variety of renewable and non-renewable sources, and can be divided into three categories depending on its production pathway. • Black/Grey/Brown hydrogen: from fossil fuel-based production (coal, natural gas, and lignite, respectively) with CO2 released to the atmosphere. • Blue hydrogen: from fossil fuel-based production with carbon capture, utilization, and storage (CCUS). • Green hydrogen: from renewable sources, commonly electrolysis-based production. Black/Grey/Brown hydrogen: from fossil fuel-based production (coal, natural gas, and lignite, respectively) with CO2 released to the atmosphere.Blue hydrogen: from fossil fuel-based production with carbon capture, utilization, and storage (CCUS).Green hydrogen: from renewable sources, commonly electrolysis-based production.Different hydrogen production pathways using renewable sources have been investigated, including water electrolysis [2], thermochemical (pyrolysis and gasification), or biological conversion (fermentation and photolysis) of biomass [3]. However, various techno-economic studies have demonstrated that compared to these processes, fossil fuel reforming with CCUS remains the most cost-competitive option with the highest hydrogen production efficiency [4–6]. Steam reforming, partial oxidation, and autothermal reforming are three main fossil fuel reforming technologies for hydrogen production, among which the steam reforming of methane (SMR) is by far the most deployed method. Although SMR is a mature technology, one of the most significant problems is its high CO2 emission. It is estimated that without CCUS, hydrogen from SMR has an emission factor of 222–325 gCO2eq per kWh of H2 (10 tCO2/tH2) [4,7].Apart from the traditional approach of employing a downstream amine scrubbing process, an alternative option to mitigate carbon emissions is adding a simultaneous carbon capture step to the conventional SMR process. A novel hydrogen production technology, known as sorption-enhanced steam methane reforming (SESMR), combines the conventional SMR process with a simultaneous in-situ absorption of CO2 using a solid sorbent (usually CaO). The main reactions involved in the SESMR process are as follows [8].Steam reforming of methane (1) CH 4 + H 2 O ↔ CO + 3 H 2 Δ H 298 ° = + 206 kJ / mol Water-gas shift (WGS) reaction (2) CO + H 2 O ↔ CO 2 + H 2 Δ H 298 ° = − 41 kJ / mol CO2 sorption and sorbent regeneration (3) CO 2 + CaO ↔ CaCO 3 Δ H 298 ° = − 178 kJ / mol Overall equation for SESMR (4) CH 4 + 2 H 2 O + CaO = 4 H 2 + CaCO 3 Δ H 298 ° = − 13 kJ / mol In comparison to the traditional SMR process, SESMR enables the removal of CO2 from the reaction zone, which shifts the equilibrium towards the product side, enhancing the production of hydrogen. The high-purity CO2 stream released from the sorbent regeneration step can also be easily separated from the sorbent, and transported or stored for further use. The sorption enhanced steam reforming process has been applied to other feedstocks as well, including phenol [9], glycerol [10], ethanol [11], and biomass [12]. In general, the CO2 sorbent is combined with active catalytic metal(s) to form a bi-functional material, using alumina, perovskite or mayenite as the structural support. Since both SMR and SESMR require the use of catalysts to proceed, and the types of catalysts used for both processes are in general identical, they will be reviewed holistically in this paper.The main reaction steps of SMR are listed in Table 1 , including the dissociative adsorption of the reactants, dehydrogenation and bond reformation steps. It is generally agreed that the activation of the first C–H bond of the CH4 decomposition step (step 1) is the rate-determining step of SMR [13–15]; but at lower temperatures (T < 500 °C), the CO formation step (step 7) becomes dominant [14]. The energy barrier for C–H bond activation over Ni surface is relatively low, and at the same time, the adsorption of C∗, H∗ and O∗ is not so strong that the species cannot react off the surface easily. Nickel is also widely employed commercially due to its low price and high availability. However, Ni-based catalysts are prone to sintering and coke formation [16,17], hydrogen reduction of nickel-based catalysts before the reforming process is also necessary for activating the material. It is, therefore, of interest to investigate the anti-sintering ability, coke resistance, as well as the reducibility and self-activation ability of SMR catalysts.Apart from nickel, noble metals (Rh, Ru, Pd, Pt, and Ir) are also promising candidates for SMR because of their excellent catalytic ability and resistance to carbon formation. Currently, there is no definitive conclusion as to how the noble metals are ordered regarding their catalytic activity for SMR, however, several experimental [18,19] and numerical [20,21] studies reported that the catalytic activity of noble metals follows the order of Rh ∼ Ru > Ir > Pt ∼ Pd. Despite their advantages, noble metal-based catalysts are limited by their high prices.One way to maintain the excellent performances of noble metals while maintaining a reasonable price is by combining two or more types of metals, using cheap transition metals (usually nickel or cobalt) as the base and noble metals as promoters. Bi/polymetallic catalysts have gained increasing attention in recent years, and the synergistic effect between commonly used metal elements has been investigated experimentally and numerically. Numerical studies focus on the reaction pathway, activation energies of certain reaction steps (in particular the C–H bond cleavage of the CH4 decomposition step), as well as the adsorption energies of atomic or molecular species on the catalyst's surface, which are indicators of the material's catalytic activity and stability. Some materials were also tested experimentally, usually in lab-scale reactors, and evaluated based on their methane conversion ability, hydrogen yield, etc. To the best of the author's knowledge, there has been no literature that systematically summarizes the bi/polymetallic catalysts that have been employed in (SE)SMR. The aim of this review is therefore to provide an overview of the bi/polymetallic catalysts that have been tested for (SE)SMR, to summarize their advantages and limitations, and to identify the gap in current (SE)SMR catalyst development for future studies.Amongst the eight noble metals, platinum group metals (including Rh, Ru, Ir, Pt, and Pd) have the highest SMR activity and are most commonly used as a promoter for Ni-based catalysts. Their catalytic performance arises from the partially filled d-subshells – electrons can be easily added to or removed from these orbitals, resulting in an optimal interaction between the metal surface and the gas-phase adsorbate. A lower degree of d-band filling leads to the too-strong adsorption of both reactants and reaction products on the metal surface, which easily blocks the active catalytic surface. On the other hand, with a higher degree of filling, the metal surface does not interact strongly enough with the reactants, which is the case for both Ag and Au. This results in a relatively low catalytic activity but a more stable and clean metal surface, which is why Ag and Au are usually added for their coke-resistant property. There is currently no literature available regarding the use of Os as an SMR catalyst, possibly due to its tendency to form a volatile and toxic oxide – OsO4 [18].Ru–Ni catalysts have been tested under lab-scale experimental SMR conditions by Jeong et al. [22] to study the effect of doping Ru over Ni/Al2O3 and Ni/MgAl2O4. They concluded that adding a small amount of Ru (0.5 wt.%) greatly suppressed coke formation on the catalyst surface and facilitated NiO reduction. The coke-resisting ability of Ru was studied on an atomic scale using Density Functional Theory (DFT) study [23]. It was demonstrated that when the noble metal was added, the activation energy of the CHO∗-producing step (step 5 shown in Fig. 1 ) was significantly lower than that of the C∗ and H∗-producing step (step 4), meaning CO production was favoured over carbon deposition.Results also showed that Ru-promoted Ni catalysts were able to self-activate at a temperature of 700 °C without any pre-reduction using hydrogen, which is beneficial from an economic and process operation point of view [24,25]. Ni-based catalysts doped with small (0.5 wt.%) or even trace amounts (0.05 wt.%) of Ru showed good self-activation properties without pre-reduction, and achieved a higher CH4 conversion rate compared to monometallic Ru or Ni catalyst. Ru decreased the reduction temperature of Ni by inducing hydrogen spill over on the Ni surface, a process in which H2 molecules disassociate on the noble metal surface to H species and diffuse into Ni via the catalyst support.Similar to Ru, the addition of Rh was also found to facilitate Ni reduction and produce a synergistic effect [26,27]. The bimetallic catalyst (with 0.2 wt.% Rh) showed higher activity compared to the linear combination of monometallic Rh or Ni catalyst. This was attributed to the enhanced textural properties of the bimetallic catalysts – the exposed metallic surface area and metal dispersion of the bimetallic catalysts, measured by CO chemisorption, were greatly promoted by the addition of Rh, and the increase was more significant with a higher Rh loading. The same synergistic effect was observed by Morales-Cano et al. as well [28], who studied the promoting effect of Ru, Rh, and Ir on Ni/α-Al2O3 catalyst. The catalysts were also aged for 240 h at 800 °C with an S/C ratio of 6 to induce the sintering of Ni particles. The activity of the aged Ni–Rh and Ni–Ir was found to be significantly higher than the aged monometallic catalysts, proving their ability to resist Ni sintering. This ability was attributed to the migration and diffusion of Ni into the Face Centred Cubic structure of Rh and Ir during the aging process, which enabled the formation of Ni–Rh and Ni–Ir alloys, and retained the high surface area of the materials.The optimal loading of noble metals in Ni-based catalysts was also investigated. Katheria et al. [29] tested a series of Ni/MgAl2O4 catalysts with Rh concentration varying from 0.1 to 1 wt.%. Results showed that 0.1 wt.% of Rh was sufficient to increase CH4 conversion by 20%, whereas a further increase in Rh concentration did not have a significant effect. A higher metal loading does not necessarily mean a better catalytic ability due to the less evenly distribution of active metal in the support. This observation was also verified by testing a series of Ni/MgAl2O4 catalysts with Pt loading varying from 0.01 to 1 wt.% [30,31]. Both studies reported that a Pt loading of 0.1 wt.% resulted in the highest catalytic activity. Further increase in Pt loading led to a decrease in both catalytic activity and stability. Results from the physical characterization of the materials showed that the highest surface area and maximum dispersion of active metal were achieved with 0.1 wt.% Pt loading, whereas higher Pt concentration, resulted in agglomeration on the material surface.Chaichi et al. [32] synthesized a novel supportless Ni–Pd-carbon nanotube material and compared its performance with Ni and Ni–Pd catalysts under SMR conditions. The addition of Pd facilitated the reduction of metallic oxides, whereas both Pd and carbon nanotube increased the specific surface area. The resultant CH4 conversion of the Ni–Pd-carbon nanotube material was 22% higher than the monometallic Ni/MgO catalyst. Reducibility enhancement by Pd was also reported by Batebi et al. [33] in a test of Ni–Pd/Al2O3 for combined steam and CO2 reforming of methane. By adding Pd, the reduction degree was increased from 69% to 83%, leading to higher CH4 conversion and H2 yield while reducing coke deposition. Bimetallic Ni–Pd materials have also been tested for the oxidative SMR process [34–38]. Results from these studies demonstrated that the addition of Pd promoted the reduction of Ni, Pd–Ni alloy was also found to form preferentially on the material surface, contributing to its high activity and coke resistance.Li and Miyata conducted a series of tests to study the doping effect of Ru [39,40], Rh [41,42], Pt [42–44], and Pd [42] on Ni/Mg(Al)O catalysts in a daily start-up and shut-down operation of SMR under steam purging. As was presented previously, the addition of all four types of promoters improved the reducibility of the catalyst by decreasing Ni reduction temperature and increasing the amount of hydrogen uptake on Ni. Ru, Rh and Pt were also capable of suppressing the deactivation of the catalyst due to Ni oxidation, which was attributed to the self-regeneration of Ni0 from Ni2+ assisted by hydrogen spill over on the noble metal surface and the reversible reduction-oxidation between Ni0 and Ni2+ in the Mg(Ni, Al)O periclase. Self-activation without any reduction treatment of Rh-, Pt-, and Pd–Ni bimetallic catalysts was also observed during the daily start-up and shut-down operation. Compared to the complete deactivation of the pure Ni catalyst after the first steam purging, the CH4 conversion of the bimetallic catalysts was kept at the value of thermodynamic equilibrium even after 4 cycles of steam purging. The self-activation and self-regeneration properties of these bimetallic materials proved Ru, Rh, Pt, and Pd to be useful additives to the conventional Ni-based catalysts.Although the reactivity of monometallic Ag and Au is relatively low, they have also been tested as promoters for Ni-based SMR catalysts due to their excellent stability. DFT-based studies [45,46] showed that the Ag-doped Ni surface is less prone to carbon deposition – the threefold Ag–Ni–Ni adsorption site is unstable for carbon atoms. Carbon atoms initially positioned at these sites will therefore move to the adjacent Ni–Ni–Ni site, which has lower adsorption energy. These negative interactions between the carbon atom and the Ag–Ni alloy surface indicate that Ag can be added as a coke-resistant promoter, which was also validated against experimental findings [47–49]. Both research teams studied the promoting effect of Ag (0.03–1 wt.%) on Ni/γ-Al2O3 and concluded that even the minimum Ag loading of 0.3 wt.% could reduce carbon deposition significantly. However, this also compromised the catalytic activity of the material, which decreased by 25% compared to monometallic Ni catalysts. This is due to the fact that Ag atoms are energetically favoured to replace Ni atoms on the step edges, which are the most active sites for methane decomposition, compared to the terrace sites [50]. Similar properties were found in Au-doped Ni catalysts. Both computational [51–53] and experimental [54,55] studies suggested that the overall catalytic activity of Au–Ni was affected by the higher energy barrier for C–H bond cleavage in the rate-determining CH4 dissociation step; whereas the stability of the material was enhanced due to the suppression of carbon formation.Ag and Au have also been employed as promoters to Ni electrodes for solid oxide fuel cells under internal SMR conditions. Ag [56] or Au [57] with a loading of 1–5 wt.% doped on Ni/yttria-stabilized zirconia (YSZ), as well as Au with a loading of 1–4 at.% doped on Ni/CeO2-Gd2O3 [58] anode were tested at temperatures ranging from 650 to 800 °C. Both exhibited satisfying performance with enhanced tolerance to carbon formation. However, the performance of Ag-doped materials is largely influenced by the reaction temperature. At temperatures higher than 750 °C, the Ag–Ni/YSZ cell degraded rapidly due to the low melting point of Ag [56]. At temperatures lower than 600 °C, the catalytic activity of the Ag–Ni/Al2O3 catalyst decreased significantly, whereas the Au-doped catalyst still exhibited higher activity than monometallic Ni/Al2O3 [59,60]. Sapountzi et al. [61] reported that an Au amount of 2.3 wt.% promoted the reducibility of Ni catalyst, and the Ni–Au alloy formed on the catalyst surface was able to inhibit the formation of sulphuric compounds, including nickel sulphide.Liu et al. [62] reported that Ni–Ir alloy supported on MgAl2O4 was a durable catalyst for steam and CO2 bi-reforming of methane under pressurized conditions. The bimetallic catalyst was composed of small metallic clusters (with a mean size of ∼2 nm) and the cluster size was retained for a duration of 434 h, in contrast to the significant increase in cluster size of the monometallic Ni/MgAl2O4 catalyst, showing the anti-sintering ability of Ir. The coke resistance of the bimetallic material was attributed to the combined effect of small ensemble sizes, increased surface oxophilicity, and higher activation barrier for CH4 dissociation. The number of active sites was evaluated by H2 chemisorption, it was found that an Ir loading of 0.1 wt.% was able to quadrupole the quantity of active sites of Ni/MgAl2O4, and the promoting effect increased with a higher Ir loading. The bimetallic Ir10Ni90/MgAl2O4 catalyst achieved CH4 and CO2 conversion of 95% and 98%, respectively, at 1 bar; and was able to maintain a relatively high CH4 and CO2 conversion of ∼60% when pressurized to 20 bar.Despite multiple advantages, the use of noble metals is still constrained for economic reasons. Therefore, many researchers have turned to the application of non-noble metals as potential promoters of Ni-based catalysts.Ni–Fe-based catalysts/oxygen carriers have been tested for chemical looping steam methane reforming (CL-SMR). Hu et al. [63] used Ni–Fe modified calcite as an oxygen carrier and concluded that a Fe/Ni ratio of 0.67 was optimal for the reaction, while higher Fe concentration led to sintering. The novel material exhibited good catalytic performance with the highest CH4 conversion of 98.9%, and high stability during the long-term reaction process. Garai et al. [64] tested Ni-ferrite supported on ZrO2 and CeO2, which showed high H2 and CO selectivity, as well as high CH4 conversion (93%, 98%, and 99%, respectively). No carbon deposition was observed on the used material, due to the ability of iron oxides (FeOx) to remove carbon via a surface redox cycle to produce Fe and CO2 [65]. Djaidja et al. [66] reported that although the addition of Fe slightly decreased the CH4 conversion of the (Ni–Mg)2Al catalyst from 93% to 91%, both H2 and CO yield were improved and carbon formation was significantly suppressed.In addition, the Fe-promoted Ni catalyst was also proved to be sulphur-resistant. Tsodikov's team [67,68] tested Ni–Fe/γ-Al2O3 catalysts prepared by epitaxial coating and a novel core-shell type catalyst containing Ni and Fe (Fig. 2 ) under SMR conditions in the presence of up to 30 ppm H2S. The materials showed good catalytic activity, unlike conventional Ni-based catalysts, which lose activity rapidly when the gas-vapour mixture contains H2S. This property was attributed to the core-shell structure, in which the core containing Ni–Fe nanoparticles provided the catalytic ability while the γ-Fe2O3 shell provided vacancies for H2S to decompose to elemental sulphur following the reactions below: (5) FeO + H 2 S → FeS + H 2 O (6) FeS = Fe vacancy + S (7) Fe vacancy + H 2 O → FeO + H 2 Cobalt is also considered a promising SMR catalyst additive because of its good activity for the WGS reaction, which assists in shifting the equilibrium towards higher H2 production. However, one problem related to the usage of Co is its tendency to oxidize when the temperature and steam partial pressure are in the range used for SMR [69]. Alloying it with Ni is a potential solution to this problem while preserving the advantages of both elements. A series of Ni–Co/ZrO2 bimetallic catalysts with different Co loadings were tested and compared to monometallic Ni and Co catalysts by Harshini et al. [70]. Their results suggested that a Ni/Co ratio of 1:1 was optimal for limiting both oxidation of Co and carbon formation caused by Ni. The material also exhibited long-term stability with a constant CH4 conversion of 81.8% within 50 h with no surface carbon formation. You et al. [71] tested a series of Ni–Co/γ-Al2O3 catalysts and found that at a temperature of 800 °C Co-modified catalysts exhibited the same reforming activity as unmodified ones with enhanced coke resistance. The performance of the bimetallic catalyst was not as good as Ni at low temperatures, possibly due to the formation of Ni–Co alloy, which increased the crystallite and particle size of the material, while decreasing metal dispersion and surface area, and blocking low-coordinate active Ni sites where the rate-determining CH4 dissociation step takes place.Ni–Cu bimetallic catalysts have been used for the conventional SMR process [60] as well as low-temperature SMR [73–75]. Results showed that by adding Cu as the promoter, a larger Ni crystallite size, surface area, and a better metal dispersion was obtained [74], and the overall carbon resistance of the material was enhanced [66,72]. TGA results before and after a long-term SMR test (20 h) showed that carbon formation on Ni–Cu/Al2O3 was great suppressed (8.9%) compared to commercial Ni/Al2O3 (28.3%). The addition of Cu has also been proven to enhance the activity of the WGS reaction [76], which explains the increase in CH4 conversion when using Cu–Ni as the catalyst. However, it should be noted that an upper limit exists in terms of the promoting effect of Cu. A Cu/Ni ratio equal to or higher than 5 in the material will result in reduced catalytic activity, as reported by Huang et al. [73].The promoting effect of Zirconium was investigated by Lertwittayanon et al. [77] using Ni/α-Al2O3 catalysts containing CaZrO3 nanoparticles. CaZrO3 loading between 10 and 15 wt.% showed the best catalytic performance with a CH4 conversion of 67%. Results also showed that unlike conventional SMR catalysts requiring an S/C ratio of approximately 3, an S/C ratio of 1/3 or 1 was most appropriate for the CaZrO3-modified catalyst. This is because a high S/C ratio causes an excessive amount of steam to adsorb on the CaZrO3 surface, which competes with the adsorption of CH4.Boudjeloud et al. [78] tested a series of La-promoted Ni/α-Al2O3 catalysts. The highest CH4 conversion (97%) and H2 yield (94%) were obtained with a Ni/La ratio of 7:3. The improved activity compared to monometallic Ni was credited to the decrease in Ni particle size and enhanced metal dispersion, which prevented the agglomeration and sintering of the bimetallic material. The addition of La also facilitated the reduction of Ni, however, its effect on coke resistance was not significant [27].The effect of doping Molybdenum was studied by Maluf and Assaf [79] using Mo–Ni/Al2O3 catalysts with different Mo concentrations. The addition of Mo decreased the surface area of the catalyst, possibly due to the blockage of active Ni sites by MoOx. However, the specific activity of each active site was increased, which was attributed to the transfer of electrons from MoOx to Ni particles resulting in an increase in electron density in Ni. Molybdenum carbide has also been employed in the methane reforming process and is known to have good catalytic activity and stability at high pressures [80]. However, its stability quickly drops at atmospheric pressure due to the surface oxidation of Mo2C to MoOx by CO2. This problem can be mitigated by combining Mo2C with nickel. Despite being a major reason for the deactivation of traditional Ni catalysts, carbon deposited on the bimetallic Ni–Mo2C surface promotes the carburization of NiMoOx back to its carbide form [81,82]. Ni–Mo2C catalysts have been tested for dry methane reforming [81–83], steam reforming of methanol [84], as well as steam-CO2 bi-reforming of methane [85], and have shown more promising results than unpromoted Mo2C catalyst or Ni–Mo catalyst in their reduced form.The promoting effect of the rare earth element, rhenium, was reported by Xu et al. [86]. They concluded that by coating a Ni–Re bimetallic layer on the surface of a high cell density Ni monolith catalyst, the reducibility and catalytic performance of the material were enhanced. The low hydrogen adsorption energy of Re atoms also facilitates the adsorption of hydrogen atoms on Re and adjacent Ni atoms, which suppressed the oxidation of Ni and led to enhanced catalyst stability.Silicon is one of the most studied metalloids as it is often employed as the catalyst support for the SMR process, usually in its oxide or carbide form, because of its thermal stability and potentially high surface area. Silica is generally considered an inert material, as it has weak metal-support interaction with the active metal (Ni in most cases) due to its low reducibility [87]. The lack of metal-support interaction in Ni/SiO2 is also a source of filamentous whisker carbon formation [88]. To improve the interaction between Ni and the silica support matrix, Majewski et al. [89] synthesized a core-shell typed Ni/SiO2 catalyst using the Stöber-deposition-precipitation method, and tested it under different SMR conditions. Results showed that the core-shell structure increased the coke resistance of the catalyst, as deposited carbon was only detected at a low s/c ratio (1:1) and temperature (550 °C). Other characteristics of the support material, including crystallite size and metal dispersion, are affected by the acidity/basicity of the support and are also known to influence the rate of carbon growth on the catalyst surface. The acidity of the silica support facilitates the decomposition of methane, but at the same time promotes cracking and polymerization leading to catalyst deactivation because of carbon formation [88]. To achieve a better acid-basic balance, basic metal oxides, including ceria [90,91] and magnesia [92,93] are often added to Ni/SiO2 to tune the surface acidity of the support, and it was found that a homogeneous distribution of the basic sites on the acidic silica framework improved the long-term stability of the catalyst.Elements with a similar electronic structure of carbon include tetra- and penta-valent p such as Sn, Sb, As, Ge, Pb, Ag, etc. The addition of these metals was predicted to have a coke-resisting effect, because, similar to the formation of nickel carbide (interaction between 2p electrons of carbon and 3d electrons of Ni), the reaction between these metals with Ni could potentially reduce the chance of Ni–C interaction. A few of the above candidates were tested by D.L.Trimm [94], and their coke-resistant ability followed the order of As > Ag > Sb > Sn > Pb.The addition of Sn was also investigated by Nikolla et al. [95–97] experimentally and numerically. The bimetallic Ni–Sn catalysts showed lower CH4 conversion during the first 30 min of the reaction, however, its long-term stability was greatly enhanced. The carbon resistance of the Ni–Sn/YSZ catalyst was explained by DFT calculated reaction energy barriers, which showed that the Ni–Sn alloy surface preferentially oxidizes C∗ rather than forming C–C bonds. The presence of Sn also lowers the binding of C∗ to low-coordinated sites, which is the position for carbon nucleation. The decrease in the catalytic activity of Ni–Sn alloy is possibly due to the blockage of low-coordinate Ni sites by Sn, as these sites are the most active for the rate-determining C–H bond activation step.Similar to Sn, boron-promoted Ni catalyst has also demonstrated good stability due to the reduction in carbon nucleation centres. A boron loading of 1 wt.% was sufficient to enhance the overall stability of the material without compromising its catalytic activity [98]. Apart from this, Ligthart et al. [27] also reported the structural-promoting ability of boron for obtaining small Ni particles. However, one limitation of the bimetallic material is that the addition of boron strongly impeded the reduction of Ni.Apart from the experimental work mentioned above, the SMR activity of metalloid-promoted nickel catalysts was also studied using numerical methods. In the work by Xu et al. [21], the catalytic activity of a series of bimetallic alloys was predicted based on a microkinetic model, and DFT-calculated atomic adsorption energies on the bimetallic M1M2 (211) surface. By setting the conditions as 793 °C, 12.2 bar, and with gas composition as 50% to equilibrium, alloys including Ni3M (M = Sn, Sb, Ge, and As) and Co3Ge were predicted to have the highest activity (Fig. 3 ). Although these elements showed promising results, experimental validation of the in-silico study has not been found. Further study can be carried out on these metalloids (Sb, As, Ge)-based catalysts in search of an optimal balance between activity and stability.Apart from nickel, some other transition metals, such as Co, Cu, and Fe, have been used as catalysts for reforming processes (dry and steam reforming of hydrocarbons, glycerol, or bio-derived material). Shen et al. [99] tested a series of monometallic Co/CeO2 and bimetallic Co-M/CeO2 catalysts (M = Ni, Al, and Cu) under conventional SMR conditions to study the effect of Co loading and different promoters. As mentioned previously, higher active metal loading does not necessarily mean better performance because of the uneven distribution of active compounds in the support. A Co loading of 12% was found to be optimal in terms of CH4 conversion and H2 yield. The addition of both Ni and Al increased CH4 conversion, whereas Cu slightly reduced the overall catalytic activity, possibly due to the sintering of Cu. The combination of Ni–Co was chosen over Al–Co because it exhibited higher CH4 conversion (76.1%), H2 selectivity (58.5%), and H2 yield (44.5%).The cobalt-based catalyst prepared from hydrotalcite precursors using the anion-exchange method was tested by Lucredio and Assaf [100] with low H2O/CH4 feed ratios of 2 and 0.5 to test the stability of the material under extreme conditions. For an H2O/CH4 ratio of 2, CH4 conversion was maintained at 80% during 6 h of reaction; the carbon amount on the used catalyst was found to be only 2.7 wt.% after 30 h of reaction. The catalyst began to show a deactivation tendency due to the deposition of excess carbon when H2O/CH4 ratio is further decreased to 0.5, and CH4 conversion decreased from ∼60% to 40% during 6 h of reaction.Although Co is less prone to coke formation compared to Ni, the interaction between Co and the metal oxide support is strong, leading to the formation of cobalt oxides with limited reducibility [101]. As presented in section 2.1, by adding a small amount of noble metal the reducibility of the transition metal-based catalysts can be largely improved because of the hydrogen spill over effect. Profeti et al. [102] explored the effect of noble-metal promoters (0.3 wt.% Pt, Pd, Ru, and Ir) on Co/Al2O3 catalysts. Results showed that the addition of the noble metals significantly decreased the reduction temperature of cobalt species, with their promoting effect following the order of Pd > Pt > Ru > Ir. In terms of their catalytic activity, the Co-based bimetallic catalysts did not show satisfying results. Average CH4 conversion of 50–60% was obtained for Pd-, Pt- and Ir–Co, 30% for Ru–Co, and only 7% for Co/Al2O3, which was possibly due to the partial oxidation of cobalt active sites in the presence of water molecules.Akbari-Emadabadi et al. tested a Ca–Co bi-functional catalyst/sorbent (with a mass ratio of Ca/Co = 9) in the CL-SMR process, and investigated the promoting effect of yttrium [103] and zirconium [104], with a mass ratio of Ca/Y = Ca/Zr = 4.5. Both promoted samples remained stable at 700 °C for up to 16 redox cycles, whereas the unpromoted one was deactivated after 10 cycles. The catalytic performance of the materials is summarized in the table below (Table 2 ). Both Y and Zr showed promoting effect regarding catalytic activity for SMR and H2 selectivity, and the usage of Zr was more advantageous in comparison to Y. Based on the results from catalyst characterization, the addition of Co reduced the overall surface area of the material by more than 30%, whereas the addition of Zr compensated this negative effect to some extent. Results also showed that Zr prevented the formation of Ca2Co2O5 spinel in the structure of the bimetallic material, which lowered the risk of losing active sites of Co. The study proved Y and Zr to be promising textural promoters of the bi-functional catalyst/sorbent materials employed in CL-SMR.Apart from Co-based bimetallic catalysts, catalysts combining two types of noble metal have also been studied. The research by Roy et al. [105] focused on a novel Pt–Rh (1.2 wt.%) catalyst supported on metal foam. The sample was tested in a multichannel heat exchanger platform reactor to evaluate its potential in solid oxide fuel cell application. The combination of Pt and Rh was proven to enhance the production of hydrogen by SMR, with CH4 conversion, H2 yield, and H2/CO ratio of 97.2%, 3.16 mol per mol of CH4 input and 6.03, respectively, all of which were higher than commercial monometallic Ni and Ru catalysts. The novel catalyst also showed excellent stability, negligible coke deposition was found after 200 h of SMR reaction at 800 °C. Further research from an economic point of view should be carried out to evaluate the potential of these materials in large-scale applications.Compared to the relatively simple mono and bimetallic system, the application of catalysts containing three or more types of active metals in SMR has not been investigated in detail. Existing literature mainly examined Ni-based material with the addition of two or three commonly used elements, such as Co, Cu, Ru, Pt, etc.The effect of the simultaneous presence of copper and zinc in Ni/Al2O3 catalyst was investigated by Nazari and Alavi [106]. They reported that Cu and Zn affect the Ni-based catalyst in different ways – Cu enables a better resistance to coke formation while Zn improves the catalyst's activity, stability, and H2 selectivity. The optimal combination of the three metals for SMR was found to be 15%Ni–1%Cu–5%Zn, which achieved a CH4 conversion of 94% and an H2 yield of 3.12.Jeon et al. [107] investigated the performance of a selection of bi- and trimetallic Ni-based catalysts under steam-deficient conditions. A series of bimetallic catalysts containing 5 wt.% of alkaline earth metal (Mg, Ca, Sr, Ba) or 0.5 wt.% noble metal (Ru, Rh, Pt, Pd), and trimetallic catalysts containing both alkaline earth metal and Ru were synthesized. Results from the tests demonstrated that adding Mg or Ca enhanced the coke resistance of Ni-based catalysts, whereas the effect of Sr and Ba was not significant. Among the noble metals, Ru was the best candidate for suppressing coke deposition. Based on these conclusions, a catalyst with optimized composition – 0.5%Ru–5%Mg–10%Ni/γ-Al2O3 – was selected and tested. A CH4 conversion of 96% was maintained for 250 h, proving its excellent long-term stability.Bi-functional polymetallic materials combining the catalytic ability of transition metals and CO2 sorbents are commonly employed for the SESMR process [108]. Chen et al. [109,110] found that a simple physical mixture of Ni/Al2O3 (20 wt.%) and CaO was able to improve H2 purity to above 95%, compared to 72% without in-situ CO2 removal. Based on an elemental mapping analysis of the samples, each Ni elemental point was surrounded by several Ca points, which allowed the efficient capture of CO2 produced during the reaction.Dewoolkar and Vaidya [111] synthesized hybrid Ni–CaO/Al2O3 and Ni-hydrotalcite materials by coprecipitation and incipient wet impregnation, respectively. The materials were tested at T = 500 °C with an S/C ratio of 9 for 20 cycles for stability evaluation. Results showed that CH4 conversion and H2 yield of the hybrid materials were higher than those obtained using a physical mixture of catalyst and sorbent, due to the more efficient mass and heat transfer. The hybrid materials also exhibited better stability, Ni–CaO/Al2O3 and Ni-hydrotalcite maintained high H2 purity of 90% for up to 11 and 16 cycles, respectively, compared to the sintering and deactivation of the mixed material after only 2 cycles.Di Giuliano et al. [112,113] reported the use of mayenite as for bi-functional Ni–CaO material. Results from multicycle sorption/regeneration TGA demonstrated the stable sorption capacity of the material after 20 cycles. This enhanced stability compared to commercial CaO was attributed to the presence of mayenite as an inert binding, preventing CaO from sintering. The same phenomenon was also observed by Dang et al. [9], confirming the role of mayenite as a structural stabilizer. Di Giuliano et al. also concluded that the nickel precursor used for material synthesis may affect the texture and reducibility properties, and nickel nitrate hexahydrate was found to the most suitable precursor.Kim et al. used ruthenium as the reforming catalyst, and tested the performance of Ru–CaO/Ca3Al2O6 under SESMR conditions [114]. As Ru is a highly active catalyst for SESMR, the mass fraction of CaO was able to be increased significantly compared to conventional Ni–CaO-based materials. Ca3Al2O6 acted as the structural stabilizer against sintering, and maintained the surface area of Ru–CaO/Ca3Al2O6 at 18 m2/g after 10 cycles of SESMR, compared to 5 and 6 m2/g for Ru/lime and Ru/CaO.Hafizi et al. [115] modified conventional calcium-based CO2 sorbent with CeO2, and tested it for CL-SESMR together with Co3O4/SiO2. The addition of CeO2 significantly improved the morphology of CaO by increasing its surface area and uniformly distributed pores in the sorbent structure. Combined with the Co-based catalyst, the material was able to produce high purity H2 (93–96%) for 8 redox cycles, and maintain the same CO2 removal efficiency for 3 carbonation/calcination cycles.Ghungrud et al. [116] reported a novel trimetallic bi-functional material for SESMR, consisting of Ni and Co (with concentrations varying from 0 to 40%) supported on hydrotalcite and promoted by 2.5 wt.% cerium. The hybrid material was evaluated in terms of its H2 production ability, sorption capacity, and cyclic stability. Results revealed that CH4 conversion increases with the Co content in the material, which is possibly due to the enhancement of WGS reaction by Co. Under optimal reaction conditions, CH4 conversion of 95.7% and 90.7% were obtained by two Ce-promoted Ni–Co samples (with Ni/Co ratios of 1/3 and 1, respectively). The maximum sorption capacity was found to be 1.74 and 1.51 mol CO2/kg, respectively. The material was also tested at optimal conditions for 25 cycles, samples showed good stability by maintaining an H2 concentration higher than 90%, and remained stable for 21 and 16 cycles. This property was attributed to the effective metal-support interaction and higher active metal dispersion within the promoted material. The author concluded that the hydrotalcite-supported Ce–Ni–Co (2.5, 10, 30 wt.%) trimetallic catalyst/sorbent shows good performance and could be a promising candidate for large-scale SESMR application.Similarly, Dewoolkar and Vaidya investigated the promoting effect of Ce and Zr on a Ni/hydrotalcite bi-functional material [117]. Results showed that both Ce and Zr were able to increase the surface area and surface basicity, which inhibited coke formation. Both Ce and Zr promoted materials remained stable for 13 and 17 cycles, respectively, whereas the unpromoted material became unstable after 9 cycles. The addition of Ce was found to be particularly beneficial, as the promoted bi-functional material reached a high CH4 conversion of 96.4% and an adsorption capacity of 1.41 mol CO2/kg sorbent.Apart from the type of promoter added to Ni-based catalysts, the structure of the promoted catalyst also influences its overall performance. Cho et al. [118] synthesized a bimetallic catalyst with a novel egg-shell structure by selectively placing Ni and Ru on the shell of alumina. Three types of catalysts were compared, including (a) egg-shell-type bimetallic catalyst (5 wt.% Ni + 0.7 wt.% Ru/Al2O3), (b) monometallic 1 wt.% Ru/Al2O3 catalyst with the same egg-shell structure, and (c) monometallic 1 wt.% Ru/Al2O3 synthesized by the conventional wet impregnation method. The comparison between (b) and (c) showed that the novel structure was able to improve methane conversion and maintain it at a higher level when gas hourly space velocity was largely increased. This was because the novel structure enables the active metal, ruthenium, to be mainly deposited on the outer region of the alumina pellets and can therefore be utilized more efficiently. With the addition of nickel, (a) achieved an even higher methane conversion than (b) at higher gas hourly space velocity. This proved that the novel structure enables the efficient utilization of active noble metals, reducing the metal loading necessary and, therefore, the overall cost.Obradovic et al. [119,120] proposed a novel plate-type catalyst for SESMR, as demonstrated in Fig. 4 . This material was synthesized by depositing Pt and Al2O3 on a static mixer element made of Ni alloy. During the SESMR process, it acts simultaneously as the catalyst, the distributor for solid sorbent, as well as the gas-phase radial mixer. The CH4 conversion obtained by this novel catalyst was 15 times higher than monometallic Ni under the same conditions. However, its catalytic activity rapidly decreased after 10 h of reaction, temperatures higher than 590 °C also led to activity loss due to carbon accumulation on the material surface. Further investigation on modifying relevant parameters (such as the promoter type, metal loading, and reaction conditions) to increase its stability may be of interest.Based on the above review, it can be concluded that an increase in catalytic activity is usually achieved either by modifying the textural properties of the material (e.g. increasing surface area and metal dispersion) using a second active metal; or by modifying the energetics of the surface reactions, in particular the rate-determining step (e.g. decreasing activation energy of the first C–H bond in CH4 dissociation). Although a variety of metal combinations have been studied for their performance in the SMR reaction, there exist many other combinations of elements that have been predicted to be active or have not been evaluated at all regarding their reforming activity.SMR is currently the most dominant hydrogen production technology, and extensive research on the catalytic aspect of this process has been carried out. Carbon emissions from SMR can be reduced by adding a CO2-sorption step. This review provides insights on recent developments in the use of bi/polymetallic catalysts for (SE)SMR. The performance of the bi/polymetallic catalysts presented in this review is briefly summarized in Table 3 and is evaluated based on three main factors: stability (resistance to carbon, sulphur, sintering, and oxidation), catalytic activity and selectivity, as well as their physical/chemical properties (reducibility and self-activation ability).The most widely used SMR catalyst to date is ceramic-supported nickel because of its relatively good performance and inexpensive price, but problems such as sintering and coke formation still exist. In search of catalysts with better performance, various elements have been added as promoters to conventional Ni-based catalysts. Noble metal-promoted catalysts generally have superior reactivity, coke resistance, and enhanced reducibility, but their applications are often limited by their prices. Researchers have therefore turned to non-noble metals and metalloids. The addition of iron was found to be coke and sulphur-resistant due to the surface redox reaction between Fe0 and FeOx. Both cobalt and copper can enhance the activity of the WGS reaction, thus shifting the reaction towards more hydrogen production. Elements including zirconium, yttrium, and lanthanum were found to be good textural promoters due to their ability to increase the surface area and metal dispersion of the material. Silicon is often used as the catalyst support in its oxide or carbide form, and the addition of ceria or magnesia was able to tune the surface acidity of silica for better long-term stability. The addition of other metalloids (tin, boron) led to enhanced coke-resisting ability, but often at the expense of losing catalytic activity. A DFT-based study has also predicted germanium, arsenic, and antimony to be effective promoters of nickel-based catalysts, however, further experimental verification is still necessary.Apart from the type of element selected as the promoter, the loading of each component is also a critical parameter that influences the overall performance of alloy material. A higher active metal loading does not necessarily indicate a higher activity due to the restricted distribution of active metal in the material. The influence of material structure on catalytic activity was also investigated. Although novel core-shell type, plate type, and metal foam-support structures were found to be beneficial to the overall catalytic performance, this largely complicates the synthesis process and limits the wide application of the materials. It is, therefore, necessary to find a balance between improvements to the material properties and a viable and efficient material preparation process.Siqi Wang: Writing - Original Draft. Seyed A. Nabavi: Writing - Review & Editing. Peter T. Clough: Writing - Review & Editing, Supervision.PTC and SAN thank BEIS for funding under the H2 BECCS competition – Bio-HyPER project (H2BECCS107).All data underlying the results are available as part of the article and no additional source data are required.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Blue hydrogen production by steam methane reforming (SMR) with carbon capture is by far the most commercialised production method, and with the addition of a simultaneous in-situ CO2 adsorption process, sorption-enhanced steam methane reforming (SESMR) can further decrease the cost of H2 production. Ni-based catalysts have been extensively used for SMR because of their excellent activity and relatively low price, but carbon deposition, sulphation, and sintering can lead to catalyst deactivation. One effective solution is to introduce additional metal element(s) to improve the overall performance. This review summarizes recent developments on bi/polymetallic catalysts for SMR, including promoted nickel-based catalysts and other transition metal-based bi/polymetallic materials. The review mainly focuses on experimental studies, but also includes results from simulations to evaluate the synergistic effects of selected metals from an atomic point of view. An outlook is provided for the future development of bi/polymetallic SMR catalysts.

Sustainable renewable energy devices, such as water splitting, fuel cells, and metal-air batteries, have attracted immense attention. 1–4 However, their large-scale application, such as in Zn-air batteries, has been severely hindered by the intrinsic sluggish kinetics of the oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) in rechargeable air electrodes. Cost-effective bifunctional catalysts with performance comparable (or even superior) to that of commercial Pt/C or Ru(Ir)O2 have long been sought after. 5–7 The precise atomic design of transition-metal sites and their local electron engineering through substrates have emerged as an intriguing yet challenging strategy. 8–13 Given that studies on the rational construction of single-atom transition-metal sites with high OER activity are scarce, great efforts have been devoted to regulating the reaction pathway to facilitate the rate-determining step (RDS), such as O∗ and OOH∗ formation, through optimizing the d band electron structure of active transition-metal sites. 14–19 Among them, the multi-site synergistic effects have shown the promising potential to enhance OER efficiency. 17 , 18 Specifically, Duan's group demonstrated a relatively high OER activity of Ni-N-C single-atom catalysts (SACs), in which the OH∗ adsorbed on the C site facilitated the Ni site to form OOH∗ according to the so-called dual-site mechanism. 20 Hu's group observed the crucial importance of the formation of Co–Fe dual sites by the adsorption of a trace amount of Fe3+ in KOH electrolyte on the single-atom Co site 21 . These intriguing results have, however, seldom been explored in view of the geometric effects of the configuration matching between dual sites and the reaction intermediates.In addition, the local electron modulation of active sites by substrates is also a promising way to enhance catalytic activity. Among them, C-based substrates have been extensively studied because of their unique advantages for electron-transfer facilitation, tunable molecular structures, and strong tolerance to acid and alkaline electrolysis. 22–24 Heteroatom doping, such as with N and P, endowing C substrates with more chemically active sites for metal coordination and higher electrochemical activity for local electron structure modulation is a challenging metric for achieving simultaneous enhancements of the catalytic stability and activity. 25 , 26 Currently, the most acceptable mechanism for the high activity of the dual-metallic catalysts is the formation of one specific higher-valence site through electron transfer from another promoter site, which has been proved to optimize the adsorption energy of intermediates. 27 , 28 In this way, the local electron modulation can also be achieved by engineering the first coordination sphere of active metal sites, such as metal-N and metal-O coordination provided by substrates. 18 , 29 In addition, further local electron structure modulation by the rational design of substrates providing a second coordination sphere for active metal sites remains intriguing yet challenging. Bearing these in mind, engineering metallic atomic dual sites with modulated local electron structure on N,P-co-doped C could become a potential strategy for synergistically boosting OER performance through simultaneously achieving electron structure optimization and electron- and/or mass-transfer facilitation.Herein, we report a theory-guided design and fabrication of an Fe–Ni dual-site catalyst supported on microporous C for the OER process. We first employed density functional theory (DFT) calculations to reveal the simultaneous activation of OH∗ and O∗ on Fe–Ni dual sites, and this simple Fe–Ni adjacent association regulated the RDS of the formation of intermediate OOH∗ and significantly reduced its formation energy. Moreover, through electron-rich sp3 hybridization P doping, the M-N-P-C moiety modulated both the electronic and geometric structures of the local environment to improve the catalytic properties of active metal sites. The grand canonical Monte Carlo (GCMC) simulation further demonstrated the thermodynamic stability of alternative N-coordinated Fe–Ni configuration in the ZIF-8 cage. Thereafter, the Fe–Ni dual sites on N,P-co-doped C were fabricated in situ by one-step pyrolysis of Fe/Ni-adsorbed ZIF-8 with NaH2PO2 co-feeding. Consequently, the asymmetric Fe–Ni dual sites and the local P doping in the obtained Fe-Ni-N-P-C comprehensively enhanced OER catalytic performance (superior to that of all reported mono- and diatomic transition-metal N-based catalysts) and could take the place of commercial RuO2. Finally, Fe-Ni-N-P-C also exhibited an attractive ORR activity; a high-performance rechargeable Zn-air battery with Fe-Ni-N-P-C as the bifunctional catalyst for the air cathode was achieved. This transition-metal dual-site construction approach with configuration matching and local electron modulation is a promising strategy for complex electrocatalytic reactions.We explored the atomic multi-site structure of the most favorable catalyst by comparing the formation energy (ΔG) profiles of four typical steps during the OER process (Equations S11–S14) on the geometric optimized catalysts of the representative Fe/Ni-based catalysts 30 , 31 (Figures S1 and S2). Among them, Fe-Ni-N-P-C, with the heteroatomic Fe–Ni dual sites coordinated by N and P doping in nearby C, exhibited the lowest energy barrier of 1.90 eV for the commonly accepted RDS of the OOH∗ formation (Figure 1 D). It is worth mentioning that the formation energies of OH∗ and O∗ on the SACs of Fe-N-P-C or Ni-N-P-C (Figure S2) were quite different in that the Fe site favored the stable adsorption of OH∗ (0.23 eV), whereas the Ni site facilitates\d the formation of O∗ (0.31 eV). This provides the possibility to regulate the OOH∗ formation pathway through a simple adjacent association of neighboring Fe-OH∗ and Ni-O∗, consequently boosting the OER efficiency on the asymmetric Fe–Ni dual sites. This proposed mechanism was vividly verified by the OOH∗ adsorption on Fe-Ni-N-P-C, where both O atoms were bonded with Fe–Ni dual sites through bidentate bonding (Figure 1A), whereas only one O atom was bonded with the Fe or Ni SACs through monodentate bonding (Figure 1B), resulting in a high energy barrier of the OOH∗ formation. More precisely, the much stronger binding energy of OOH∗ on the Fe–Ni dual sites (−1,632 kJ mol−1) than on the symmetric Fe–Fe (−1,517 kJ mol−1) and Ni–Ni (−1,267 kJ mol−1) diatomic sites (Figure S3) further demonstrates the importance of the heteroatomic Fe—Ni dual sites on the precious configuration matching. Moreover, Bader analysis of the electron density 32–34 further revealed the contribution of the local electronic engineering through the electron-rich sp3 hybridization P doping: the electron donation to OOH∗ significantly increased from 0.62 e (without P doping) to 1.05 e (with P doping) such that the charge redistribution on Fe–Ni dual sites endowed them with stronger metallic activity (Figures 1A, 1B, and S4 and Table S1). In contrast to the bidentate bonding form on Fe-Ni-N-P-C, the OOH∗ adopted the monodentate binding form on Fe-Ni-N-C as a result of its relatively lower electron density without P doping, resulting in an increase in formation energy (1.95 eV). Therefore, the asymmetric Fe–Ni dual sites and the local electron engineering by P doping were able to synergistically achieve an optimized configuration matching with OOH∗, reducing the formation energy of the RDS in OER.As a crucial factor in practice, the thermodynamic stability of this promising Fe–Ni dual-site configuration was demonstrated by a GCMC simulation (simulation details can be found in the supplemental information). The full-atomic models with two different initial configurations of alternative and parallel distributions of Fe3+ and Ni2+ in the typical template of a ZIF-8 cage, 35 , 36 corresponding to the asymmetric Fe–Ni dual sites and symmetric diatomic Fe–Fe and Ni–Ni sites, respectively, were constructed (Figures 1C and S5). However, ZIF-8 has proved to be an ideal sacrificed template for providing N coordination sites to stabilize Fe–Ni dual sites and the C matrix. When the configuration geometry approaches equilibrium, the alternative distribution of Fe3+ and Ni2+ is much more stable in the ZIF-8 cage, whereas the parallelly distributed Fe3+ and Ni2+ ions repel each other and consequently redistribute themselves in different cages. This provides clear evidence that the formation of asymmetric Fe–Ni dual sites is more thermodynamically favorable than the symmetric diatomic Fe–Fe and Ni–Ni sites. Moreover, an out-of-plane geometric structure caused by P sp3 hybridization is able to reduce the strain of C matrix (Figure 1A) and hence improve the stability. 37 Accordingly, we precisely fabricated the Fe–Ni dual sites in N,P-co-doped C through a simple in situ adsorption-pyrolysis method 38 , 39 by using porous ZIF-8 as the cage for alternative N-coordinated Fe–Ni dual sites and the template of the C matrix, as well as by co-feeding NaH2PO2 for P doping (synthesis details can be found in the supplemental information). X-ray diffraction (XRD) and scanning electron microscopy (SEM) revealed that the intrinsic highly uniform rhombic dodecahedron crystal structure of ZIF-8 was well preserved after the incorporation of Fe3+ and Ni2+ (Figures S6A and S6B). After the pyrolysis, Fe-Ni-N-P-C still maintained the morphology of the parent ZIF-8 with a slight size shrink (Figures 2A and 2B). The Raman spectra of Fe-Ni-N-P-C possessed two D and G band peaks at 1,362 and 1,587 cm−1, respectively, and the intensity ratio of the D1 to G bands (ID1/IG) for Fe-Ni-N-P-C was calculated to be 3.79, which is higher than that for Fe-Ni-N-C without P doping (3.13), indicating the formation of a defect-rich C matrix due to the heteroatom doping (Figure S6D). In addition, the hierarchical porosity with a high Brunauer-Emmett-Teller (BET) surface area of 625 m2 g−1 and a pore size of 1–4 nm (Figure S8 and Table S3) made the catalytic active sites highly exposed and facilitated the adsorption of OH− and the diffusion of O2. Only typical graphite peaks at 24° and 43° 40 (no characteristic peaks of either Fe or Ni species) were observed in its XRD pattern (Figure S6C). The transmission electron microscopy (TEM) image of Fe-Ni-N-P-C showed a hollow structure due to the typical Kirkendall effect 41 , 42 without the formation of any metal nanoparticles (Figure 2C). Closer observation in the aberration-corrected high-angle annular dark-field (HAADF) scanning TEM (STEM) revealed many highly dispersed small bright dual dots (marked as red circles) throughout the C matrix (Figures 2J, 2K, and S7E–S7H), indicating the formation of atomic dual sites. Energy-dispersive X-ray (EDX) mapping and inductively coupled plasma (ICP) emission spectrometry showed that Fe, Ni, N, and P were highly uniformly dispersed in the C matrix (Figures 2D–2I) with mass contents of 1.3%, 1.1%, 6.75%, and 1.7%, respectively.The bonding configurations of Fe-Ni-N-P-C are illustrated by X-ray photoelectron (XPS) spectroscopy (Figure S9), in which the coordinated N-M peak at 399.6 eV can be clearly observed in the high-resolution N 1s spectrum, 43 , 44 suggesting the formation of N–Fe and N–Ni sites. The co-existence of Fe2+ (708.8 and 724.6 eV) and Fe3+ (711.8 and 726.1 eV) and the mixed valence (855.3 eV) between Ni0 (825.5 eV) and Ni2+ (855.7 eV) are evident in the high-resolution Fe 2p and Ni 2p spectra, 45 and the absence of peaks belonging to Ni3+ species can be ascribed to the partial reduction of Ni2+ by adjacent C and N atoms during the pyrolysis at high temperature. Because the prior stable N-coordinated Fe–Ni sites were in the ZIF-8 cage before the pyrolysis, the P 2p spectrum shows no characteristic peaks of P-M bonds—only P–C (132.8 eV) and P–O (133.9 eV) bonds. 46 , 47 Notably, no characteristic peaks of FePx or NiPx species are evident, which could be attributed to the prior formation of metal–N bonds in the Fe/Ni-ZIF-8 precursor during the pyrolysis; thereafter, Fe or Ni species can hardly react with the PH3 generated by NaH2PO2∙2H2O. Therefore, the designed configuration of Fe-Ni-N-P-C suggests that Fe–Ni dual sites are coordinated with N in the C matrix with P doping nearby (Figure 3 A).This proposed atomic configuration of Fe-Ni-N-P-C (Figure 3A) was further verified by X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine-structure (EXAFS) measurements. Fe-Ni-N-P-C showed a Ni K-edge XANES spectrum similar to that of Ni phthalocyanine (NiPc), in which a stronger peak occurs at around 8,334 eV (Figure 3B) as a result of the distorted D4h symmetry of Ni caused by the metal-metal path. 48 , 49 Moreover, the EXAFS spectrum of Ni in Fe-Ni-N-P-C (Figure 3C) exhibits two main peaks at 1.58 and 2.44 Å, suggesting the co-existence of Ni–N and metal–Ni sites, respectively, neither of which exists in the spectrum of SAC Ni-N-C or Fe-N-C 50 or is the location of the Ni-Ni or Fe-Fe peak in Ni or Fe foil, respectively; thus, this peak can be attributed to Fe–Ni dual sites. Similar results were also obtained from the Fe K-edge XANES and EXAFS spectra (Figures 3D and S10A), which show distinct Fe–Ni dual sites at 2.48 Å. It is worth noting that the peaks of Ni–N (1.58 Å) and Fe–N (1.52 Å) show significant shifts compared with NiPc (1.30 Å; Figure 3C) and FePc (1.45 Å; Figure 3D), and this can be attributed to the metal-metal interaction between Fe–Ni dual sites and the sp3 hybridization P doping with a relatively large atomic radius. It is worth mentioning that the shifted Ni–N and Fe–N peaks might overlap the reported Ni–O (1.6 Å) and Fe–O (1.5 Å) peaks, 18 , 51 but it is hard to form Ni-O and Fe-O bonds because of the lack of O-rich coordinated sites in the ZIF-8 cage and the much weaker metal-O coordination than metal-N coordination. 52 Therefore, we think that Fe3+ and Ni2+ are coordinated with N sites in the ZIF-8 cage and convert to Fe/Ni–N species during the pyrolysis. More specifically, the DFT calculation also confirms this proposed atomic configuration of Fe-Ni-N-P-C. The first shell-fitting results reveal that the coordination number of Fe–N and Ni–N is 3.65 and 3.35, respectively, suggesting the co-existence of N3–Fe–Ni–N3 dual-metallic coordination and Fe–N4 and Ni–N4 single-metallic coordination (Figure S10B and Table S4). We also calculated that 66% of metal–N bonds in Fe-Ni-N-P-C are derived from Fe–Ni dual sites, and the rest of the N is coordinated with single Fe or Ni atoms (details in Figure S11). More importantly, the calculated Fe–Ni path length of 2.37 Å (Figure S10C) is in good agreement with the EXAFS spectra of 2.44 Å for Ni–M and 2.48 Å for Fe–M (Figures 3C, 3D, and S10D). Both second shell-fitting results of Ni and Fe K-edge EXAFS also fit well with the experimental spectra (Figures 3E and 3F).As a proof of concept, determined by the rotating disc electrode (RDE) approach, Fe-Ni-N-P-C with an Fe/Ni mass ratio of 1:1 exhibited the smallest overpotential of 337 mV at a current density of 10 mA cm−2 in 0.1 M KOH (Figure 4 A) and further decreased to 250 mV in 1 M KOH (Figures S12 and S13A). More significantly, the calculated mass activity was as high as 8,894 A g metal−1 at an overpotential of 350 mV, and the corresponding turnover frequency (TOF) reached 0.66 s−1. To the best of our knowledge, this exceeds all the reported transition-metal N-based catalysts (M-N-C) (Table S5). It is worth mentioning that, compared with single-atom Fe-N-C and Ni-N-C, the physical mixture of Fe/Ni-N-C-PM, and N-C with trace Zn (0.015%) counterparts, Fe-Ni-N-C exhibited a much smaller overpotential of 395 mV to deliver the current density of 10 mA cm−2, indicating that excellent OER activity originates in the atomic Fe–Ni dual sites rather than single-atom Fe–N/Ni–N sites or the trace Zn site. In addition, the smaller OER overpotential of Fe-Ni-N-P-C (335 mV) than of Fe-Ni-N-C (395 mV) demonstrates the important role of P doping for enhanced OER activity, which is consistent with the results of the DFT calculation.In addition, the slope of the Tafel plot of Fe-Ni-N-P-C was calculated to be only 76 mV dec−1 (Figure 4B), much lower than that of commercial RuO2 (120 mV dec−1); moreover, the larger slope of Fe-Ni-N-C without P doping (100 mV dec−1) further demonstrates the synergistic effect of Fe–Ni dual sites and P doping on facilitating the OER kinetics through configuration matching and local electronic environment modulation. In addition, the intrinsic charge-transfer resistance of Fe-Ni-N-P-C was also measured by electrochemical impedance spectroscopy (EIS) analysis. The Nyquist plot of Fe-Ni-N-P-C showed the smallest semicircle in the low-frequency region (Figure S14), indicating the charge-transfer superiority over other controlled samples. Moreover, attributed to the thermodynamically stable Fe–Ni dual sites, Fe-Ni-N-P-C demonstrated excellent durability over 1,000 cyclic voltammetry (CV) tests (Figure S13B), and no obvious peak shifts were observed in the high-resolution Fe 2p or Ni 2p spectra of Fe-Ni-N-P-C after the OER cycle (Figure S15). A 12 h, the chronoamperometric response remained 86% of the initial current density (Figure 4C), superior to that of commercial RuO2 (66%).More particularly, Fe-Ni-N-P-C also exhibited an attractive ORR activity with a half-wave potential (E1/2) of 0.823 V (versus reversible hydrogen electrode [RHE]; Figure S16A), close to that of commercial 20% Pt/C (0.830 V); its lower Tafel slope of 87 mV dec−1 also demonstrates the favorable reaction kinetics compared with those of Pt/C (90 mV dec−1; Figure S16B). Moreover, further analysis of the Koutecky-Levich plot and rotating ring-disk electrode (RRDE) testing (Figures S16D–S16F) both showed the near-four-electron ORR pathway for Fe-Ni-N-P-C with low H2O2 yield (4%). Importantly, no obvious decay was observed in E1/2 after 3,000 continuous potential cycles, indicating superior long-term stability to Pt/C (Figure S16C). Attributed to both excellent OER and ORR activities, the reversible overpotential (ΔE) achieved on Fe-Ni-N-P-C (0.744 V; Figure 4D) was lower than that on commercial Pt/C (1.015 V), RuO2 (0.960 V), and many studied OER and ORR bifunctional electrocatalysts (Table S6). Inspiringly, a rechargeable Zn-air battery was assembled to illuminate the orange light-emitting diode (LED) with an open-circuit voltage of 1.458 V (battery assembling details can be found in the supplemental information) (Figures 5A, 5B, and S17). Other excellent characteristics include high constant current discharge potential (1.24 V at a current density of 10 mA cm−2), excellent discharge peak power density (120 mW cm−2), small charge-discharge potential gap at high current density, and long-term cycle stability with an energy efficiency of 64% after 90 h, which is superior to that of the Pt/C + RuO2 mixture (57%; Figures 5C–5E and S17). It is worth mentioning that although the polarization curve of Pt/C + RuO2 coincided with that of Fe-Ni-N-P-C at a lower current density, as the current density increased (>225 mA cm−2), the voltage of Fe-Ni-N-P-C stayed relatively high while the voltage of Pt/C + RuO2 cathode dropped dramatically (Figure S17). This superiority is consistent with its small charge-transfer resistance and large active surface area. Attributed to its porous hollow structure, Fe-Ni-N-P-C provides facilitating channels for O2 diffusion and electrolyte transportation and ultimately enhances electron transport and the electrode reaction. The assembled battery performance based on the Fe-Ni-N-P-C cathode was superior to that of not only commercial Pt/C + RuO2 but also many other reported cathode catalysts (Table S7). The promising application of the Fe-Ni-N-P-C-based cathode in the rechargeable Zn-air battery provides clear evidence that it is a practical catalyst design and engineering strategy at the atomic level.In summary, we have demonstrated a DFT-guided strategy for OER catalytic active-site design at atomic precision. The constructed Fe–Ni dual sites with P doping are able to facilitate the formation of OOH∗ through geometric matching because of their asymmetric affinities with OH∗ to O∗, as well as the local electron environment engineering for strong electron donation. Confirmed by the GCMC simulation, the thermodynamically stable catalyst of Fe-Ni-N-P-C has been precisely fabricated through the pyrolysis of alternative N-coordinated Fe–Ni dual sites in the ZIF-8 cage by co-feeding NaH2PO2 for P doping. The resultant Fe-Ni-N-P-C has demonstrated a superior OER activity with a very low overpotential of 250 mV at a current density of 10 mA cm–2 and a high TOF of 0.66 s–1 at an overpotential of 350 mV, largely exceeding those of commercial RuO2 and all reported transition-metal N-based catalysts. Moreover, a high-performance rechargeable Zn-air battery has been achieved with attractive ORR activity (half-wave potential of 0.82 V). We believe that this robust strategy of atomic configuration matching with local electron modulation opens a new window for electrocatalyst construction.Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Jun Hu (junhu@ecust.edu.cn).All materials generated in this study are available from the lead contact without restriction.All data needed to support the conclusions of this manuscript are included in the main text or supplemental information.The Fe–Ni dual sites in N,P-co-doped C were synthesized by a double-solvent method. Typically, ZIF-8 was dispersed in n-hexane (10 mL) under ultrasound for 10 min at room temperature. Subsequently, a controlled amount of FeCl3∙6H2O aqueous solution (100 mg mL−1, 30 μL) and Ni(NO3)2 aqueous solution (100 mg mL−1, 30 μL) was slowly injected into the ZIF-8 n-hexane solution under stirring. Then the whole solution was subjected to ultrasound for another 10 min. After that, the mixed solution was vigorously stirred for 3 h at room temperature. Finally, the yellow powder of Fe/Ni-ZIF-8 was centrifuged and dried in vacuum at 65°C for 6 h. The Fe/Ni-ZIF-8 sample and NaH2PO2 H2O were separately placed into two porcelain boats with NaH2PO2∙H2O at the upstream side of the furnace. The mass ratio of Fe/Ni-ZIF-8 to NaH2PO2∙H2O was 1:5. The annealing was performed at 1,000°C for 2 h at a heating rate of 5°C min–1 in N2 atmosphere. The resultant catalysts were denoted as Fe-Ni-N-P-C. For comparison, by adjusting the mass ratio of Fe and Ni, Fe-N-C, Ni-N-C, and Fe-Ni-N-C, we synthesized FexNiy-N-P-C via a similar procedure without adding some specific species as precursors. We also pyrolyzed ZIF-8 at 1,000°C for 2 h in N2 atmosphere to investigate the existence of Zn and its contribution to OER performance.This work was supported by the Natural Science Foundation of China (nos. 91834301, 21676080, and 21878076) and the Science and Technology Commission of Shanghai Municipality (no.19160712100).F.P. conceived the idea, prepared and characterized the catalysts, performed the catalytic measurements, and wrote the manuscript under the supervision of J.H. and H.L. T.J. analyzed the data and revised the manuscript. X.D., Y.C., and X.Z. provided valuable discussions and suggestions for manuscript revision. W.Y. and H.L. helped to perform the DFT calculation and GCMC simulation. All authors contributed to the preparation of the manuscript and gave approval to the final version of the manuscript.The authors declare no competing interests.Supplemental information can be found online at https://doi.org/10.1016/j.checat.2021.06.017. Document S1. Supplemental experimental procedures, Figures S1–S19, and Tables S1–S7 Document S2. Article plus supplemental information

The principle of how the active sites of catalysts match the reaction intermediates has long been sought after. Herein, we report a theory-guided atomic design and fabrication strategy of a C-based catalyst with diatomic Fe–Ni and N,P co-doping for the oxygen evolution reaction (OER). The configuration matching (with O∗ on the Ni site and OH∗ on the adjacent Fe site) and the local electron engineering by P doping significantly facilitate the rate-determining step of OOH∗ formation. Such diatomic Fe–Ni is demonstrated to be thermodynamically stable and is precisely constructed through the pyrolysis of Fe3+/Ni2+-adsorbed ZIF-8 under NaH2PO2 co-feeding. The synergistic effects endow the catalyst with a low overpotential and high turnover frequency, exceeding all transition-metal N-based catalysts so far as we know, which provides a deep understanding of the OER mechanism on heteroatomic metal-based catalysts. This strategy will pave the way for novel catalyst design and the replacement of noble-metal-based catalysts.

Data will be made available on request. Data will be made available on request.Plastic waste pollution is one of the most serious environmental issues worldwide today. A sustainable waste management strategy is necessary to overcome this problem. One of the most investigated methods for plastic waste treatment is pyrolysis. The term “pyrolysis” refers to a variety of thermal and thermo-catalytic conversion processes and technologies aimed at producing (a) liquid and gas products composed mainly of hydrocarbons (HC), oxygenated HC (OHC), and hydrogen (H2), and (b) carbon-rich solids. The pyrolysis-catalysis of plastic waste has already been extensively researched for the production of carbon nanomaterials (CNMs) and H2 [1]. In a two-stage process, the plastic is pyrolyzed to produce different hydrocarbons that pass through a catalytic bed for cracking and catalytic decomposition. So far, evidence has been obtained that converting waste plastics into valuable products via pyrolysis-catalysis processes could be a promising alternative for plastic waste management [2].CNMs, especially carbon nanotubes (CNTs), have remarkable and valuable properties, including high electric and thermal conductivities [3]. These materials can be used in several applications, such as composites [4], catalysis [5], adsorption for environmental cleaning [6–8], field-effect emitters [9], and electrode materials for supercapacitors [10]. Carbon-based light molecules, such as methane, ethylene, and acetylene, are usually used as carbon precursors for synthesizing CNMs. However, the carbon contained in plastic waste is a better alternative as the raw material has a negative value. Its use as feedstock renders the intended processes environmentally friendly and provides a sustainable approach to the mass production of CNMs.As expected, catalysts play an important role in the production of CNMs. Metal-based catalysts are the most commonly used. Accomb et al. [11] have investigated the effect of using different transition metals (Fe, Ni, Co, and Cu) supported by alumina on CNT production from low-density polyethylene (LDPE). Fe- and Ni-based catalysts gave the largest yield of CNTs and hydrogen, followed by Co. Cu gave no filamentous carbon because of weak metal–support interaction [11]. Fe-based catalysts have a desirable catalyst–support interaction and large carbon solubility [12]. Moreover, regarding the degree of graphitization, Fe-based catalysts also perform better [13]. Other authors [6,14] have tested bimetallic catalysts to understand the synergistic effect between the metals. Increasing the Ni to Fe molar ratio in an Fe-Ni catalyst enhances the thermal stability and graphitization of the formed carbon. Moreover, the yields of CNM and hydrogen also increased. Ratkovic et al. [15] have demonstrated that the yield of CNTs from catalytic decomposition of ethylene over Fe-Ni/Al2O3 is three times higher than over Fe/Al2O3. The performance of the bimetallic catalyst has been attributed to the strong metal–support interaction that leads to well-dispersed small particles.The size and the shape of the metal particles are the most important factors determining the type of CNM produced. When the particles are a few tens of nanometers and well-dispersed, CNTs are produced instead of carbon nanofilaments CNFs [12]. Aboul-Enein and Awadallah [16] have studied the production of CNMs using Fe-Mo/MgO catalysts during catalytic pyrolysis of waste PE. They found that the Fe/Mo ratio has a key effect on the type and morphology of the produced CNMs. With high loads of Fe and Mo, large-diameter carbon nanofilaments (CNFs) and hollow CNTs were formed due to the aggregation of metal particles. In another study [17], the same authors studied the effect of adding Cu to Ni-La2O3 on the decomposition of the non-condensable gases produced from the pyrolysis of polypropylene (PP). The bimetallic Cu-Ni particles were in a quasi-liquid state, which increased the size of the metal particles and led to the formation of large-diameter cap-stack CNFs, while highly dispersed Ni particles were responsible for the growth of multi-walled carbon nanotubes (MWCNTs).The type of plastic has also impacted the production of CNMs. Polyethylene (PE) and PP are usually used because of their abundance and high carbon content. Aboul-Enein et al. [18] have demonstrated that the morphology and crystallinity of the CNM depend on the type of plastic waste. LDPE and PP produced MWCNTs with a high degree of graphitization, while high-density polyethylene (HDPE) produced MWCNT with a rugged surface. A small yield with low quality and purity was obtained from polystyrene (PS) and polyethylene terephthalate (PET). Similar results have also been obtained by Veksha et al. [19], who also found that the influence of plastic type is more pronounced at lower temperatures. Real-world plastic waste was also investigated [13] and the carbon produced was mainly filamentous with some amorphous carbon. Moreover, contaminants in the feedstock are known to affect the pyrolysis process and products [20]. Wu et al. [21] have demonstrated that 0.3 wt% of polyvinyl chloride (PVC) in the feedstock led to a significant reduction in the quality and purity of CNTs.Despite abundant research into CNT production using catalysis-pyrolysis, there is a lack of data collected in continuous feeding mode. To fill this gap, this work investigates the synthesis of CNMs from waste plastic using a new catalyst in a continuous feeding mode. The support used for synthesizing the catalyst is UGSO (UpGraded Slug Oxides), a negative-value mining residue. These oxides contain a significant amount of Fe, making them very attractive for synthesizing CNMs. The combination of Ni with these oxides enhances their catalytic activity, as shown in previous studies [22,23]. The main targets of this study are: 1) Investigate the performance of the Ni-UGSO catalyst during the pyrolysis-catalysis of waste HDPE for the synthesis of CNMs and compare its activity with a catalyst widely used in literature (Fe/Al2O3). 2) Study the effect of plastic waste on CNM synthesis by using different plastics (virgin HDPE, used HDPE, and mixed plastics). Investigate the performance of the Ni-UGSO catalyst during the pyrolysis-catalysis of waste HDPE for the synthesis of CNMs and compare its activity with a catalyst widely used in literature (Fe/Al2O3).Study the effect of plastic waste on CNM synthesis by using different plastics (virgin HDPE, used HDPE, and mixed plastics).This work presents a new concept for synthesizing CNMs from waste plastic. The process has an ecological impact because it valorizes waste plastic. Moreover, the catalyst used is also made from a negative-value mining residue.Virgin HDPE was purchased from McMaster-Carr. The pellets are spherical, with a diameter of 6.35 mm. The post-consumer HDPE and the mixed polyolefin samples were obtained from KWI Solutions Polymers, Inc. The HDPE waste particles have different shapes and dimensions, varying from 3 mm to 1 cm. The post-consumer mix of polyolefins has an average composition of 80 wt% HDPE and LDPE, 15 wt% PP, 4 wt% PC and PET, and 1 wt% metals and wood. A photograph of the different plastics is provided in Appendix A (Fig. A.1). All plastics were used without further grinding or treatments. UGSO, which is a mix of different oxides, was provided by Rio Tinto Iron & Titanium and it is the same material used in previous publications [22–24] Nickel and iron nitrates were purchased from Sigma-Aldrich, and γ-Al2O3, with particle size of 210 µm, was purchased from McMaster-Carr.Both UGSO and γ-Fe2O3 were ground and screened using a sieve with a size of 53 µm. Ni-UGSO and Fe/Al2O3 were prepared by incipient wetness impregnation. Both catalysts were dried for 3 h at 105°C. Ni-UGSO was calcined for 16 h at 900°C, according to a previous study [24], while Al2O3 was loaded with 10 wt% of Fe and calcined for 2 h at 750°C, as reported in the literature [25]. According to a previous study, the Ni loading of 10 wt% is a concentration that favors carbon formation [23]. After calcination, both catalysts were crushed and sieved to give particles of size below 53 µm.The pyrolysis-catalysis of the plastic waste was performed in a two-stage fixed-bed quartz reactor. Pyrolysis takes place in the first stage. The second stage consists of a catalytic bed for catalysis of the produced gases. Temperature can be controlled in both stages separately. The experimental setup is described in more detail in previous work [26]. For each experiment, 5 g of catalyst was dispersed in quartz wool positioned homogeneously in the middle of the catalysis zone of the cylindrical reactor. Another quartz wool piece was placed in the middle of the pyrolysis stage to receive the waste plastic which falls inside the reactor. In this process, metals and other contaminants were retained in the pyrolysis step as char residue. For each experiment, the catalyst was activated with high purity H2 at a flow rate of 0.10 SLPM for 3 h at 600°C. After 3 h, the hydrogen flow was stopped and replaced by N2 at a flow rate of 0.03 SLPM. Following a previous study [22], for all experiments, the temperature of the catalysis stage was fixed at 650°C, while that of the pyrolysis zone was fixed at 700°C to favor gas production. When both temperatures reached their set point, the reaction began by feeding the plastic using a two-stage feeder at a rate of 0.33 g·min−1. The duration of each reaction was 2 h; this restriction is due to the fact that the carbon formed accumulated in the reactor and caused a pressure rise beyond the maximum acceptable level. Condensable liquids were recovered in a cold trap immersed in an ice bath. The gas exiting this condenser passed through a charcoal column so that all possible liquid hydrocarbons were retained as a mist. Gas samples were taken every 10 min for gas chromatography (GC) analysis using a SCION 456-GC equipped with flame ionization detector FID and thermal conductivity detector TCD. The experiments were duplicated and the error is below 5%. Average values were taken for all data provided.At the end of each reaction, the solid accumulated on the quartz wool and catalytic bed (catalyst on quartz wool) was weighed to estimate the yield of solid products, Ysolid using Eq. (1). (1) Y s o l i d ( w t % ) = m s o l i d m p l a s t i c × 100 The cold trap and the adsorbent were also weighed to determine the yield of the liquids and wax, Yliquids. (2) Y l i q u i d s ( w t % ) = m l i q u i d s + w a x m p l a s t i c × 100 The gas yield, Ygas, was determined according to the following equation: (3) Y g a s ( w t % ) = 100 − Y l i q u i d s − Y s o l i d The yield of filamentous carbon was determined as: (4) Y f i l a m e n t o u s c a r b o n ( w t % ) = m f i l a m e n t o u s c a r b o n m c a r b o n c o n t a i n e d i n t h e p l a s t i c × 100 The yield of hydrogen was determined as: (5) Y h y d r o g e n ( w t % ) = m h y d r o g e n × 2 m h y d r o g e n c o n t a i n e d i n t h e p l a s t i c × 100 X-ray diffraction (XRD) was used to analyze both catalysts before and after the pyrolysis-catalysis experiments; XRD provided information about the different crystalline phases in the fresh and used catalysts. A Philips X'pert PRO diffractometer (PANalytical) with a CuKα radiation source producing at 40 kV and 50 mA was used. The diffraction spectra were collected in the range of 15°–80° with a step of 0.05° per 700 s. The data were analyzed with the MDI JADE 010 software.The Brunauer–Emmett–Teller (BET) multipoint method was used to compare the specific area and pore volume of both fresh catalysts. Samples were exposed to N2 physisorption at 110°C for 18 h with an accelerated surface area and porosimetry system (ASAP 2020 V4.01).A temperature-programmed reduction (TPR) was used to quantify the amount of reduced oxides in both catalysts. Samples of ⁓30 mg of catalyst were deposited in a U-shaped quartz tube. The sample was pretreated with Ar at a flow of 20 mL·min−1 for 1 h at 140°C. The catalyst was reduced using a gas mixture (10% H2 in Ar), with a flow rate of 20 mL·min−1, while it was heated from room temperature to 1100°C at 3°C·min−1. To remove the H2O formed during the reduction period, the gas goes through a cold trap containing 2-propanol on liquid N2. The H2 variation was detected by a cathetometer, and its consumption is proportional to the peak area.In order to examine the morphologies of the CNMs, a Hitachi SU8230 was used in scanning transmission electron microscopy (STEM) mode. The energy dispersive X-ray spectroscopy (EDXS) detector used was a FlatQuad 5060F (Bruker, Germany). SEM images were taken using an in-lens SE detector with an acceleration voltage of 2 kV. Samples were prepared by dispersing the particles into ethanol with ultrasound, then dropping the suspension on a copper grid coated with amorphous carbon.Thermogravimetric analysis (TGA) was used to quantify the amount of carbon deposited on the surface of the catalysts and distinguish filamentous carbon from amorphous carbon. The analyses were performed using a Setaram Setsys 24 analyzer under a stream of 20% O2 and 80% Ar, from room temperature to 1000°C at a heating rate of 10°C· min−1.Raman spectroscopy (SP2500 Acton spectrometer) with a 30-mW 414-nm laser was used to determine the intensity ratio (G/D) and provide information about defects present in the graphene sheets.The physicochemical properties of the fresh catalysts are presented in Table 1 . Ni-UGSO shows the highest surface area and pore volume. However, both Ni-UGSO and Fe/Al2O3 have similar average pore volumes. The difference in terms of H2 chemisorption is very important. After reduction, Ni-UGSO has three times more metal-active sites than Fe/Al2O3. This is expected, as UGSO contains nearly 30 wt% of iron, according to its elemental analysis [24], plus the added Ni.The XRD results of both catalysts are shown in Fig. 1 . UGSO is a mix of different oxides, most of which are spinels. When Ni is added to UGSO, the peaks at 36.9, 42.9, 62.4, and 78.7° increase in intensity. These angles correspond to the spectra of NiO and MgO. EDXS mapping from a previous study [23] has shown that NiO formed a solid solution with the MgO present in UGSO. The presence of the Ni-Mg-O solid solution offers the highest dispersion of Ni in a basic environment [27]. Thus, the metal–support interaction is very strong. For Fe/Al2O3, alumina (Al2O3) and hematite (Fe2O3) were detected in the XRD spectrum. However, the formation of hercynite (FeAl2O4) is not significant. The most intense peak in the FeAl2O4 spectrum is at 36.5°, and it also appears in the spectrum of the Fe/Al2O3 catalyst at low intensity. This indicates that a minor quantity of Fe formed FeAl2O4, and the rest is present in the form of Fe2O3. This may be due to the calcination method: a higher temperature and longer calcination times would favor the formation of FeAl2O4, as reported in the literature [28]. Table 2 presents the results of the pyrolysis-catalysis of post-consumer HDPE using Ni-UGSO. The yields of filamentous carbon and H2 are 68 wt% and 79.4 wt%, respectively. The conversion is seen to decrease in the gas composition slightly over time, shown in Fig. 2 . This is expected as the access to the active sites of the catalyst becomes more and more limited as the filamentous carbon forms. These results are quite similar to those obtained during the catalytic cracking of ethylene using the same catalyst [22]. At the beginning of the reaction, the yield of the solid is high, and that of the liquid is low. Over time, the hydrocarbons produced from the pyrolysis of waste HDPE have less access to the metal sites, and the rate of catalytic cracking decreases. As a result, more liquids and waxes are produced from thermal cracking activity. At these conditions, the performance of Fe/Al2O3 is very limited, with a yield of only 3.72 wt% of filamentous carbon and a 29 wt% yield of liquids. Previous studies have shown that Fe/Al2O3 performs better for producing CNMs and H2 during the catalytic pyrolysis of waste plastic [11–13]. However, all the experiments reported in these studies were conducted at higher temperatures (700, 800, and 900°C). Higher temperatures up to 800°C are known to increase the production of filamentous carbon, as the kinetics of hydrocarbon decomposition and diffusion through the metal particles rates increases [13]. As shown in the TPR results, the differences in composition influence the reducibility and the dispersion of active metals in the tested catalysts. Table 3 compares the results of this work with other reported results. Ni-UGSO produces a high yield of carbon and H2 at a lower temperature (650°C). In the case of this work, average yields during 2 h of reaction are presented. However, the results reported in previous works are from batch regime experiments. Nevertheless, the relatively excellent catalytic performance of Ni-UGSO can be attributed to the presence of Fe in UGSO. In this study, the temperature is significantly lower than the other temperatures reported in previous reports, which makes this process more economical. Moreover, the catalytic temperature affects the quantity and quality of the produced CNMs. An increase in temperature might enhance the yield and the quality of CNMs [13]. The reduction of Ni-UGSO with H2 prior to each experiment leads to the formation of metallic Fe, Ni, and Fe-Ni alloys, as has been proven in previous studies [22,23]. These metals are the active phases that produce atomic carbon, leading to CNMs [12]. This is discussed in more detail in Section 3.2.1.After the reaction, the XRD spectrum of Ni-UGSO in Fig. 3 shows the presence of metallic Ni, Fe, and Fe-Ni alloys. This result confirms previous claims regarding the presence of these metals at the surface of the catalyst and their major role in decomposing hydrocarbons. The Ni in the Ni-Mg-O solid solution is reduced to metallic Ni, whereas the Fe in hematite (Fe2O3) was reduced to wüstite (FeO) and then to metallic Fe. The same reduction of Fe occurs in Fe/Al2O3; however, the peak of metallic Fe is not as intense as in the case of Ni-UGSO, confirming the TPR results about the difference in metal content at the surface of both catalysts. The diffraction peak at 2θ= 26° corresponds to the d002 of graphitic carbon. In the case of Ni-UGSO, this peak is intense and sharp, indicating the presence of crystalline carbon. This peak is also present in the case of Fe/Al2O3 at a lower intensity.From these results, Ni-UGSO has a metal–support interaction sufficient to avoid catalyst sintering but not high enough to inhibit CNM formation. A strong metal–support interaction reduces the amount of the surface accessible to hydrocarbons [12]. Moreover, a strong metal–support interaction prevents the migration of carbon into the subsurface of the metal layer [29] and disturbs the distribution of carbon atoms over the catalyst particle, leading to the formation of defects in the CNMs.The morphologies of the filamentous carbon formed on Ni-UGSO are presented in Fig. 4 . The filaments are of different diameters (between 8–90 nm). The high-angle annular dark-field (HAADF) image shows some metal particles located at the tip of the filaments and others trapped inside them. The trapped metal particles have a smooth morphology, whereas the metals at the tip of the CNMs have an angular form. The non-uniformity of the filaments is expected, as UGSO is a mining residue containing different particle sizes. Thus, the metal particles formed during the catalyst activation are not uniform in size. Furthermore, the size of these metals can change because of sintering phenomena, especially at high temperatures. It has been proven that the size of CNMs is directly related to the particle size of the metal crystals [30,31].EDXS analysis of a metal particle at the tip of a filament is presented in Fig. 5 . The peak of Fe is intense; Ni and C are also present. The Cu comes from the support, not the sample. This indicates that the metal particle contains both metals, and from the bright field scanning transmission electron microscopy (BFSTEM) image, it is also covered by filamentous carbon. Ni interacts very efficiently with hydrocarbon molecules and promotes their decomposition, whereas carbon diffusion and nucleation occur on the surface of Fe. In other words, Ni dehydrogenates the adsorbed hydrocarbons more quickly, while Fe solubilizes carbon better than Ni [32]. Yao and Wang [28] have demonstrated the synergetic effect of Fe and Ni during the pyrolysis-catalysis of PP for the production of CNMs. They found that the presence of both metals led to the production of highly graphitized bamboo-like MWCNTs compared to monometallic catalysts.The state of the metal particles in the BFSTEM image in the inset of Fig. 5 is in accordance with the observations of Krivoruchko and Zaikovskii [33]. These authors have demonstrated that the metal–carbon particle is in a quasi-liquid state. A nanometer-scale metal particle has very high surface energy, which leads to weaker bonds between metallic atoms. Consequently, the metallic surface melts and spreads on the carbon [34]. This reshaping of the metal particles explains the different shapes of metallic particles seen in Fig. 4. The mechanistic model suggested for the growth of these filaments is called the vapor–liquid–solid (VLS) model [35]. The process starts with hydrocarbon adsorption and decomposition on the active sites to produce atomic carbon. The latter diffuses inside the metal as liquid metastable carbides until saturation is reached. Finally, carbon precipitates to grow a filament.TGA analysis of both catalysts is presented in Fig. 6 . The weight loss for Ni-UGSO is estimated at 75 wt%, confirming the yield of filamentous carbon calculated by the mass balance. Meanwhile, the mass loss of Fe/Al2O3 is about 15.3 wt%, which is an overestimate for this small sample because the mass balance gave a smaller carbon yield. The derivative plot shows that the weight loss peak for Ni-UGSO occurs at 620°C, while for Fe/Al2O3, it occurs at a lower temperature. The higher the degree of structural order of the filaments, the higher the oxidation temperature is. The oxidation of pure graphite occurs at 645°C [36]. The absence of a peak at a temperature lower than 600°C indicates that no amorphous carbon was produced during the pyrolysis-catalysis of used HDPE using Ni-UGSO as a catalyst. Consequently, all the deposited carbon on the catalyst surface is crystalline. The STEM results discussed in Section 3.3.1 will help identify the type of CNM produced. Table 4 presents the total filamentous carbon and H2 yields for each feedstock. There is no significant difference between the products of virgin and used HDPE. This result indicates that the used HDPE might not be highly contaminated. However, for the mixed plastics, the yield of deposited carbon decreases by 10 wt%. This is attributed to impurities present in mixed plastics. Moreover, The differences in polymers, size, and physical properties of mixed plastics should be considered potential, or at least partial, causes for the observed differences. The average carbon production rate decreases, and the mass of solid residue found in the quartz wool increases. Metal and wood particles are also observed on the quartz wool. Catalyst poisoning occurs because of the different impurities, which might cause a decrease in catalyst activity. A comparison of the detailed molar composition of the gaseous streams is provided in Fig. A.2 of Appendix A.BFSTEM analysis of filamentous carbon produced from different plastics is presented in Fig. 7 . These filaments have different diameters, as explained previously. Most of the filaments have a tubular shape, where the graphene layers are parallel to the growth axis. These filaments are not smooth and have a large diameter, which indicates that they are tubular CNFs. The bamboo-like CNFs are also observed in experiments using virgin and used HDPE, as shown in Fig. 8 . Fig. 7(b) shows a torn filament to look like a helical nanofiber. The same types of CNMs have been reported in the literature reporting the pyrolysis-catalysis of plastic waste [2,14,25,37]. The distance between the graphitic sheets is around 0.33 nm for all produced CNFs. A closer observation shows a layer of amorphous carbon covers the CNF in Fig. 7(f). This layer has an average thickness of 3.7 nm near the tip of the CNF, which increases to 4.1 nm on the other side. This phenomenon is not observed in virgin and used HDPE, only for mixed plastics. This suggests that the quality of CNFs produced by mixed plastics is lower than those from a single type of plastic. The lower carbon yield and the presence of amorphous carbon are attributed to different contaminants, such as wood, oxygenated plastics such as PET, and PS. The presence of PET and PS in the feedstock is known to decrease the quality and quantity of CNMs, as reported in other studies [18,19]. PET and PS produce smaller quantities of gases during pyrolysis [38]. In addition, the pyrolysis of PET generates oxygenated compounds, while the pyrolysis of PS produces a high liquid fraction containing mainly styrene and aromatics [39].According to Chen et al. [40], a balance between the dissociation rate of hydrocarbons and carbon diffusion rate must be maintained to ensure the continuous growth of CNFs. When the decomposition rate of hydrocarbons is high due to factors such as high metal loading or high temperatures, the balance is not maintained, and this causes some defects in carbon structure, such as carbon onions. According to the results of this work and the absence of carbon onions, Ni-UGSO demonstrates a good balance between the dissociation and diffusion rates, assuring the continuous growth of CNMs. Fig. 9 shows the derivative plots of the TGA analysis of the spent catalyst used for different types of plastics. One weight loss peak for virgin and used HDPE appears above 600°C. This indicates that the carbon produced from these two plastics is crystalline and not amorphous. For the mixed plastics, two peaks are present: one at 554°C and the other at 600°C. The first peak is attributed to amorphous carbon, and the second to crystalline carbon. These results are in accordance with the BFSTEM images.The Raman spectroscopy results are presented in Fig. 10 . Peaks are seen at around 1350 and 1585 cm−1 for all samples. The peak at 1350 cm−1 (D peak) is the scattering peak of the disordered component, while the peak at 1585 cm−1 (G peak) is the resonance peak of graphite. The G´ peak observed at ⁓2700 cm−1 is associated with the process of two-photon elastic scattering, indicating the purity of the CNM. The G/D ratio measures the quality and crystallinity of the CNMs, with a higher G/D ratio indicating a higher purity. For the virgin HDPE, the G/D ratio is low (0.94), indicating that the CNFs are of low quality. The CNFs from mixed plastics also have low quality, as the G intensity equals the D intensity. In addition, the G´/G intensity is the lowest, and the G´ is broad and not as sharp as the other G´ peaks. However, this is because of the presence of filamentous carbon seen in the STEM results. The CNFs from used HDPE showed the highest quality compared to the others, with a G/D ratio of 1.13. The low G/D ratio for the CNFs from virgin plastic is not expected as it is supposed to be the purest feedstock, particularly compared to the used HDPE and mixed plastics. The analysis was repeated several times, and the results were consistent, suggesting the presence of defects in the graphitic lattice, such as edge dislocations.To improve the quality of the CNMs, some studies suggest adding steam or other oxygenated compounds, such as CO2, to the pyrolysis gases [21,41,42]. Acomb et al. [42] have increased the G/D ratio of CNTs (produced from PS) from 1.08 to 1.43 by adding steam at a rate of 0.25 g·h−1. However, the steam decreases the quality of CNTs produced from PP and LDPE. Wu et al. [21] have reported that adding steam reduces the quantity of amorphous carbon. The MWCNT Raman peaks are smoother, but the D peak increases. The authors have observed that the nature of the CNTs changes, becoming more tangled. Azara et al. [22] have demonstrated that the presence of CO2 enhances the quality of the CNFs produced from the dry reforming of ethylene using Ni-UGSO as a catalyst. Compared to other results from the literature, the CNFs produced in this work have low Raman indicators, as shown in Table 3. Optimization is required to enhance the quality of the produced CNFs by either varying different parameters (such as temperature and Ni content) or adding oxygenated precursors (such as steam and CO2).Plastics decompose in the first stage of the reactor to give different hydrocarbon products. The thermal cracking HDPE at 700°C gives 79 wt% of gas, composed of 37 wt% C2H4 and 32 wt% of CH4, as shown in Table A.5 of Appendix A. For polyolefins, the decomposition mechanism is β-scission [20]. Then, the gases pass through the catalytic bed at 650°C, where the macromolecules decompose to lighter components. The yield of liquids for the pyrolysis test without a catalyst is 20.0 wt%, and that of the pyrolysis-catalysis tests is 8.4 wt%. This confirms the further degradation of macromolecules in the catalytic zone. In the presence of Ni-UGSO, the C-C and C-H bonds are cleaved mostly by the Ni, which is present either alone or alloyed with Fe. Carbon atoms form the graphene layers, and hydrogen atoms are released as H2 gas at the outlet of the reactor.A previous study [22] has shown that for Ni-UGSO catalyst, the Ni (111) plane is responsible for the adsorption and dissociation of hydrocarbons, while the diffusion of atomic carbon occurs mainly on the Fe (110) plane. The atomic carbon produced has a greater affinity for dissolving into Fe nanoparticles than Ni nanoparticles. This diffusion process reduces the particle melting point to below the melting point of pure iron (1538°C), reshaping the particle [43]. The diffusion of atomic carbon continues until saturation is reached and the formation of Fe3C begins. The high concentration leads the carbon to dissolve out as graphene layers, forming CNFs [12]. Therefore, the good performance of Ni-UGSO for carbon diffusion and formation of carbon layers is attributed to the synergistic effect of the Fe-Ni alloy formed from the Fe present in UGSO and the Ni added to the support.CNMs and hydrogen were produced from waste plastic in a two-stage pyrolysis-catalysis reactor operating in continuous mode. The catalyst is made from negative-value mining residues containing a significant amount of Fe. The main conclusions can be listed as follows: • Ni-UGSO demonstrated excellent catalytic performance in the synthesis of CNMs and production of H2 with yields of 56.6 and 6.6 g /100 gplastic, respectively. • In the same conditions and catalytic temperature (650°C), Ni-UGSO showed far better performance than Fe/Al2O3. • The CNMs produced were mostly tubular CNFs with different diameters and irregular shapes. • The type of plastic affected the quantity and quality of the produced CNFs. There was no significant difference in the quantity of CNMs and H2 produced from virgin and used HDPE. • When mixed plastics were used as a feedstock, the yield of CNMs decreased by ⁓10 wt%, and amorphous carbon was produced. This was attributed to the presence of contaminants and non-polyolefenic plastics. Ni-UGSO demonstrated excellent catalytic performance in the synthesis of CNMs and production of H2 with yields of 56.6 and 6.6 g /100 gplastic, respectively.In the same conditions and catalytic temperature (650°C), Ni-UGSO showed far better performance than Fe/Al2O3.The CNMs produced were mostly tubular CNFs with different diameters and irregular shapes.The type of plastic affected the quantity and quality of the produced CNFs. There was no significant difference in the quantity of CNMs and H2 produced from virgin and used HDPE.When mixed plastics were used as a feedstock, the yield of CNMs decreased by ⁓10 wt%, and amorphous carbon was produced. This was attributed to the presence of contaminants and non-polyolefenic plastics.In light of these results, it is concluded that Ni-UGSO, which is made from waste, can be used to treat waste plastic at low temperatures for the mass production of CNMs and H2. Some of the future prospects can be listed as follow: • Study the effect of CO2 and vapor on the quality of CNFs. • Study the effect of pyrolysis and catalysis temperatures on the yield and quality of CNFs. • Apply the produced CNFs in composite materials and catalysis. Study the effect of CO2 and vapor on the quality of CNFs.Study the effect of pyrolysis and catalysis temperatures on the yield and quality of CNFs.Apply the produced CNFs in composite materials and catalysis.This work was supported by the National Science & Engineering Research Council of Canada (NSERC) and KWI Polymers.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to thank M. Nicolas Brodusch from McGill University for the STEM micrographs and the University de Sherbrooke's Platforme de Recherche et d'Analyse des Matériaux (PRAM). We thank Marc Couture, Karen Bechwaty, Mohamed Hossam Eldakamawy, Sabrina Bahia Karakache, and Marc-Alexandre Fortin for their technical support.Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.ceja.2022.100424. Image, application 1

Pyrolysis and in-line catalytic decomposition of plastic waste were performed for the production of carbon nanomaterials (CNMs) and hydrogen. A new catalyst, Ni-UGSO (Nickel-UpGraded Slug Oxide), was used in a continuous mode, and its activity was compared to that of a typical Fe/Al2O3 catalytic formulation. Moreover, the pyrolysis-catalysis of different plastics (virgin high-density polyethylene (HDPE), used HDPE, and mixed plastics) was studied to investigate the effect of the plastic type on the quantity and quality of the produced CNMs. Ni-UGSO exhibited the highest catalytic activity for the production of CNMs and H2 of the two formulations tested, with yields of 56.6 kg/100 kgplastic and 13.2 kmol/100 kgplastic, respectively. The high activity and performance of Ni-UGSO were attributed to the synergistic effect of the Ni and Fe in Ni-UGSO. Scanning transmission electron microscopy (STEM) results revealed that most of the produced carbon was in the form of carbon nanofilaments (CNFs) of different diameters, ranging from 8 to 90 nm. The use of mixed plastics as a feedstock decreased the yield of CNMs by 10 wt%, and a layer of amorphous carbon covered the CNFs. This layer is due to the presence of polystyrene (PS), polyethylene terephthalate (PET), and other contaminants in the feedstock. Raman spectroscopy showed that the CNFs produced from used HDPE had the highest intensity ratio G/D (1.13).

Data will be made available on request.The extensive utilization of fossil fuels for daily life and industrial applications has led to significant emissions of greenhouse gases (GHGs) including carbon dioxide (CO2) and methane (CH4) [1]. CO2 or dry reforming of methane (DRM), which converts both GHGs into syngas (H2 and CO), is an attractive process to produce sustainable fuels with low carbon emissions [2]. However, the activation and dissociation of the CO bond in CO2 and the C-H bond in CH4 both require high energy input (750 kJ mol−1 and 439.5 kJ mol−1, respectively) [3,4]. DRM can be driven by sustainable solar energy, and efforts have been spent on researching on solar-driven thermochemical process in which solar energy is used to reach the required high temperatures [5]. On the other hand, solar-driven photothermochemical DRM (PTC-DRM) is an emerging approach that can incorporate photocatalysis into thermochemical DRM [6–9] . Compared to conventional solar thermochemical DRM process, the PTC-DRM process has a couple of additional unique aspects in utilizing light irradiation when a specially designed catalyst is applied: (1) light-driven thermocatalysis due to surface plasmon resonance effect when a plasmonic metal catalyst such as nickel is used, and (2) photo-excited electron-hole pairs from a semiconductor support that actively participate in the redox reaction [1,6,10]. Thus, by combining both thermocatalysis and photocatalysis in one reaction system, PTC-DRM activities are largely enhanced compared with traditional thermochemical DRM [1,7,9,11–13].Metal supported on metal oxides has been one of the most widely researched catalyst structures for PTC-DRM process [14–29]. Ni is an attractive metal candidate in terms of low cost, abundance, and high PTC-DRM activities [21–29]. However, high-temperature DRM environment is prone to cause Ni sintering and carbon formation, leading to deactivation of Ni-based catalysts [30]. As regard to photocatalytic DRM, which was conducted under ultraviolent or visible light irradiation at low temperatures. For example, with ultraviolet light irradiation of 150 mW/cm2 at 100 °C, production rates of CO and H2 can reach 750 μmol g−1 h−1 and 1126 μmol g−1 h−1, respectively on Ni-montmorillonite/TiO2 [31]. With visible light irradiation of 790 mW/cm2 at 400 °C, 2Ni/CeO2−x yielded CH4 and CO2 rates of 0.21 and 0.75 mmol (gcat • min)−1, respectively [32]. However, these works achieved low reaction rates, far from industrial application requirements, thus exploring photocatalysts for more efficient photothermal reaction under high temperature and full-spectrum solar irradiation is very necessary.On the other hand, adjustable bulk and surface components of metal oxides, such as perovskite oxides of general formula ABO3, is a promising catalyst candidate [33]. In the perovskite oxide framework, the A-site is generally a rare earth or alkaline earth metal, such as La [34–37] and Sr [38], while transient metal, such as Ni [34,35,37] and Co [39] can occupy B-site. Compared to a metal oxide-supported metal catalyst system, perovskite oxides have the advantages of uniform dispersion of active metal sites and highly tunable oxygen vacancies concentration, thus can counter deactivation [34,35]. Another beneficial advantage of the perovskite oxide catalyst is its considerable photocatalytic activity due to a desirable narrow band gap (e.g., LaNiO3: ∼2.2 eV) [40,41]. This makes perovskite oxides also attractive in the field of photocatalytic organic compounds degradation, water splitting and N2 fixation [42–44]. From existing research, the base perovskite structure, LaNiO3, was reported to decompose completely during the DRM reaction, and Ni-La2O3 alone was unable to resist Ni sintering and carbon deposition, thus resulting in inefficient DRM reaction [36,45]. Partial substitution at the A-site of perovskite presents an effective approach, as this modification may remarkably enhance the catalytic activities by altering the electronic state of B-site cations and/or introducing oxygen vacancies [46], thus both average Ni oxidation states and Ni particle size will be reduced, and carbon deposition can be suppressed. Wang et al. reported that Ce substitution at the A-site of LaNi0.5Fe0.5O3 perovskite introduced more oxygen vacancies and activated B-site cations, thus the DRM activity was enhanced [36]. Valderrama et al. also reported that partial substitution of La by Sr at A-site in LaCoO3 structure improved Co metallic phase dispersion, leading to high DRM activities and coke resistance [47]. Therefore, LaNiO3 perovskite catalyst with partial substitution at A-site can be a promising candidate for enhanced PTC-DRM performance due to its photoactivity and enhanced properties to resist metal sintering and coke deposition.This work aims to conduct a systematic exploration of efficient PTC-DRM activities and mechanism studies over efficient La1−xCexNiO3 catalysts (x = 0.0 – 1.0). The catalyst morphology, surface chemical states of catalysts before and after PTC-DRM reaction, and optical properties of fresh catalysts were characterized to understand the promoting effects. Then, the in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) was performed on the catalysts within the temperature range of 25–600 °C under light and dark conditions to understand the intermediate change relation with the catalyst activities. Finally, the roles of each component of PTC-DRM were discussed to understand the PTC-DRM mechanism.La1−xCexNiO3 catalysts were synthesized by the Pechini method [48], and the corresponding metal nitrates were utilized with appropriate stoichiometry. Specifically, to synthesize La0.9Ce0.1NiO3, 79.9 mg La(NO3)3•6 H2O, 8.9 mg Ce(NO3)3•6 H2O and 59.6 mg Ni(NO3)2•6 H2O along with 78.8 mg citric acid were dissolved in 5 ml water at a metal cations to citric acid ratio of 1:1, denoted as solution A. Another 78.8 mg citric acid was dissolved in 2 ml ethylene glycol and denoted as solution B. Solution B was added dropwise to solution A. The resulting solution was stirred for 15 min at 400 rpm and was then heated to 120 °C to form a viscous gel and finally yielded a solid precursor. This product was then transferred to an oven to be calcinated under air at 750 °C for 5 h to produce the corresponding catalyst samples.Morphology, structure, and composition of the catalysts were characterized by transmission electron microscopy (TEM, FEI Tecnai G2 F20ST), high-angle angular dark-field scanning transmission electron microscopy (Hitachi 2700 C), X-ray diffraction (XRD, BRIKER D8), and X-ray photoelectron spectroscopy (XPS, Omicron), Raman spectroscopy (Horiba Jobin-Yvon LabRam HR, 633 nm laser source). UV–vis diffuse reflectance spectra were collected by a Hitachi U4100 UV–vis–NIR Spectrophotometer with Praying Mantis accessory. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) spectra were collected on a Nicolet 6700 infrared spectrometer (Thermo Electron) equipped with a Praying Mantis DRIFTS accessory and a reaction chamber (Harrick Scientific, HVC-DRP). The maximum allowable operating temperature of the chamber is 600 °C. Because PTC-DRM activities were evaluated after reducing catalysts in H2/Ar mixture at 700 °C for 2 h, TEM, XRD, XPS, UV–vis characterization was conducted on H2-reduced catalysts. H2 temperature-programmed reduction (H2-TPR, Micromeritics, AutoChem II 2920) was performed on 0.15 g fresh catalysts under a 10% H2/90% Ar gas flow of 40 standard cubic centimeters per min (sccm) with a heating rate of 10 °C/min from 200° to 700°C. The thermogravimetric analysis (METTLER TOLEDO, TGA) was performed on 20 mg spent catalysts under an air flow of 40 sccm with a heating rate of 10 °C/min from 25° to 800°C and kept at 800 °C for 3 h.A similar experimental setup applied in the PTC-DRM performance measurements was reported in our previous works [14,16,49], and the reactor configuration is shown in Fig. S1. The concentrated solar irradiation can be operated as high as 1200 W, and the corresponding light intensity was measured to be 3.6 W/cm2 (Fig. S2), which resulted in 420 °C on the catalyst surface. Auxiliary heat was supplied from a tube furnace to reach higher temperatures. A thermocouple was in contact with the center of the catalyst surface and connected to the furnace to provide feedback to the heating program, thus ensuring catalyst surface temperature was the same under light and dark conditions once a designated temperature was set. For PTC-DRM experiments, 5 mg catalyst was dispersed in 5 ml deionized water and sonicated to form a uniform ink. The ink was then dropped onto a piece of Whatman™ Quartz filter paper and placed on the catalyst holder and transferred into the tube reactor. The reactor was first purged with 150 sccm Ar for 30 min to remove impurities under room temperature, followed by reducing the catalyst under a mixed flow of 23 sccm H2 and 28 sccm Ar for 1 h at 700 °C. Then, the reactor was purged with 150 sccm Ar to remove the remaining H2. After that, the reactant gases (10% CO2/10% CH4/80% Ar) were introduced into the reactor with a flow rate of 14 sccm. Only CO and H2 were detected as the products by an on-line gas chromatograph (GC 2010, Shimadzu) equipped with a thermal conductivity detector (TCD) and a methanizer-assisted flame ionization detector (FID). The production rates of CO and H2 (n, mol g−1 h−1) were calculated using the following formula: n = P ∙ V ∙ v i m ∙ R ∙ T ∙ 3600 Where P is the pressure (1.01 ×105 pa), V is the gas volumetric flow rate (2.3 × 10−7 m3 s−1), v i is the volume concentration of each gas, which is converted from GC measurements and calibration, T is the temperature (298.15 K), R is the gas constant (8.314 J mol−1 K−1), and m is the loaded catalyst weight (5 × 10−3 g).The CO2 and CH4 conversion % were calculated using the following formula: Conversion % = [ X ] in − [ X ] out [ X ] in × 100 Where [ X ] in is the concentration of each original reactant gas (CO2, CH4), and [ X ] out is the measured concentration of each gas at the outlet. In situ DRIFTS spectra were recorded on a Nicolet 6700 spectrometer (Thermo Electron) equipped with a liquid nitrogen cooled HgCdTe (MCT) detector, a Praying Mantis DRIFTS accessory and a reaction chamber (Harrick Scientific, HVC-DRP) [16]. The reaction cell was equipped with a sample cup to load powder samples and a heater and temperature controller to control the reaction temperature. The maximum operation temperature of the reaction chamber is 600 °C. The dome of the DRIFTS cell has two ZnSe windows allowing IR transmission and a third (quartz) window allowing transmission of light irradiation. Light was introduced into the DRIFTS cell through an optical fiber connected to the solar simulator operated at 1200 W. The intensity of the light measured at the outlet optical fiber was close to 0.1 W/cm2. After loading 10 mg of catalyst sample on the sample cup, the sample was first reduced with a gas mixture of 23 sccm H2 and 28 sccm Ar for 10 min at 600 °C, and no spectra change was observed, meaning the complete reduction of the catalyst occurred. After the reduction process, the reaction chamber was cooled down to room temperature and simultaneously purged by Ar. The reaction temperature was then set to targeted temperatures (25–600 °C) with either light or dark condition, and at the same time, DRM gases were introduced. The DRIFTS data were then taken continuously until no spectra change was observed (10 min). The final DRIFTS spectra were collected and presented.Reducibility of the catalyst determines the active form of the catalyst [46], thus as-prepared fully oxidized catalysts were examined by H2 temperature-programmed reduction (H2-TPR) analysis ( Fig. 1). For LaNiO3, the reduction was observed to happen in 3 steps, as the peaks appeared at 262 and 382 °C along with a broad peak ranging from 457° to 613 °C. The peaks can be associated with different intermediary species of Ni, and La2O3, as the possible reduction steps of perovskite structures are as follows [50–52]: (1) 2LaNiO3 + H2 → La2Ni2O5 + H2O (2) La2Ni2O5 + 2 H2 → La2O3 + 2Ni + 2 H2O In general, the lowest-temperature peak is associated with reduction of Ni3+ to Ni2+, and the peaks at higher temperatures are due to the reduction of Ni2+ into Ni0+ and partial reduction of Ce4+ or La3+. The reduction peaks of La0.9Ce0.1NiO3 shifted to lower temperatures, namely, 357 °C and 486 °C, corresponding very well with previous literature findings [50,53]. This result showed that the partial Ce substitution promoted the reaction between H2 molecules and NiO species to occur at lower temperatures. It is likely the partial incorporation of Ce in perovskite lattice resulted in the distortion of the perovskite structure [54], thus making the reduction of perovskite easier. Additionally, the overall H2 consumption was measured to increase from 2327 μmol/g on LaNiO3 to 3310 μmol/g on La0.9Ce0.1NiO3, thus indicating surface oxygen species and bulk lattice oxygen amounts being higher on La0.9Ce0.1NiO3, which demonstrated an enhanced oxygen mobility upon Ce substitution [55]. On the other hand, La0.5Ce0.5NiO3 and CeNiO3 have only one reduction peak, which is likely due to the reduction of NiO and partial reduction of ceria-based oxides [56]. By calculating the total oxygen storage capacity (OSC) of each catalyst (Table S1), it was found the existence of Ce increased the total OSC on all La1−xCexNiO3 (x = 0.1, 0.5, 1.0) catalysts compared with LaNiO3. It was also observed that at 700 °C, both catalysts have been completely reduced. Therefore, 700 °C was chosen as the reduction temperature to fully reduce Ni.X-ray diffraction (XRD) patterns of reduced catalysts were then characterized and presented in Fig. 2. For all catalysts, the absence of NiO (JCPDS 89–3080) and presence of Ni (JCPDS 04–0850) indicated that Ni element has been fully reduced, thus can act as the active metallic sites for DRM [9], which agrees with the H2-TPR profiles. In addition, for La1−xCexNiO3 with x = 0.0, 0.1, 0.5, La2O3 (JCPDS 71–5408) peaks are clearly resolved. CeO2 (JCPDS 81–9325) was weakly observed on La0.9Ce0.1NiO3, which is likely due to the relatively low concentration and uniform distribution of CeO2. Peaks at 2θ of 30.2°, 39.7°, and 47.4°, characteristic of LaNiO3 structure (JCPDS 33–0711) [57,58], have been only identified on x = 0.0 and 0.1 samples. Wang et al. proposed the “self-regeneration” effect that Ce cations (Ce3+/Ce4+) will reversibly shuttle between CeO2 and perovskite structure depending on the local redox fluctuations [36]. The absence of perovskite structure on La0.5Ce0.5NiO3 is likely due to its transition to CeO2. For CeNiO3 sample, only Ni and CeO2 phases are identified. These results revealed the existence of perovskite structure only on LaNiO3 and La0.9Ce0.1NiO3.Transmission electron microscopy (TEM) and energy dispersive X-ray spectrometry (EDS) tests were then conducted on the reduced La1−xCexNiO3 samples to investigate the morphology and elemental compositions ( Fig. 3). The EDS elemental mapping evidenced the presence of each element (actual metal atomic fraction listed in Table S2) and demonstrated the uniform distribution of La and Ce elements. It is widely accepted that small particle size highly benefits the coking resistance and light absorption properties of Ni-based catalysts for the DRM process [59]. While most of the particle sizes ranged from 5 nm to 40 nm, with higher Ce substitution, the average Ni particle size gradually increased. Specifically, the average particle sizes increased from 12.3 ± 5.7 nm on LaNiO3 to 22.3 ± 8.0 nm on CeNiO3. In addition, La0.5Ce0.5NiO3 showed clearly separate and large metallic Ni particles. The appearance of the Ni particles is likely due to the weak interaction with the La2O3 and CeO2 with an excess amount of Ce substitution. Moon et al. also reported on La0.5Ce0.5NiO3, segregated phases of singular NiO, CeO2, and La2O3 were observed, and poor activities of steam CO2 reforming of CH4 were received due to the weak metal-support interaction [50]. These results confirmed that Ni distribution is optimal on La0.9Ce0.1NiO3 among these samples.Perovskite structure materials are reported to desorb part of the lattice oxygen at high temperature in the reducing environment; concurrently oxygen vacancies (VO) will be formed and partial valence change of the B-site ions will happen [60]. The oxygen vacancies have been generally believed to benefit DRM performance in both CO2 adsorption and coke mitigation [12,61]. The dissociation of C-O of CO2 can happen on the VO then produce CO and O, in which O becomes mobile and can thus participate in the removal of deposited coke by oxidizing C into CO [62]. Therefore, X-ray photoelectron spectroscopy (XPS) analyses were conducted on reduced La0.9Ce0.1NiO3 and LaNiO3 catalysts to investigate the surface elemental compositions. XPS analysis of La (830 ∼ 860 eV), Ni (850 ∼ 880 eV) and Ce (880 ∼ 925 eV) were not discussed as the peaks of these three elements overlap, making it impractical to reach meaningful conclusions. O 1 s deconvolution was then performed in Fig. 4, and two types of oxygen species were located, lattice oxygen (OL) at ∼530 eV, and chemisorbed oxygen (OA) related to the presence of oxygen vacancies (VO) at ∼532 eV [63]. The OA concentration was calculated as the peak area ratio of OA, and the values on LaNiO3 and La0.9Ce0.1NiO3 were 42.8% and 82.0%, respectively. Therefore, the partial Ce substitution clearly introduces more VO on the reduced catalyst.The optical properties were then characterized by UV–vis absorption spectra ( Fig. 5). All catalysts showed strong UV light absorption abilities, and the band gaps of LaNiO3, La0.5Ce0.5NiO3, CeNiO3 were calculated to be 2.19 eV, 2.94 eV, and 2.76 eV, respectively, demonstrating their semiconductor properties. However, according to the Tauc plot of La0.9Ce0.1NiO3, it is not plausible to determine the band gap value. LaNiO3 and CeNiO3 showed very similar light absorption across the wavelength range from 200 to 800 nm and a characteristic adsorption peak centered at around 280 nm was identified, which is similar to reported absorption curves [64,65]. La0.9Ce0.1NiO3 expressed strongest light absorption ability, especially across visible light wavelength range. The reason is likely due to the small and well-distributed Ni particles (indicated by TEM and EDS) since black Ni particles have a dominating light absorption ability and hinder the light transmission to other components [22].The PTC-DRM activities on La1−xCexNiO3 (with x = 0.0–1.0) were evaluated at 700 °C under illumination by a 1200 W concentrated solar simulator and were compared at the same temperature under dark conditions. The results are presented in Fig. 6 and Fig. S3. All catalysts showed DRM performance enhancement under light conditions compared with those under dark conditions, demonstrating the photoactive nature of perovskite catalysts. It was found that with a small amount of Ce substitution (0.1, 0.3), the PTC-DRM activities were largely enhanced. By comparing LaNiO3 and La0.9Ce0.1NiO3, the average CO and H2 production rates increased from 363 mmol g−1 h−1 and 258 mmol g−1 h−1 to 550 and 545 mmol g−1 h−1 in the dark with Ce substitution, while under light illumination, the average CO and H2 production rates even increased from 468 mmol g−1 h−1 and 395 mmol g−1 h−1 to 616 mmol g−1 h−1 and 620 mmol g−1 h−1. However, when Ce concentration is rich (x = 0.5 and x = 1.0), the performance of the catalyst did not further increase. Wang et al. [36] reported the similar catalyst performance as the function of the Ce substitution degree on La1−xCexNi0.5Fe0.5O3 and suggested that Ni phase provides the primary catalytic activity, and the (LaCe)(NiFe)O3 perovskite further enhances the catalytic activity, while CeO2 is not responsible for the activity enhancement.It was also observed that CO production rates are more stable than H2 production rates during the reaction time. Specifically, on La0.9Ce0.1NiO3 under light condition, the 10th-h CO production rate was only 2.23% less than initial value, while H2 production rate showed a 15.7% decrease. The CH4 conversion rate also decreased relatively faster than CO2 (Fig. S4). These results may be attributed to the consumption of the Ni active sites for CH4 conversion to produce H2 and the sufficient concentration of Vo active sites for CO2 conversion to produce CO [12].Ultimately, La0.9Ce0.1NiO3 exhibited good catalytic DRM stability in both light and dark conditions. Furthermore, it was found that La0.9Ce0.1NiO3 produced an average H2/CO ratio of 1.01 under light conditions, which is optimal in achieving near-unity syngas production. Overall, these results present the promotional effects of Ce substitution and light illumination to improve both activity and stability for DRM.In our previous studies, we have conducted PTC-DRM on both Pt-based and Ni-based catalysts [7,8,12,49]. Comparing the Pt-based and Ni-based catalysts, the required loading of Ni is generally larger than Pt to achieve optimal DRM activity. However, under photo-illumination when electron-hole pairs are generated, the too large an amount of metal may result in a stronger charge recombination, leading to a lower photocatalytic contribution [22,66,67]. We further compared this work with the state-of-the-art results in the literature, including both PTC-DRM and DRM. In literature, different light sources and formats of results were reported (e.g., H2 and CO production rates, CO2 and CH4 consumption rates, average CO2 and CH4 conversion percentage, etc.), thus making it difficult to compare the performance directly. However, from Table 1, it is still clear to see the La0.9Ce0.1NiO3 catalyst ranks among the top ones in terms of average CO2 and CH4 conversion percentages. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) analysis was conducted to investigate the intermediates’ change under concentrated solar irradiation or in the dark conditions in the DRM reactant gas atmosphere at temperatures ranging from room temperature (25 °C) to maximum operating temperature (600 °C) (Fig. S5). Upon exposure to the CO2 and CH4, several peaks were observed. According to previous reports, the peaks centered at 3016 cm−1, 1338 cm−1, and 1308 cm−1 were identified as gaseous CH4 [74], and the broad peak centered at 2313 cm−1 was identified as gaseous CO2 [75]. The peak intensities of gaseous CH4 and CO2 were observed to decrease as temperature increased, indicating the endothermic characteristic of DRM. The adsorption bands located at 2178 cm−1 and 2111 cm−1 are assigned to gaseous CO [76,77], which appeared only after 400 °C in the dark, but was observed starting from 200 °C under light, indicating that light irradiation can activate the CO production at lower temperature.To clearly observe the intermediates’ change, the in situ DRIFTS spectra at the wavenumber ranging from 1200 to 2000 cm−1 are presented in Fig. 7. It was observed that the solar irradiation led to an intensity increase in formate, m-CO3 2-, and b-CO3 2- bands, which are all active intermediates in CO2 reduction reactions [78,79]. The stronger peak intensities under light are likely due to the generation of VO under light that promoted CO2 adsorption and formation of these peaks [8,49]. This aligns with the promoted reaction rates in the DRM process under light irradiation.It was also noticed that under light condition, the peak at 1753 cm−1, which is attributed to La2O2CO3 species [80], only appeared under light irradiation when temperature was between 100 and 600 °C. As previously reported, two parallel routes for CO2 activation may be occurring on the La-based catalysts: (1) Through direct decomposition on oxygen vacancies and (2) Through the formation and decomposition of La2O2CO3 [60,81]. Thus, La2O2CO3 has been widely recognized to be an active intermediate formed by the reaction between La2O3 and CO2, further reacting with carbonaceous intermediates on Ni at the metal-support interface to produce CO and regenerate the La2O3. The generation of La2O2CO3 under light irradiation aligns well with the higher DRM activities. Similarly, Akula et al. also observed the formation of La2O2CO3 on La2O3/TiO2 during photocatalytic water splitting reaction, which can highly improve the photo-induced electro-hole pairs separation and speed up the photocatalytic methanol decomposition [82]. Gao et al. also reported the presence of La2O2CO3 during the PTC-DRM process, which can facilitate the coke mitigation on metal surface [83].All the carbonate peaks intensities are also weaker on LaNiO3, and La2O2CO3 was not observed on the Ce free sample (Fig. S6). This result indicated that partial Ce substitution can benefit the adsorption of CO2 and formation of La2O2CO3 upon light irradiation. It is likely that the rich VO density from the partial Ce substitution boosted the generation of intermediates.Characterization of spent catalysts was conducted to reveal information about catalyst stability, as both Ni sintering and coke formation are generally believed to be responsible for deterioration of DRM performance [59]. TEM measurements were first carried out on spent LaNiO3 and La0.9Ce0.1NiO3 after 10 h DRM reaction at 700 °C under the dark and light conditions ( Fig. 8). By comparing these TEM images with Fig. 3, Ni sintering can be observed on the spent LaNiO3 after DRM reaction at 700 °C under both dark and light conditions (large Ni particles are circled in Fig. 8), while no obvious Ni sintering was observed on the spent La0.9Ce0.1NiO3. To quantify the extent of Ni sintering, we analyzed the Ni particle size distribution over 100 particles for each spent sample and presented in Fig. S7 . Ni with much larger particle sizes were clearly observed on spent LaNiO3 (average value of 32.0 ± 14.8 nm and 37.6 ± 17.1 nm for light and dark, respectively), while La0.9Ce0.1NiO3 showed much smaller values (average value of 23.2 ± 6.3 nm and 25.0 ± 8.4 nm for light and dark, respectively), which matched the PTC-DRM performance difference presented in Fig. 6. These results suggest that the presence of Ce benefited Ni stabilization, and mitigated Ni sintering, resulting in better PTC-DRM performance on La0.9Ce0.1NiO3. As widely reported, Ce-containing Ni-based catalysts showed strong metal support interactions that prevented Ni from sintering and deactivating [84–86]. However, no obvious differences were observed on both catalysts between light and dark conditions, meaning the Ni sintering is mainly the result of thermal stability and independent of light conditions.Large amounts of carbon filaments (e.g., CNTs) were clearly observed on both samples under dark conditions, yet under light conditions, the coke species were mainly active carbon. We conducted further analyses of additional TEM images of spent LaNiO3 and La0.9Ce0.1NiO3 after DRM reaction at 700 °C under the dark condition (Fig. S8). It was observed that on La0.9Ce0.1NiO3, the majority of Ni particles were still closely attached to the supports, whereas, on LaNiO3, many Ni particles were detached from the support and likely encapsulated in carbon (as circled yellow in Fig. S8a). This indicates severer Ni deactivation on LaNiO3 and agrees with its lower DRM performance compared to La0.9Ce0.1NiO3. Liu et al. also observed that light irradiation can tune the carbon deposition behavior of the Ni-based catalysts during PTC-DRM process [25].Thermogravimetric mass spectrometric (TGA) analysis and Raman spectroscopy was then performed to determine the deposited coke amount on spent catalysts after DRM reactions under the light and dark conditions (Fig. S9, S10). Specifically, a weight loss of 47.0% and 51.6% was observed on LaNiO3 under light and in the dark condition, respectively. While on La0.9Ce0.1NiO3, the values reduced to 20.2% and 27.6%. The difference in coke formation amounts on the two catalysts agrees with the visual observations on TEM images. With regard to Raman analysis, two peaks, with D band at ∼1330 cm−1 and G band at ∼1580 cm−1 were observed on all catalysts, which are assigned to amorphous carbon and graphitic carbon, respectively, where carbon filaments are usually composed of graphitic carbon [25,87]. We calculated the intensity ratios of D- and G-band (ID/IG) on spent catalysts after DRM reaction at 700 °C under dark and light conditions (Table S3). Clearly, the ID/IG ratio is larger under light than that in the dark, and the ID/IG ratio of spent La0.9Ce0.1NiO3 is larger than that of spent LaNiO3. The ratio of ID/IG is highest on spent La0.9Ce0.1NiO3 under light, while it has the highest DRM performance. The positive correlation of the fraction of amorphous carbon with DRM performance agrees with the literature that amorphous carbon is more reactive and easier to be gasified so that Ni can continuously serve as the active sites for DRM reaction [88,89].The effect of Ni NPs size on the nucleation and growth of coke in DRM is well reported in the literature [60,90,91]. In our case, the average Ni NP size on fresh La0.9Ce0.1NiO3 and LaNiO3 (16.5 ± 7.3 nm and 12.3 ± 5.7 nm, respectively) are similar, but TEM images indicated the severe agglomeration of Ni NPs on spent LaNiO3. Thus, the partial substitution of Ce can prevent the Ni from aggregation, reducing the coke formation, and thus improving the PTC-DRM activities and stability.Then, XRD analysis was performed on the spent La0.9Ce0.1NiO3 and LaNO3 and compared with the reduced samples to observe the structure change (Fig. S11). It is clearly observed that on La0.9Ce0.1NiO3, the perovskite LaNiO3 structure was still present after the DRM reactions, while on LaNiO3, the structure was destroyed and became Ni and La2O3. Similarly, Das et al. reported that the perovskite structure was maintained in the DRM atmosphere by partially substituting Ni with Fe on La0.9Sr0.1NiO3 [60]. Wang et al. also reported that the stability of perovskite (LaCe)(NiFe)O3 structure are credited to the preserved perovskite structure during the DRM reaction environment [36].Furthermore, we conducted XPS analysis on spent La0.9Ce0.1NiO3 after DRM reaction at 700 °C under the dark and light conditions and presented the deconvolution of O 1 S XPS spectra in Fig. S12. Similar to that observed in Fig. 4, two types of oxygen species, lattice oxygen (OL) at ∼530 eV, and chemisorbed oxygen (OA) related to the presence of oxygen vacancies (VO) at ∼532 eV, were identified. By calculating the peak area ratio, the OA concentration value was determined to be 58.6% and 73.3% under the dark and light conditions, respectively, both are lower than the value of fresh La0.9Ce0.1NiO3 (82.0%), indicating the consumption of VO during reaction process. More importantly, the higher VO concentration under light can possibly enhance CO2 adsorption and the adsorbed O on VO can oxidize deposited coke, thus improved PTC-DRM activity and stability were achieved under light condition. Several other publications also reported that light illumination can induce the generation of VO on CeO2 [12,49,92], SrTiO3 [93], or TiO2 [94], because photogenerated electrons may weaken and break metal-O bonds.Multiple compositions are responsible for the enhanced PTC-DRM activities on La0.9Ce0.1NiO3. The effects of Ni metallic phase, CeO2, and perovskite structure LaNiO3 are thus discussed.The Ni metallic phase is undoubtedly the primary catalytic site for PTC-DRM reaction, and previous study also confirmed that performance on Ni-free supports was extremely poor [6,7]. Our TEM analysis indicated that Ni NPs were uniformly distributed on La0.9Ce0.1NiO3, and the aggregation of Ni NPs was mitigated during the reaction. It is likely that the Ce promoter reduced the chemical interaction between Ni and support, leading to increased reducibility, evidenced by H2-TPR results (Fig. 1), thus better dispersion of Ni [95]. The tiny Ni NPs enhanced the CH4 conversion and the interaction of Ni-ceria improved CHx oxidization, avoiding complete decomposition of CHx to carbon, thus less carbon was observed on La0.9Ce0.1NiO3 [96,97]. Ye et al. also reported that Ni LSPR property enhances PTC-DRM activities with smaller particle size, while large Ni particles can lead to weak optical property [22]. The light absorption on La0.9Ce0.1NiO3 extended to light of visible and near infra-red region, therefore it can strongly harvest solar energy and the PTC-DRM activities under light irradiation was significantly boosted.CeO2 was reported to generate electron-hole pairs and oxygen vacancies under light irradiation [7,12]. CO2 adsorption and conversion can happen on oxygen vacancies to produce CO, and CH4 or intermediates (CHx) can be oxidized to produce CO and H2, thus PTC-DRM activities were highly enhanced by CeO2. As indicated from the O 1 s XPS spectra (Fig. 4), oxygen vacancies concentrations are much higher on La0.9Ce0.1NiO3 than LaNiO3, and the light absorption ability is also stronger on La0.9Ce0.1NiO3 (Fig. 5). Therefore, it is likely that the Ce partial substitution retains the perovskite structure in the DRM environment, and the generated oxygen vacancies and electrons can boost CO2 adsorption and activation, thus promoting the generation of active intermediate La2O2CO3 to mitigate carbon formation and boost the DRM activities. However, for Ce-rich catalysts (x = 0.5, 1), although higher values of oxygen storage capacity were obtained, they showed declined activities. Therefore, CeO2 functions as an optimal promoter at a lower concentration (x = 0.1).The perovskite structure can also be the active sites, in which the dominant defects during DRM process are oxygen vacancies, which can act as the active sites for adsorption and activation of the reactants and intermediates [36]. In our study, the observed preservation of perovskite structure on La0.9Ce0.1NiO3 is likely benefited by two factors: (1) replenishment of oxygen vacancies from Ce substitution; (2) uniform distribution of Ni to boost the oxidation process [60,98].For comparison purpose, the La0.9Ce0.1 oxides supports were first synthesized, and same amount of Ni was wet impregnated on the supports to yield Ni/La0.9Ce0.1Ox. By conducting PTC-DRM on La0.9Ce0.1NiO3 and Ni/La0.9Ce0.1NiOx at the same DRM reaction conditions (Fig. S12), it was found that Ni/La0.9Ce0.1NiOx received inefficient and unstable PTC-DRM performance, proving the catalytic properties of perovskite structure. The perovskite catalyst was also widely used in photocatalytic CO2 reduction since it has strong light absorption and can generate electron-hole pairs for CO2 reduction at surface sites [99].Additionally, to verify the potential photocatalysis on La0.9Ce0.1NiO3 in the solar-driven DRM process, a control experiment was conducted at 700 °C with a 495 nm long-pass filter applied. The comparison of 10 h average DRM performance of La0.9Ce0.1NiO3 under full spectrum, 495 nm long-pass filter, and dark conditions was presented in the Table 2. The DRM performance is almost the same under the 495 nm long-pass filter and dark conditions, indicating the existence of photocatalysis that boosted the DRM reaction on La0.9Ce0.1NiO3 under full spectrum irradiation.Based on the above experimental results and discussion, the possible PTC-DRM reaction pathways on La1−xCexNiO3 catalyst are proposed as follows: CH4 dissociation takes place on Ni reaction sites to form C* and H* intermediates (Steps 1–2) [60]. Two H* can couple and form H2 (Step 3), and C* can be oxidized by lattice oxygen (OL) to generate CO (Step 4). On the other hand, CO2 can directly dissociate over oxygen vacancies (VO) or hydrogenate yielding CO, O, carbonate, or formate as intermediates (Steps 5–7), which can further react to form CO (Steps 8–9) [100–102]. CO2 can also be adsorbed on the La2O3 surface and subsequently convert La2O3 into La2O2CO3 intermediate, which can actively react with deposited coke (Steps 10–11) [100,103]. Additionally, under light illumination, high-energy electron (e-) and holes (h+) will be generated on the perovskite catalysts [11,31,69]. The e- can generate VO on the catalyst surface and enhance the CO production, while h+ can boost CH4 dissociation towards higher H2 production (Steps 12–14). (1) CH 4 → CH x * + ( 4 − x ) H * (2) CH → C* + H* (3) 2 H* → H2 (4) C* + OL → CO + VO (5) CO2 + VO → CO + OL (6) CO2 + H* → HCOO* (7) CO2 + OL → CO3 2- (8) HCOO* → CO + OH* (9) CO3 2- + C* → 2CO + OL (10) La2O3 + CO2 → La2O2CO3 (11) C*+ La2O2CO3 → 2CO + La2O3 (12) LaxCe1−xNiO3 + hv → e- + h+ (13) Ce4+ + e- → Ce3+ + VO + O (14) CH4 + xh+ → CH(4−x) + xH+ In summary, promoting effects of partial substitution of Ce into LaNiO3 perovskite catalysts on the PTC-DRM activities were presented. Ce, as a promoter, was found to benefit Ni NPs active sites distribution and perovskite structure retention on La0.9Ce0.1NiO3. Therefore, Ni aggregation can be mitigated during PTC-DRM process due to stronger metal-support interaction. In addition, by conducting in situ DRIFTS analysis and control experiment with 495 nm long-pass filter, the light irradiation was found to enhance CO2 adsorption and formation of active intermediate La2O2CO3, induce photocatalytic activities on La0.9Ce0.1NiO3, assisted by generated oxygen vacancies and electron-hole pairs. These advantages led to carbon mitigation and promoted PTC-DRM activities. As a result, at 700 °C under 30 suns light irradiation, the La0.9Ce0.1NiO3 showed highest PTC-DRM activities with CO and H2 production rates of 616 and 620 mmol g−1 h−1, respectively. This work systematically advances the design of cost-effective catalysts and the study of the light contribution mechanism thus promoting efficient solar-powered conversion of greenhouse gases. Zichen Du: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization, Writing – review & editing. Cullen Petru: Investigation, Methodology, Writing – review & editing. Xiaokun Yang: Validation, Resources, Writing – review & editing. Fan Chen : Validation, Resources. Siyuan Fang: Validation, Resources. Fuping Pan: Validation, Writing – review & editing. Yang Gang: Validation, Writing – review & editing. Hong-Cai Zhou: Resources, Writing – review & editing. Yun Hang Hu: Resources, Writing – review & editing. Ying Li: Conceptualization, Writing – review & editing, Supervision, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by U.S. National Science Foundation (Grant No. 1924466). The use of Materials Characterization Facility (MCF) at Texas A&M University is acknowledged.Reactor configuration, irradiation spectrum of concentrated solar, additional PTC-DRM activities comparison, in situ DRIFTS spectra, TGA, XRD, Raman characterization.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jcou.2022.102317. Figure S1 Supplementary material .

Solar-driven photothermochemical dry reforming of methane (PTC-DRM) is a promising technique to produce syngas using greenhouse gases (CO2 and CH4). In this work, Ce-substituted LaNiO3, i.e., La1−xCexNiO3 perovskite catalysts were synthesized for PTC-DRM reaction under concentrated sunlight. At 700 °C under 30 suns light irradiation, the CO and H2 production rates were at 616 and 620 mmol g−1 h−1, respectively, over the La0.9Ce0.1NiO3 catalyst, notably higher than those obtained in dark at the same reaction temperature and higher than those over LaNiO3 under the same light irradiation condition. The CO2 and CH4 conversion by the La0.9Ce0.1NiO3 catalyst are among the top-performing catalysts reported in the literature. The Ce substitution of La at a small fraction (x = 0.1) was found to benefit Ni active sites distribution and retention of the perovskite structure, which led to mitigation of both Ni sintering and carbon formation, thus promoting light absorption and PTC-DRM activities. A higher fraction of Ce substitution (x ≥ 0.5), however, did not show any beneficial effects. By conducting in situ DRIFTS at PTC-DRM reaction conditions and control experiment using 495 nm long-pass filter, light irradiation was found to induce photocatalytic activities on La0.9Ce0.1NiO3 and enhance CO2 adsorption and formation of active lanthanum oxycarbonates intermediates (La2O2CO3), possibly due to the generation of oxygen vacancies and electron-hole pairs. This work reports a new catalyst design and mechanistic studies for PTC-DRM reaction, and the findings are of importance for the application low-carbon fuel generation from sunlight.

No data was used for the research described in the article.Over the last decades, the increasing incidence of Diabetes Mellitus in both developing and affluent countries [1] has further highlighted the need for new approaches to glucose sensing. Currently, most of the commercially available products are still based on the same principle behind the enzymatic biosensor proposed by Clark in 1962 [2,3]. Although enzymatic electrochemical sensors have the potential to be produced at a low cost and be compact, they suffer from a limited functional period which complicates their use for continuous monitoring applications [4]. The poor stability of the signal can chiefly be attributed to fouling by non-target species, and in particular to the use of the enzyme, whose activity progressively reduces over time [5]. Additionally, the enzyme's sensitivity to temperature, pH and humidity changes and the high risk of enzyme leaching increases the complexity of the manufacturing process and imposes more delicate conditions of usage [6]. To overcome these drawbacks, mechanisms of non-enzymatic electrochemical glucose sensing have been studied [5,7,8]. In this regard, nanomaterials play a major part [9]. Different nanostructured electrode materials have been developed over the years and in particular noble metals (Au, Pt, Pd etc.) [10] played a significant role in the advancement towards fourth-generation glucose sensors.Considerable effort has been dedicated in understanding the mechanism of direct electrooxidation at metal electrodes. Accordingly, two models have been proposed: the chemisorption model [11] and the incipient hydrous oxide adatom mediator model [12].Despite the extensive studies on the oxidation process, the use of bulk noble metals for glucose sensing is limited, due to the presence of many shortcomings [13,14]. First, the competition of anions such as halides for glucose adsorption, leading to catalyst poisoning [15,16]. Secondly, the electrode's degradation as a result of fouling by species commonly found in physiological solutions. Lastly, the much greater cost of noble metals compared to other materials, due to their scarcity and use as assets in the financial system [17].For these reasons, the research community has focused its attention on finding alternative materials capable of electrooxidizing glucose without being as affected by poisoning and fouling, while costing a fraction of the noble metals. Examples are metal oxides (NiO, CuO, Co3O4 etc.) [18–20], metal sulfides (NiS, FeS2, CuS2, etc.) [21–23], metal organic frameworks [24–26] and carbon materials (graphene, carbon nanotubes etc.) [27]. Alloying [28], doping [29] and carbon composites [30] are common approaches being explored to fine tune the catalytic activity and/or poisoning resistance.In particular, in the field of glucose sensing the element nickel has shown exceptional potential, finding application in its pure form, in compounds and also recently in conductive metal organic frameworks. Its widespread use can be mainly attributed to the highly oxidizing power of the Ni(II)/Ni(III) couple [31,32] and to its intrinsic resistance to halogen poisoning (both by itself and in combination with noble metals) [33].Furthermore, nickel is a naturally occurring element that can be readily found in the environment, existing in over a 100 different mineral forms [34]. It belongs to the transition series with other group 10 elements such as Pt and Pd. Sometimes considered as the “impoverished” sibling of Pt, its high activity towards sluggish reactions in homogeneous [35] and heterogenous catalysis [36] proves this characterization to be limited. Additionally, the wider availability of Ni makes it vastly more affordable than common noble metal catalysts (e.g., the price of Ni is presently 2500-fold lower than for Au). As a result, more researchers are turning to nickel-based materials for application as either supports or catalysts due to their outstanding yet highly tunable properties.This review article summarizes the most recent progress in the development of Ni-based non-enzymatic electrochemical glucose sensors. The fundamentals of glucose electrooxidation at Ni electrodes are highlighted, which can in some part be extended to other Ni-based systems. The literature on the chloride poisoning is briefly presented and a possible explanation for its resilience compared to noble metals is proposed. Additionally, the analytical performance of sensors based on Ni compounds, bimetallic nanostructured systems and metal organic frameworks with Ni centers is discussed.The electrochemical oxidation of glucose on Ni electrodes has been reported in numerous studies to occur at the same potential of the Ni (III) oxide formation [37]. Accordingly, the electron transfer appears to be mediated by the Ni ( OH ) 2 /NiO(OH) redox couple. As a consequence, the overall electrocatalytic activity of Ni towards glucose electrooxidation is positively correlated with pH.In alkaline solutions, the following mechanism has been proposed by Fleischmann and others: (1) Ni + 2 OH − → Ni ( OH ) 2 + 2 e − (2) NiO + H 2 O → Ni ( OH ) 2 (3) Ni ( OH ) 2 + OH − → NiO ( OH ) + H 2 O + e − (4) NiO OH + glucose → Ni OH 2 + gluconolactone The Nickel(III) acts as a strong oxidant towards glucose (and other organic compounds [32]) by participating in the rate limiting step of the electrooxidation: the hydrogen abstraction from the C α to the functional group. The irreversible nature of the oxidation reaction is confirmed by the increase in peak current for increasing glucose concentrations limited only to the anodic scan (Fig. 1 -a).To achieve reproducible results, a good strategy is to first perform a few scans of cyclic voltammetry between 0V and 0.7V (vs Ag/AgCl/KCl(3 M) reference electrode), which guarantees the complete conversion of Ni and NiO species to Ni ( OH ) 2 and NiO(OH) [39].For the sake of comparison, many researchers evaluate their Ni-based electrodes’ performance at pH 13. Nevertheless, an optimization study should always be considered in order to evaluate the hydroxyl concentration that leads to the highest peak current and the lowest background current (current response in the absence of the analyte). In a recent work, Ko et al. [38] proved via chronoamperometry that the concentration of NaOH producing the best signal to noise ratio was 0.5 M. Clearly, care should be taken not to generalize their results to other systems having different substrates or additional catalysts.Two different forms exist for Ni ( OH ) 2 ( α and β ) and for NiO(OH) ( β and γ ) [40]. Visscher and Barendrecht [41] as well as Hahn et al. [42] observed that α - Ni ( OH ) 2 slowly transforms in β - Ni ( OH ) 2 as a result of electrochemical ageing. In particular, the transformation from α - Ni ( OH ) 2 to β - Ni ( OH ) 2 is associated with a shift of the anodic peak position to higher potentials. As a result, for a fixed potential the current density for water oxidation will be reduced for aged electrodes. This polymorphism can account for the diverse electrochemical response of different Ni electrodes.One of the main disadvantages of non-enzymatic glucose sensors is the lack of a selective recognition mechanism. Nickel-based catalysts and non-enzymatic glucose sensors as a whole, once polarized can oxidize many other molecules in addition to glucose. This can be attributed to the easily accessible catalytic sites (e.g. NiOOH) which on a flat electrode surface can be reached by small and large molecules alike. As a result, the electrode's current response may poorly correlate with the real serum glucose concentration. In particular, the main interferents are reportedly: ascorbic acid, uric acid, acetaminophen and dopamine [43]. Acetaminophen and dopamine can be especially problematic when their oxidation products have an affinity for the electrode surface (usually carbon-based) because they can effectively block the catalytically active sites [44,45]. Other interferents include mono and disaccharides such as fructose and sucrose, which possess a similar chemical structure to glucose. Disaccharides do not exist in the bloodstream [46], since they get separated into their constituents by enzymes during digestion (e.g. sucrose gets separated by sucrase into glucose and fructose). However, in the case of fructose its blood concentration is usually 10–1000 times lower than that of glucose [47], thus posing no particular issue from an analytical standpoint.Overall, nickel-based non enzymatic glucose sensors can achieve a satisfactory level of selectivity and this can be attributed to a number of reasons. First of all, the concentration of interferents in the blood serum is usually more than 10 fold lower than that of glucose (which for healthy individuals is around 4 mM [48]). Accordingly, even supposing to oxidize the interferents with the same kinetic rate, this would at most cause a 10% uncertainty in the current signal. Second of all, the majority of the proposed electrode configurations in the literature involve some measure of surface nanostructuring [9], which can act as a size control for molecules larger than glucose. Additionally, the high degree of hydroxylation of Ni-based electrodes renders the surface extremely hydrophilic, which naturally repels hydrophobic species such as most fouling agents [49]. Moreover, the electrochemical technique being used also has an impact. Accordingly, good selectivity can be achieved in certain systems with potentiostatic approaches (chronoamperometry) by choosing a potential at which glucose is oxidized, but not the interferents. Lastly, the use of membranes (e.g. Nafion) has been shown to effectively protect the surface from fouling, while maximizing the overall sensor's selectivity mostly through a size selection mechanism [50,51].The chloride ion (Cl−), with a molar mass of 35.45 g/mol, is the most abundant anion in the human serum with a concentration around 97–107 mM [52]. It plays a significant role in the fluid homeostasis, electrolyte balance, conservation of electrical neutrality and acid base status [53].Its significant concentration in the extracellular fluid is well known to impair the performance of non-enzymatic glucose sensors in a process referred to as halide poisoning, where the surface-active sites of the catalyst are being blocked. Noble metals, such as Pt and Au, are reported to be particularly affected by halide poisoning for glucose and methanol electrooxidation [54–57]. The reason for the loss of catalytic activity has been attributed to the specific adsorption of Cl−. Moreover, the presence of chlorides causes the formation of soluble species instead of an oxide layer from Au [16,58], inevitably leading to corrosion. Similarly, electrochemical cycling in chloride rich solutions is known to cause Pt etching [15].Conversely, the surface of metals such as Ni spontaneously forms a passivation film composed of an inner NiO with an hydroxylated outer layer, which protects it against corrosion [59]. In certain conditions, the interaction with halides can still lead to the breakdown of the passivation film, and the subsequent corrosion by pitting.Nonetheless, NiOx is drastically less susceptible to poisoning compared to noble metals and the reason for this has generally been attributed to a difference in adsorption energy of the halide ion on the different surfaces [60,61]. A deep evaluation of the poisoning resistance of NiOx electrodes was first performed by El-Rafaei and colleagues [33], who recorded the glucose electrooxidation peak in 0.5 M NaOH before and after the addition of 0.1 M Cl−. After chloride addition, the peak current decreased to a minimum at the 4th cycle (losing 4% of the initial value), and then it increased again almost to the original value after the 15th cycle. This behavior may be explained by the low absorbability of Cl− ions, and possibly by the high reversible oxidation potentials of the Cl2/Cl− couples [62].A deeper understanding of the mechanisms at play is provided by computational approaches. In a pivotal work, Bouzoubaa et al. [63] described the interaction of an hydroxylated defect-free NiO(111) surface with different halides (F−,I−,Cl−,Br−). Fig. 2 shows the modeled slab for different OH- substitutions by halide ions X−.In particular, as the chloride coverage increases and the Cl− progressively substitutes more OH− groups, the lateral anion-anion repulsions gradually reduce the OH− substitution energy (becoming endothermic at more than 75% substitution). This result is confirmed by previous studies [64]. Moreover, since the ionic radius of Cl− is larger than that of OH−, the NiCl bond length is larger than that of NiO, thus leading to a splitting in the mixed topmost plane. Fig. 3 may be read as follows: at low chloride concentrations (<10−4 M) the NiOH termination is the most energetically favorable, whereas above 10−4 M the 25% OH− substitution is the most stable configuration even at high Cl concentrations, due to the strong anionic repulsion between Cl ions. This calculation may explain why NiOx electrodes are intrinsically resilient to chloride poisoning: approximately three quarters of surface hydroxyl groups are maintained, which during polarization in an alkaline electrolytes can still give rise to the highly oxidizing Ni(III).The small difference in electronegativity between Ni (χ = 1.91) and Se (χ = 2.55) allows the formation of different nickel selenides [65]. A wide range of phases have been reported to be stable at room temperature: NiSe2, NiSe, and Ni3Se2. Based on the prediction of Horn and Goodenough's [66], an increase in the covalency of the metal-oxygen bond significantly influences the binding of oxygen-related intermediate species, which in alkaline solutions are important pathways to glucose electrooxidation.In the field of electrocatalysis of the different nickel selenides, a prominent role is played by Ni3Se2 and NiSe2, which boast a narrow band gap and high conductivity, making them intriguing candidates for non-enzymatic glucose sensors. Moreover, metal selenides possess higher electrical conductivity, compared to the respective oxides and sulfides, due to the strong metallic character of selenium [67,68]. Different selenylation approaches have been proposed, but most are based on techniques such as hydrothermal [69] or solvothermal [70] synthesis, electrodeposition [71], solution chemical process [72] or solid state synthesis [73]. Although the literature on nickel selenides for glucose sensing is still scarce, the current reports are highly promising. The first study on the use of NiSe2 as an electrode modifier for OH− mediated glucose electrooxidation was done by Mani et al. [74]. Here, the researchers employed a hydrothermal synthesis method and then a drop cast of the nanosheets in an alcoholic dispersion on a glassy carbon electrode. The authors proposed the following redox process for the electrode in the absence of glucose in alkaline solution: (5) NiSe 2 + OH − ↔ NiSeOH + Se + e − During the cyclic voltammogram the Ni(III) and Se(II) species are oxidized to NI(IV) and Se(III), which readily oxidize glucose to glucolactone. The enhancement in the electrocatalytic activity may be attributed to the ability of the Se constituent to increase the charge transport efficiency between the Ni center and the electrode substrate. Although reasonable, in the absence of a Ni electrode control such assertions still need to be verified. The calculated sensitivity for the electrode after amperometric calibration in 0.1 M NaOH was limited to 5.6 μA mM−1 cm−2 with a limit of detection (LOD) of 0.023 μM.In order to improve the electron transfer capabilities of nickel selenide-based electrodes, some researchers have combined them with carbon nanostructures. For instance, in a recent work, Xu et al. [75] fabricated a hierarchical electrode composed of carbon nanorods and NiSe2, synthesized though a facile thermal route. After dispersion in a Nafion solution, it was drop cast on a glassy carbon electrode. The reported sensitivity was 3636 μA mM−1 cm−2 with a LOD of 0.38 μM, after amperometric calibration in 0.1 M NaOH. Evidently, the mechanical stabilization of the NiSe2 nanosheets with a conductive membrane greatly improves the overall performance of the sensor. Fig. 4 clarifies the effect of the carbon nanorod incorporation, having a significant and positive influence on the current increment as a result of glucose addition.While electrode fabrication procedures such as drop casting and thin film coating are common in the literature, they are time consuming and require the use of polymeric binders in order to fix the catalyst on the electrode's surface. In these cases, the catalytic active centers can get significantly blocked by the polymer, thus hindering the electron transfer capabilities of the electrode as a whole [76].A more scalable solution, which has been explored not only for Ni selenides, is the synthesis of the active catalyst directly from the corresponding metal (e.g., Ni). In this way, there is no need for a membrane to guarantee the direct catalyst-electrode contact, which, as previously stated, impairs the electrical conductivity.As an example, Ma et al. [77] fabricated through hydrothermal routes Ni3Se2 nanosheets (NS) supported on a Ni foam, with a reported sensitivity of 5962 μA mM−1 cm−2 and a LOD of 0.04 μM. Fig. 5 highlights the improvement in the current density ( Δ j ), going from a bare Ni foam to a Ni3Se2/Ni Foam. The authors attributed the increase in sensitivity to the synergistic interaction between the Ni3Se2 nanosheets and the Ni support. Table 1 lists all the non-enzymatic glucose sensors described in this section.Compounds of Ni with chalcogenides for glucose sensing are not solely limited to selenides. Ni sulfides have been closely investigated as well. The main reported advantages of these materials are their high redox ability, good electrical conductivity and thermomechanical stability [80].Electronic and band structure calculation suggest that as the ratio of S to Ni increases, the Ni d-band centers become more negative and the S p-band centers become more positive [81]. The presence of a band gap between Ni d-band and the S p-band explains why nickel sulfides are generally less conductive than pure Ni. It has been shown that the phase of nickel sulfide has a meaningful effect on the catalytic activity for the hydrogen evolution reaction [82]. As of now, no studies have definitively clarified its impact towards glucose electrooxidation.Nickel sulfides can exist in different crystalline structures and stochiometric ratios, such as Ni3S2, NiS, NiS2, Ni3S4. They find application in different fields, ranging from dye-sensitized solar cell [83] to supercapacitors [84] and electro(photo)catalytic oxygen/hydrogen evolution [28,29]. The current scientific research on nickel sulfides for glucose sensing is mainly focused on exploring the catalyst-support interaction which provides the greatest sensitivity, stability and reproducibility. The first work on an electrodeposited NiS film was performed by Kannan et al. directly on indium tin oxide (ITO) electrodes [85]. The chosen synthesis method was straightforward and easily scalable, granting a sensitivity in 0.1 M NaOH of 7430 μA mM−1 cm−2.In alkaline solution the proposed reversible redox reaction is the following [86]: (6) NiS + OH − ↔ NiSOH + e − Given that the morphology has a strong effect on the catalytic activity, the majority of the published research works report catalyst nanostructuring. Accordingly, hollow spheres of α-NiS have been studied due to their good electrocatalytic activity and stability, ease of synthesis, and environmental compatibility [87]. Interestingly, the authors observed a significant difference between the α-NiS and the β-NiS hollow spheres, with the former leading to a stronger electrocatalytic response. However, the abovementioned sensor had a sensitivity of only 155 μA mM−1 cm−2. This was likely the result of aggregation, which reduced the active sites and the non-exceptional electrical conductivity of NiS. To overcome these obstacles many authors make use of a conductive matrix, such as functionalized carbon black [88]. Relatedly, Ni3S2/carbon composites have also been fabricated, as done by Lin and colleagues [89], by using an hydrothermal method where the carbonaceous matrix consisted of multiwalled carbon nanotubes.With an interesting approach, Meng et al. [90] grew a Ni3S2 nanoworm (NW) network directly on a poly (3,4-ethylenedioxythiophene)-reduced graphene oxide hybrid films (PEDOT-rGO HFs) modified on glassy carbon electrode. A schematic of the fabrication process is illustrated in Fig. 6 . The sensor showed good sensitivity (2123 μA mM−1 cm−2) and low LOD (0.48 μM), that the authors attributed to a combination of high surface area, morphology, hydrophilic nature allowing easy OH− adsorption and good coupling between the different electrode layers.As of now, one of the most promising technological methods is to directly grow the nickel sulfide catalyst directly on a nickel foam. For instance, Huo et al. [91] fabricated a 3D Ni3S2 nanosheet array supported on a Ni foam by hydrothermal synthesis using nickel nitrate and thiourea. After amperometric calibration in 0.5 M NaOH, the measured sensitivity was 6148 μA mM−1 cm−2, with a LOD of 1.2 μM, mainly ascribed to the open channel structure, combined with a fast electron and ion transport. Alternatively, as illustrated in a similar work by Kim et al., Ni3S2 nanostructures can be hydrothermally grown on a Ni-foam by having it react with thioacetamide in an alcohol and water medium [92]. By adjusting the solvent composition, a hierarchical cauliflower-like structure was obtained (Fig. 7 b). The sensor showed superior sensitivity (16 460 μA mM−1 cm−2) and good LOD (0.82 μM) in a 0.5 M NaOH solution. Table 2 describes the detection parameters of non-enzymatic glucose sensors based on nickel sulfides.Nickel nitride (Ni3N) has recently shown to be a promising electrocatalyst for glucose sensing. Ni3N is a low temperature solid state phase at the boundary between the hcp and hcp + fcc zones, where the nitrogen atoms occupy the octahedral interstitial sites of the nickel lattice in a way that minimizes the N–N interactions [94].Calculated Density of States (DOS) studies indicate that bulk Ni3N is intrinsically metallic and that the carrier concentration can be additionally enhanced when dimensional confinement was applied along with nanoscale structure [95].Accordingly, the nitridation process induces a contraction in the d-band near the Fermi level, thus favorably changing the electronic structure for catalytic purposes.Different nitridation techniques are reported in the literature, such ammonolysis in a NH3 atmosphere [96], reactive physical vapor deposition in N2 [97], direct liquid injection chemical vapor deposition with NH3 as a co-reactant [98], nitrogen ion implantation [99], plasma based nitridation [100] or a solvothermal process with highly reactive azide or hydrazine [101,102]. When going from bulk to nanostructured materials, multiple stoichiometries of nickel nitride have been reported (e.g., NiN, Ni2N, Ni4N, Ni8N), due to the fact that phase boundaries can change when the characteristic size is at the nanoscale [103]. The first investigation of Ni3N as a glucose electrocatalyst was done by Xie et al. [104]. The authors synthesized nickel nitride nanosheets on a Ti mesh by ammonolysis of a previously deposited Ni layer. The sensor after amperometric calibration in 0.1 M NaOH displayed a very high sensitivity of 7688 μA mM−1 cm−2 and a LOD of 0.06 μM. However the linear range was only up to 1.5 mM, which limits its potential application since the usual blood glucose concentration is around 4–7 mM [105].As seen with nickel selenides (Section 4.1) and sulfides (Section 4.2), and many other catalytic systems [106–108] the integration of a catalyst, such as Ni3N, in a conductive carbonaceous matrix can lead to significant enhancements in the sensing capabilities due to the well-known improvements in the electron transfer afforded by carbon materials. In this regard, Liu et al. [109] investigated how a change in the structural parameters of different carbon matrices affected the electrocatalytic activity of a Ni3N nanosheet/carbon electrode as a whole. The authors concluded that a hollow/tubular 3D porous architecture leads to the best sensing performance towards glucose, due to a greater dispersion of the nanosheets in the inner and outer walls of the carbon fibers. The calculated sensitivity in 0.1 M NaOH was 1620 μA mM−1 cm−2 until 1.75 mM and 856 μA cm−2 mM−1 (from 1.75 to 9.18 mM), with a LOD of 0.05 μM for the lower concentration range.To further increase the sensing performance of composite Ni3N/carbon electrodes, a current trend is to work on improving the synergy between the carbon matrix and the nickel nitride catalyst. Nitrogen doping of carbon materials allows to fix the metal sites and to facilitate the catalytic process by regulating the electronic structure of the carbon matrix [110]. In a recent work, Chen et al. [111] successfully fabricated a sensor based on nickel nitride decorated nitrogen doped carbon spheres (Ni3N/NCS). Fig. 8 shows the morphological details of the bare and modified N-doped carbon nano spheres, and also the XRD spectrum of the synthesized material. A facile, eco-friendly one pot nitridation process was employed and after amperometric calibration in 0.1 M NaOH the calculated sensitivity was 2024.18 μAmM−1cm−2 (up to 3 mM) and 1256.98 μA mM−1cm−2 (from 3 to 7 mM) with respective LOD of 0.1 μM and 0.35 μM.The sensing parameters of the above-described nickel nitride-based sensors are listed in Table 3 .A commonly employed strategy to maximize the electrode's sensitivity is to introduce nanoparticles in the design, in order to take advantage of their high specific surface area and augment the number of active sites. A fairly recent avenue for glucose sensing is represented by bimetallic alloy nanomaterials, which by definition are comprised of two or more metals. Due to synergistic interactions, bimetallic nanomaterials are considered to be more electroactive than their monometallic counterpart [116,117]. Depending on the metal combination, significant improvements have been reported in terms of sensitivity, stability, biocompatibility and specificity due to biomimetic behavior [118]. The change in reactivity of a metal as a result of alloying can either be due to a change of the electronic structure, increased number of possible bonding geometries for adsorbates, or more indirectly, due to a change in the lattice parameters [119]. In general, a useful, albeit simple, descriptor for a metal's reactivity has been recognized to be the position of d-band center ε d . The higher (lower) the d-band center, the stronger (weaker) the affinity of an adsorbate to the metal site [120–122]. As a molecular adsorbate interacts with the metal's d-band it gives rise to bonding and antibonding molecular orbitals. As a consequence of alloying an upward (downward) shift in the metal d-band is produced, which leads to a decreased (increased) filling of the metal-adsorbate anti-bonding orbitals. It should be noted that there are exceptions to this rule [123], and more refined models that take into account the shape of the d-band have been proposed [124].The main synthetic methods for Ni-based bimetallic nanostructures are co-reduction [125,126], thermal decomposition [127], seed mediated growth [128], galvanic replacement reaction [129] and electrodeposition [130–132].The tendency of Ni to oxidize complicates its synthesis in aqueous solutions [133]. An additional hurdle arises due to the magnetic properties of Ni nanoparticles, causing them to cluster together. For this reason there are few reports on the synthesis of monodispersed size distributions for Ni nanoparticles [134,135].Due to the strong electrocatalytic activity of both copper and nickel in alkaline solution towards glucose electrooxidation [136,137], many researchers have tried to explore how these two metals interact at the nanoscale and how their catalytic activity may change as a result. In a recent study, Wei et al. [138] developed a dendritic Cu@Ni on a Ni foam (NF) electrode for glucose sensing through a facile electrodeposition method. A schematic representation of the system is shown in Fig. 9 .The sensor, after amperometric calibration in 0.1 M NaOH, showed a very high sensitivity of 11340 μAmM−1cm−2 with a LOD of 2 μM. NiO(OH) and CuO(OH) both contribute to a single anodic peak, due to the closeness of their respective oxidation potentials, as also observed in a similar work by Bilal et al. [139]. The strong electrochemical response was attributed to the synergistic interaction the between oxide shell and the conductive metal core, combined with the high surface area allowing easy glucose diffusion. The investigation of Lin et al. [140] on electrodeposited Ni and Cu nanoparticles on multiwalled carbon nanotubes shed light on two important aspects. First, that the multiwalled carbon nanotubes provide a large conductive area onto which Ni and Cu ions can electrodeposit without competing, thus leading a more ordered and active structure. Secondly, there is a ratio of Cu:Ni which gives the highest glucose oxidation current increment, which was observed to be 1:1.Analogously, Ammara et al. explored the combination of a Ni–Cu nanocomposite with carbon nanotubes, by sequential electrodeposition on electrophoretically deposited carbon nanotube film [141]. In this way, the oxidation of Cu and Ni is lessened, while guaranteeing good mechanical stability and improved electrical conductivity though the formation of a percolation path. Here, the researchers also noted that the ideal ratio Ni:Cu to be 1:1.In a recent work, Xu and colleagues [142] fabricated through a one-step hydrothermal synthesis method Ni–Cu bimetallic alloy nanoparticles on reduced graphene oxide. Surprisingly, after running an optimization study on the molar ratio of Ni:Cu, the strongest sensitivity was associated with a 4:1 ratio, in contradiction with the results of Lin et al. [140] and Ammara et al. [141]. In fact, a 1:1 ratio caused a decrease in the sensitivity compared to the bare metal surfaces. The difference in the optimum ratio between the studies is not trivial to explain, but may be due to the different synthetic methods being employed. Possibly, co-electrodeposition might cause the blocking of active sites of Cu at lower molar ratios, compared to hydrothermal methods.Bimetallic nanostructured systems composed of Ni and a noble metal are a promising solution to the current limitations of pure noble metal electrodes for glucose sensing: namely, the sensitivity to chlorides, which impairs their long term stability [16], and their high cost. Moreover, the addition of a second metal can modulate the catalytic activity and facilitate the reactant adsorption and product desorption, as implied above.Simultaneously, integrating a noble metal to a Ni electrode produces two main beneficial effects. First, it allows to extend the sensors’ range of activity to neutral pH. This is because noble metals (such as Pt, Au) are able to directly electrooxidize glucose without the need for a high concentration of solution hydroxyls in the initial rate-limiting step of C1 dehydrogenation [43,143]. Secondly, it causes an increase in the electrical conductivity to the bare Ni electrode. In particular, many authors [144–147], have explored the combination of Au and Ni for electrochemical glucose sensors, noting the presence of synergistic interactions between the two metals.As an example, the group of Yang synthesized spherical Au@Ni nanoparticles with a core-shell structure through a seed-mediated growth in oleylamine [148]. The core shell structure of the nanoparticles can be clearly appreciated in the TEM image shown in Fig. 10 c. The oxidation of glucose on core-shell Au@Ni nanostructures was noted to be analogous to that of a pure Au particle. A similar observation was also done by Guo et al. [149] with a 4 nm electrodeposited Ni(OH)2 layer on nanoporous Au. The fabricated core-shell nanostructures were able to shield the active sites on the particle surface from Cl− and intermediates adsorption. At the same time the formed Ni layer allowed the formation of metal-OH sites, similarly to the Au–OH sites at more negative potentials. In this way it was possible to avoid the oxidation of other interfering substances present in the electrolytic solution.With the use of electrodeposition, Zhou et al. [150] constructed a NiPt nanosheet array on carbon paper and concluded that the Pt:Ni ratio which gave the highest anodic peak in alkaline solution was 1:160. For higher amounts of Pt a decrease in current was observed, likely because the benefits in terms of improved electrical conductivity did not counterbalance the substitution of NiO(OH) centers with the less active PtOH.Other Ni-based bimetallic nanoparticles for glucose sensing include Ag-Ni [151] for which the Ag:Ni ratio giving the strongest activity was observed to be 1:4.In the future, Ni-noble metal systems might play a larger role in the field of continuous non enzymatic glucose monitoring devices. However most of the literature still performs the amperometric calibration in highly alkaline conditions, which is far from those of clinical applications requiring continuous glucose monitoring [105]. Therefore, a calibration in phosphate buffer solution at neutral pH should be the end-goal. Table 4 summarizes the sensing characteristics of enzyme-free glucose sensors using bimetallic nickel-based nanomaterials. Metal organic frameworks (MOFs), have emerged as a new class of materials that combine unique properties such as microporosity, high apparent surface areas, and exceptional thermal and chemical stability [152]. For all these reasons MOFs are highly attractive as potential materials for the development of sensors. However, their low electrical conductivity and instability in the aqueous media has limited their applications for electrochemical sensing [153,154]. However, enthralling opportunities are provided by the relatively novel field of conductive MOFs, which are characterized by highly conjugated and delocalized π-bond in the ligand. Such a structure facilitates electron transport and greatly enhances its electrical conductivity with high sensing capability [155].In a recent work, Zeraati and colleagues synthesized a Ni-MOF with an ultrasonic assisted reverse micelle synthetic route, with a sensitivity of 2859.95 μA mM−1 cm−2 and a LOD of 0.76 μM [156]. Using a facile one pot solution process Xiao et al. [157] demonstrated that the Ni-MOF nanobelt morphology is favorable to glucose oxidation, in particular due to its reduced thickness which maximizes the surface area.An interesting avenue, proposed by Wang et al. [158], consists in the combination of a hierarchical flower-like Ni-MOF with single walled carbon nanotubes (SWCNT) used to enhance the electrical conductivity. The authors noted that the addition of the SWCNT not only led to an increase in faradaic current density but also led to a decrease in the peak-to-peak distance for the Ni(II)/Ni(III) couple suggesting an improvement also in the electrochemical reversibility. Analogously, Zhang et al. [159] achieved an extremely high sensitivity of 13850 μA mM−1 cm−2 by combining a Ni-MOF with carbon nanotubes. An even greater sensitivity of 21 744 μA mM−1 cm−2 was reported by Qiao et al. [160] for a Ni-MOF synthesized via solvothermal methods on carbon cloth.With an exciting approach, Xue et al. [161] fabricated a 2D Ni@Cu-MOF by simple room temperature stirring. After performing electrochemical impedance spectroscopy analysis, the researchers concluded that the addition of Ni to the Cu-MOF caused an overall decrease in electrical resistivity. The as obtained sensor showed a sensitivity of 1703 μA mM−1 cm−2 and a LOD of 1.67 μM in 0.1 M NaOH. In an comparable study, Kim et al. [162] developed a Ni@Cu MOF though a two-step hydrothermal method.Bimetallic MOF based on Co have also been synthesized with different morphologies and supports [163,164], due to the notable performance afforded by the synergistic interactions between the two metals.As an example, in the solution proposed by Xu et al. [165], a nanorod-like bimetallic Ni/Co MOF was grown on a carbon cloth support. The sensor boasted a high sensitivity of 3250 μA mM−1 cm−2 and a low LOD of 0.1 μM. The authors attributed the notable performance to the Ni/Co synergy and to the high surface area of the open framework structure. Cao and colleagues [166] investigated the effect of the integration of Ag nanoparticles in a matrix of Ni-MOF nanosheets. Compared to other systems in the literature, their sensor displayed a lower sensitivity (160 μA mM−1 cm−2).A fascinating proposal by Lu et al. [167] consists in the use of core-shell MOF@MOF by internal extended growth of a shell of Ni-MOF on a core UiO-67. A schematic of the novel synthetic process is presented in Fig. 11 . The researchers compared the electrochemical response of the composite with that of Ni-MOF and noted a decrease in the peak-to-peak distance and an increase in the glucose oxidation peak. This was attributed to the excellent electrical conductivity of UiO-67, which in turn improved the electron transfer rate constant. Table 5 provides a summary of the sensing performance of nickel MOF-based sensors.Ni-based materials are attracting the attention of the scientific community for their outstanding performance towards glucose electrooxidation. Ni selenides, sulfides and nitrates are only recently being studied and have already shown promising results due to their strong redox capabilities and good electrical conductivity. The combination with a conductive carbonaceous matrix to form composite electrodes is a common solution to achieve a stronger electrocatalytic response due to an improved charge transfer constant. As a general rule, the highest sensitivity values for electrodes based on Ni-based compounds are obtained when the catalyst is grown directly on a Ni foam, instead of being drop cast and/or fixed on the surface with a binder.A promising avenue is the use of bimetallic nanosystems where Ni is a component (e.g., Cu/Ni, Co/Ni, Ag/Ni, Au/Ni and Pt/Ni) due to their superior activity, biocompatibility and fouling/poisoning resistance.Conductive MOF with Ni centers results in strong current responses thanks to their open pore structure combined with the oxidative power of Ni(III). The integration of Ni-MOF with nano-carbon based materials significantly improves the electrical conductivity and the combination with Cu and Co engenders synergistic interactions.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Nickel-based catalysts are currently the subject of intensive study in the search for novel electrode materials for non-enzymatic glucose sensing. Their strong activity towards glucose electrooxidation and intrinsic resistance to chloride poisoning makes these catalysts ideal candidates for the development of affordable and stable glucose sensors. In this review, the mechanism of glucose electrooxidation at Ni electrodes is described, clarifying the effect of the different phases of Ni on their catalytic activity. Moreover, a brief background on chloride poisoning is provided, supplemented by computational studies. Furthermore, this article details the most intriguing compounds of Ni (selenides, sulfides, nitrates) and the analytical performance of the respective sensors. Additional focus points of this work are multimetallic nanosystems where Ni is a component, and the growing field of conductive metal organic frameworks with Ni centers. This review will be beneficial for researchers who aim at delving deeper into the potential of Ni-based materials for glucose sensing.

Data will be made available on request. Data will be made available on request.Growing concerns about international dependencies and environmental impacts of fossil fuel sources are driving a national transition towards more sustainable and renewable energy sources, including bio-derived liquid fuels (Schiffer, 2022; O'Riordan and Sandford, 2022). Moreover, rising mobility demands are leading to intensified efforts to identify renewable drop-in replacements for both diesel and aviation fuel (Yan et al., 2021). The aviation industry in particular has high investment for biofuel substitutes considering electrification isn't as viable option for decarbonization as it is in the ground transportation sector. These shifting priorities are reflected in recent federal legislation that incentivizes biofuel production and use through tax credits and grants (Sustainable Aviation Fuel Grand Challenge, 2023; Brownley, 2023; Yarmuth, 2022). It follows that many groups are examining the potential of waste streams as reliable carbon sources that can serve a circular economic agenda (Awogbemi et al., 2021; Al-Muhtaseb et al., 2021). Among these, waste oils, fats and greases (e.g., used cooking oil) have drawn attention due to the high energy content and chemical similarity of long chain fatty acids to conventional diesel fuels (Borugadda and Dalai, 2018; Orsavova et al., 2015). Direct use of long chain fatty acids, and even their conversion to esterified biodiesel, is challenging due to flash point and viscosity discrepancies that limit mixing with or wholesale replacement of petroleum-derived fuels (Dey and Ray, 2020). However, fatty acid conversion to long-chain alkanes through deoxygenation and decarboxylation mechanisms offer potential for improved diesel engine compatibility with preferred ignition quality owing to their high cetane number (Yanowitz et al., 2017). This provides a compelling reason for identifying strategies for selectively deoxygenating fatty acids to hydrocarbons, especially fatty acids prevalent in food waste streams, e.g., oleic acid.Adding to the challenge in deoxygenating fatty acids with conventional refinery catalysts, waste fatty acid streams are often characterized by high moisture contents (Peng et al., 2008; Lizhi et al., 2008; Peterson et al., 2008), and the hydrodeoxygenation process, itself, generates water as a stoichiometric byproduct. As a result, there has been significant efforts directed towards development and commercialization of aqueous-phase catalytic processes for conversion of fatty acids, most often performed under hydrothermal conditions (250–400 °C, 10–18 MPa) (Peterson et al., 2008; ReadiJet- ARA 2023). Using a suitable hydrogen source, fatty acids like oleic acid can be converted in hydrothermal media to linear alkanes by hydrogenation (Eq. (1a)) and decarboxylation (Eq. (1b)) mechanisms: (1a) (1b) While there is a growing number of reports on hydrothermal fatty acid-to-alkane conversion, most studies have employed expensive noble metal catalysts, including platinum, palladium, and rhodium (Mäki-Arvela et al., 2007; Murata et al., 2010). Recent costs of these active metals has ranged from $900 to 2400/oz (Daily Metal Price: Free Metal Price Tables and Charts 2023). Moreover, significant negative environmental impacts are often associated with mining and processing of these rare metals (Burnett et al., 2021; Amatayakul and Ramnäs, 2001), something that is counter to the broader sustainability goals of biorenewable fuels. Some of the present authors recently reported on successful hydrothermal fatty acid-to-hydrocarbon conversion using supported ruthenium catalysts as a lower cost noble metal substitute, but market prices for this metal have also grown dramatically in recent years, highlighting additional challenges associated with the price volatility of these trace metals.The above discussion has led to renewed interest in identifying more earth-abundant and low-cost metals that might also be effective catalysts for hydrothermal fatty acid conversions. A recent report by Zhang et al. (2018a) showed hydrothermal conversion of oleic acid to heptadecane can be accomplished using a Ni-Cu bimetal catalyst supported on ZrO2, where hydrogen was supplied by in situ aqueous phase reformation of methanol. Nickel and copper are priced at $0.74/oz and $0.25/oz, respectively, making them an obvious price cognizant replacement for aforementioned noble metals (Daily Metal Price: Free Metal Price Tables and Charts, 2023). ZrO2 has shown promise as a stable support material for use in harsh hydrothermal environments (Papageridis et al., 2020; Joshi et al., 2014). Furthermore, zirconium may have a functional role in reactions since it has been shown to facilitate hydrogen production from water gas shift reactions of liquid hydrogen sources (Stekrova et al., 2018; Amatayakul and Ramnäs, 2001; Lytkina et al., 2015). Though Cu itself is not thought to be catalytic on its own, its interactions with Ni have been shown to enhance the latter's activity for a variety of upgrading processes, including gasification, hydrogenation and pyrolysis (Rashidi and Tavasoli, 2015; Wang et al., 2020; Kumar et al., 2019). For example, Luo et al. used core-shell structured Ni-Cu nanocrystals on a carbon support for hydrodeoxygenation of 5-hydroxymethylfurfural and showed that the incorporation of copper resulted in >30% higher selectivity for 2,5-dimethylfuran as compared to the corresponding supported Ni mono-metal catalyst (Stekrova et al., 2018; Luo et al., 2017). Modeling of this co-metal has shown its ability to increase the rate and selectivity of the water gas shift reaction, leading to higher H2 production in aqueous systems (Stekrova et al., 2018; Lytkina et al., 2015; Gan et al., 2012) and promote active site clearing through carbon deposit oxidation (Stekrova et al., 2018; Boualouache and Boucenna, 2020).The present report revisits the recent findings by Zhang and coworkers (Rashidi and Tavasoli, 2015; Zhang et al., 2018a) to further examine the reactions of oleic acid and related fatty acids (stearic acid and linoleic acid) over Ni- and Cu- catalysts in hydrothermal media. Surprisingly, initial experiments comparing mono-metal and bimetal catalysts activity show similar oleic acid conversion with Ni/ZrO2 as Ni-Cu/ZrO2, inspiring a deeper investigation of Ni/ZrO2 reactivity. Thus, we conducted the first thorough evaluation of Ni/ZrO2 for deoxygenation of multiple long-chain fatty acids under hydrothermal conditions. Method of catalyst preparation (co-precipitation versus wet impregnation), hydrogen source and concentration, and catalyst deactivation pathways were all studied. Through relation of the highest performing catalysts’ abilities under incrementally changing conditions to their morphological distinctions, we build the foundation for more directed future catalyst design. Findings point to a path forward towards development of lower cost and more environmentally sustainable materials that will be critical to efforts targeting valorization of waste carbon streams like used oils, fats, and greases.Zirconyl(IV) nitrate hydrate (99.5%) was obtained from Acros Organics. Glycerol (ACS grade) was purchased from Merck. Sodium hydroxide (20 N) and methanol (optima grade) were obtained from Fisher Chemical. Boron trifluoride (20% in methanol solution), hexane (>97.0%), stearic acid (>98.5%), linoleic acid (99%), nickel(II) nitrate hexahydrate (>94.5%), copper(II) nitrate trihydrate (analysis grade), oleic acid (90%), ethylene glycol (>99%), glucose (>99.5%), formic acid solution (1 M), dichloromethane (>99.8%) and sodium carbonate (>99.5%) were obtained from Sigma-Aldrich.Metal co-precipitation and wet impregnation methods were used to synthesize Ni- and Cu-based mono-metal and bimetal catalysts using tetragonal ZrO2 as a hydrothermally stable support material. Co-precipitation methods were adapted from Zhang et al. (2018). A 0.1 M solution of metal precursors, with the final Cu, Ni, or Cu-Ni (1:1 molar ratio) loading of 15 wt% relative to Zr salt was mixed with a second solution containing 0.15 M NaOH and 0.045 M Na2CO3. The two solutions were mixed in a round bottom flask to obtain a pH of 9.5, and stirred at 1000 rpm overnight. The resulting solid was collected by vacuum filtration and washed with deionized water before drying in an oven for 12 h at 110 °C. Tetragonal ZrO2 was synthesized in the same manner except that Ni and Cu salts were omitted from the preparation. Catalysts prepared by wet impregnation were synthesized using procedures adapted from previous reports (Freitas et al., 2018). Cu or Ni nitrate salt solutions (either 6.336 g Cu or 5.191 g Ni salts) were immobilized onto 6.667 g of the pre-synthesized tetragonal ZrO2 support material in 50 mL of water while sonicating for 30 min, followed by stirring at room temperature overnight at 250 rpm. The resulting solid was recovered and washed three times with water and ethanol before drying in an oven at 110 °C for 12 h. Independent of synthesis procedure, dried catalysts were ground with a motor and pestle before calcining in a furnace at 600 °C for 4 h with 1 h ramp time. Following calcination, individual catalysts were activated in a tube furnace under flowing H2 for 1 h at 650 °C after a 1 h ramp time (∼10 °C/min).Elemental compositions of the catalysts and selected aqueous phase samples following oleic acid conversions were obtained by inductively coupled plasma-atomic emission spectrometry (ICP-AES; Perkin Elmer). Prior to analysis, solids (3.5 mg) were microwave digested in 3 ml HCl + 9 ml HNO3 before diluting to 100 mL in deionized water for analysis. Phase composition was determined by X-ray diffraction (XRD, PANalytical PW3040 X-ray diffractometer) between 10 and 80° (2θ) at a scan rate of 37° min−1. HighScore spectral analysis was used for spectrographic assurance. To supplement ICP-AES and XRD for compositional confirmation, energy dispersive scanning transmission electron microscopy was carried out using a FEI Talos F200X STEM operating at 200 kV. To test for differences in fresh vs spent catalyst surface chemistry, temperature programmed reduction (TPR) was performed using a Micromeritics AutoChem II 2920 unit. Prior to TPR analysis, 35 mg of catalyst was pretreated to 550 °C under He/O2 gas mixture for 30 min. After which, the TPR was carried out to 600 °C under H2/Ar atmosphere at a gas flow rate of 50 ml min−1, temperature ramp of 10 °C min−1, and thermal conductivity detection (TCD).Mono- and bimetal catalyst activity was evaluated for hydrothermal conversion of oleic acid to stearic acid (hydrogenation, Eq. (1a)) and subsequent conversion to heptadecane (decarboxylation, Eq. (1b)) and possibly other hydrocarbon products. Hydrothermal batch reactions were performed in a stainless steel Swagelok microreactors (1.27 cm outer diameter × 10 cm length, 0.12 cm wall thickness) heated by submerging in a fluidized sand bath. Previous tests showed that the microreactors reach setpoint temperatures in <3 min (Li and Strathmann, 2019). Activity of the different catalysts were first compared at baseline conditions where 30 mg of catalyst was reacted with 100 mg of oleic acid and 20 mg methanol (as a source for in situ hydrogen production) in 1 m water at 350 °C for 5 h. Controls containing no catalyst were also performed at the same conditions. Activity of the most promising catalyst was further examined at a range of reaction conditions, including organic hydrogen source and concentration, temperature, and time. For comparison, reactions were also conducted with stearic (saturated fatty acid analogue) and linoleic acid (polyunsaturated fatty acid analogue) as the starting reactant. Reactions were quenched by removing reactors from the heated sand bath and submerging in a bath of water at room temperature. Once cooled, reactor contents were removed and washed three times with dichloromethane (DCM, 10 ml total) to extract residual fatty acid and conversion products. All reactions were carried out at least in duplicate.Fatty acids and hydrocarbons were analyzed by gas chromatography using a flame ionization detector (GC-FID; Thermo Scientific TRACE 1310 equipped with an Agilent DB-Wax 30 m × 0.25 mm × 25 µm capillary column). Hydrocarbon products were directly analyzed after injecting DCM extracts. For fatty acid analysis, samples were first subjected to a fatty acid methyl esterification (FAMES) procedure (Araujo et al., 2008) before GC-FID analysis. The injection and detection temperatures were 250 and 280 °C, respectively. Column temperature was increased from 50 to 200 °C at a ramp rate of 25 °C min−1 and then further to 230 °C at a ramp rate of 3 °C min−1. High purity H2 served was used as a carrier gas.Oleic acid conversions were calculated in molar yield percentages to gage the activity and selectivity of respective catalysts. Eqs. (2-3) were used to determine such activity and are as follows: (2) C o n v e r s i o n ( m o l % ) = ( 1 − m o l o l e i c a c i d f i n a l m o l o l e i c a c i d i n i t i a l ) × 100 (3) P r o d u c t Y i e l d ( m o l % ) = ( m o l p r o d u c t m o l i n i t i a l o l e i c a c i d ) × 100 % Fig. 1 shows results from initial experiments screening reactivity of oleic acid, a representative long-chain mono-unsaturated fatty acid, with mono-metal (Ni and Cu) and bimetal (Ni-Cu) ZrO2-supported catalysts. ZrO2 alone only led to minimal conversion of oleic acid to the corresponding saturated fatty acid, stearic acid, after 5 h of reaction. ZrO2 co-precipitated with Cu, however, saw more than double the conversion of oleic acid and production of stearic acid. No catalysts saw heptadecane production in the absence of nickel (Fig. 1a, Table 1 ). These findings track with those found by Zhang et al. when employing Cu on ZrO2 for hydrodeoxygenation of oleic acid with methanol as a hydrogen source. In comparison, complete conversion of oleic acid was observed during reactions with Ni/ZrO2 and Ni-Cu/ZrO2, with a mix of stearic acid and heptadecane products, the latter being the product of stearic acid decarboxylation. The GC-FID chromatograms (Fig. 2 ) showed minimal formation of other hydrocarbon cracking products, estimated to be <5% of product% yield. These products were relatively irrelevant for alkane production productivity and thus were not included in the final yields reported. It should be noted that, even when considering the cracking products, the quantified products don't lead to mass balance closure. This finding is not atypical for hydrothermal reactions, (Papageridis et al., 2020; Zhang et al., 2018c; Miao et al., 2016; Cai et al., 2022) particularly those conducted over nickel catalysts given its lack of controllability (Ananikov, 2015). For example, oleic acid is susceptible to aromatization under these conditions (Tian et al., 2017). Additionally, if acetic acid is generated via hydrocarboxylation of the methanol precursor, which has been shown to be promoted by ruthenium based catalysts, its presence can promote the peracid mechanism for the epoxidation of oleic acid (Qian et al., 2016; Jalil et al., 2019). Oleic acid epoxides can then polymerize to form polyesters and other long chain products with low vapor pressures making them difficult to detect by common analytical techniques (Japir et al., 2021; Borugadda and Dalai, 2018; Yeh et al., 2015). The presence of an unsaturated bond is necessary for these reactions, which is supported by the relatively higher mass balance achieved when starting with a saturated fatty acid (e.g., see Fig. 5). It is assumed that the products of these side reactions are consuming the remainder of the oleic acid, and no mass is being lost between reactor loading and product analysis. Surprisingly, heptadecane yields were slightly higher for the mono-metal Ni/ZrO2 (25.3%) compared to the bimetallic (20.5%) catalyst. This finding contrasts with the recent report by Zhang and co-workers (Zhang et al., 2018a), where heptadecane yields were markedly higher for Cu-Ni/ZrO2 (32.2%) compared with Ni/ZrO2 (22.7%) at the same reaction conditions used here and 3 h of reaction. The promotional reaction effects attributed to Cu in bimetallic catalysts are often recognized as electronic manipulation of the active metal catalyst via bimetal alloying (Kim et al., 2014). If metals are alloyed, their coordination state shifts, which is hypothesized to impact reactivity characteristics like intermediate adsorption time (Jin and Choi, 2019). XRD results in Fig. 3 a suggest that Cu and Ni are alloyed. Therefore, alloying cannot be independently claimed as the key to reactive synergy. It is possible that the addition of copper interferes with this interaction, possibly intercepting the support and the nickel at their interface. In future discussion, we acknowledge the significance of support interactions with the active metal, justifying how this interception could inhibit nickel's activity. It should also be noted that the study by Zhang et al. yielded higher in-situ pressure generation due to a higher loading to reactor headspace ratio, which could have promoted activity. The requirement of Ni for high overall activity is consistent with the metal's documented behavior in hydrogenation applications (Ananikov, 2015; Wang et al., 2020; Li et al., 2022). However, the requirement for decarboxylation is less clear given the non-reductive nature of decarboxylation reactions. Vardon and co-workers (Vardon et al., 2014) also observed elevated rates of stearic acid decarboxylation by Pt/C and Pt-Re/C catalysts when applying a reducing H2(g) headspace compared to inert N2(g) headspace despite the non-reductive nature of the reaction mechanism. It was hypothesized that the reducing conditions served to maintain the active metals in their catalytic form.Further tests supported the use of co-precipitation as an effective method for catalyst synthesis, as both the Ni and Ni-Cu catalysts prepared by alternative methods of wet impregnation of the ZrO2 support proved to be less active for the decarboxylation step critical to heptadecane formation (Fig. 1b). The high reactivity of co-precipitated Ni-Cu/ZrO2 has been attributed to increased nickel dispersion and metal-support contact, both initially due to nucleation rate differences of the two metals and long term due to stabilization effects by the bimetal (Liang et al., 2017; Elliott et al., 2006). Fig. 3 and Table 2 summarizes characteristics of the mono- and bimetallic catalysts, including Ni and Cu content from ICP-AES analysis. Catalysts were prepared with theoretical loadings of 15 wt% for each metal on the ZrO2 support, with ICP-AES analysis showing some divergence from these values. Higher than expected values can be attributed to less than theoretical ZrO2(s) recovery from solution. Nonetheless, active metal contents in the most active co-precipitated formulations of Ni/ZrO2 and Ni-Cu/ZrO2 were close to the nominal values.XRD analysis confirmed that all synthesized materials were tetragonal ZrO2 (t-ZrO2). These were used, in part, because commercial sources of ZrO2 are typically the less active monoclinic form of ZrO2 (Samson et al., 2014). The presence of a t-ZrO2 phase is indicated by XRD diffraction peaks at 2θ of 30.2°, 50.4°, 50.6°, and 60.0° (full scan XRD data provided in Fig. S1 in Supplementary Materials). For m-ZrO2, peaks would have been observed at 2θ = 28.7° and 34.2°. Additionally, XRD showed peaks at 43.9°, 50.6°, and 74.9°, which are representative of nickel and copper hybridized face-centered-cubic lattice peaks at signature (001), (200), and (220) planes respectively. The lack of a distinguished Ni peak at 45° in Fig. 3a-2 is either due to high dispersion leading to low diffractive definition, or peak overlay by ZrO2 at 45.5° (Fig. 3a-4). Regardless, the peak shift to a centralized 44° position in Fig. 3a-3 indicates a hybrid diffraction between Ni and Cu which suggests metal-metal interaction. The orientation of such peaks are slightly offset between what would normally be characteristic locations of these lattice dimensions for both nickel and copper, possibly suggesting an alloyed bimetal structure.Dispersion of Ni and Cu on the synthesized materials was variable as shown in the STEM-EDS images provided in Fig. 3b–e. The co-precipitated Ni-Cu bimetallic catalyst shows finite nickel agglomerates embedded in the support matrix with copper more evenly dispersed. In comparison, wet impregnation of the Ni and Cu salts leads to larger agglomerated clusters around the ZrO2 support. These findings are consistent with reports by Zhang et al. that co-precipitation of Ni (and Cu in the case of bimetallic formulation) with Zr maximizes active metal-support interfaces that may be important for hydrothermal catalytic activity with oleic acid (Zhang et al., 2020). Fig. 3d, e show STEM-EDS and SEM of the fresh Ni/ZrO2 (CP) catalyst synthesized via co-precipitation. SEM shows the morphology to be clumpy with slight geometric features and looks of deposits/growth on a larger support (Fig. 3e).Differences in reactivity observed with oleic acid (Fig. 1) go beyond elemental composition. Ni/ZrO2 prepared by co-precipitation showed higher heptadecane yields than the catalyst prepared by wet impregnation of Ni onto ZrO2, despite the latter formulation having significantly higher Ni content. The low activity of the bimetallic catalyst prepared by wet impregnation methods may result from incongruent deposition wherein Cu overlays Ni, limiting interactions with fatty acids or the hydrogen donor. Bimetal deposition is highly sensitive to synthesis conditions and slight deviations of metal collocation has shown to make large differences in the adsorptive behavior of the catalyst (Zhao et al., 2016; Yang and Cheng, 2014). During co-precipitation, the concerted nucleation of all metals promotes mixed metal deposition, which can prevent active site blockage (Tang et al., 2018). Additionally, coprecipitation of the support in the presence of simultaneously nucleating metals can also promote the formation of a higher concentration of oxygen vacancies in the support ZrO2, since the introduction of structural impurities can induce reverse oxygen spillover. Such vacancies may strengthen support-metal interactions and stabilize the support by lowering the dielectric constant as a result of lattice distortion around the vacancy (Samson et al., 2014; Chen et al., 2014; Gao et al., 2011).While these findings support earlier reports by Zhang and co-workers, the higher than anticipated reactivity observed for co-precipitated Ni/ZrO2 calls for a more in-depth examination of the material's hydrothermal reactivity with oleic acid and related fatty acids. Fig. 4 shows the effects of reaction temperature (Fig. 4a), reaction time (Fig. 4b), and catalyst loading (Fig. 4c) on oleic acid conversion and yields of heptadecane and stearic acid products. Complete conversion of oleic acid with Ni/ZrO2 within 5 h was observed at all temperatures >200 °C, but decarboxylation to produce heptadecane was not appreciable until temperatures reached 325 °C (Fig. 4a), reaching a maximum of 42% at 370 °C. Interestingly, side products decreasing stearic acid and heptadecane yields were already significant at a temperature of 200 °C, but only increased slightly at higher temperatures reflecting competition among parallel pathways for oleic acid conversion.A reaction temperature of 350 °C, intermediate between conditions where heptadecane production was observed, was used when examining other system variables. Complete conversion of oleic acid at this temperature was observed within 2 h, yielding principally stearic acid, but longer reaction times were required to further convert the saturated fatty acid to heptadecane (Fig. 4b). Further increases in heptadecane yield were limited after 20 h. Dosing additional methanol at t = 20 h also had limited effect on further conversion, so availability of excess hydrogen was not believed to be responsible for the stalled conversion. This is more likely the result of hydrothermal water-induced catalyst deactivation during the first 20 h of reaction. This would be consistent with previous reports (Wang et al., 2014; Elliott, 2008; Champon et al., 2020). Reasons for reduced activity are explored in Section 3.6.The yields shown in Fig. 4a and Table 1 should be contextualized in the literature of materials used for hydrothermal deoxygenation. Noble metals are the most prominent for these reactions and their higher yields are not unprecedented. Reports by Fu and coworkers (Fu et al., 2010, 2011) evaluate platinum and palladium on carbon for hydrothermal deoxygenation of fatty acids. At a reaction temperature of 330 °C, they observed a 9.3% yield of heptadecane over Pt/C from oleic acid after 1.5 h with no added hydrogen source. Interestingly, when starting with palmitic acid they observed a 55% yield for the decarboxylation product, pentadecane, after just 1 h. When starting with Pd/C, this yield dropped slightly to 50% pentadecane. Additionally, Vardon et al. observed a 37% heptadecane yield after 9 h of reacting a Pt-Re/C bimetal catalyst at 330 °C with oleic acid using glycerol as a hydrogen donor (Vardon et al., 2014). Ni/ZrO2 has also been explored for stearic acid hydrothermal deoxygenation by Miao et al. in 2018, which yielded 37% heptadecane after 9 h at 330 °C in the absence of an external H2 source (Miao et al., 2018). This agrees reasonably well with the yield we observed after reacting stearic acid for 5 h at 350 °C, albeit in the presence of methanol as a H2 source (Table 1). In 2021, Zeng at al. was able to achieve 60% heptadecane yield from stearic acid using Co3O4 nanoparticles and a carbon matrix shell, underscoring the possibilities that advanced catalyst design can have for the future of non-noble metal deoxygenation catalysis (Zeng et al., 2022).In the absence of a catalyst, around 18% of oleic acid was converted to other products, 27% was converted to stearic acid and none of it was converted to heptadecane (Fig. 4c). However, when 15 mg of the catalyst was loaded, rather than the typical 30 mg, the heptadecane yield dropped about 80% and the stearic acid yield almost doubled from 26% to 40%. Counterintuitively, when doubling the catalyst weight from 30 mg to 60 mg, heptadecane and stearic acid yields dropped to 17% and 12%, respectively, with complete oleic acid conversion retained. This optimum concentration suggests parallel pathways (e.g. polymerization, saponification, isomerization, etc.) are promoted by Ni ZrO2 and such are able to overcome decarboxylation in the presence of excess material. It should be noted that products produced from these pathways were not identified or quantified, however, there is a possibility that aromatics make up a portion of these products (Zhang et al., 2018b). Though these products have been found to be minimal in aqueous solvent systems (Tian et al., 2017), if produced they could be valorized for their high energy density and low smoke point heavily valued for diesel and jet fuel.Hydrothermal conversion experiments were also performed for the corresponding saturated fatty acid, stearic acid, and polyunsaturated fatty acid, linoleic acid, both with and without Ni/ZrO2. Results shown in Fig. 5 show that heptadecane production over the 5 h increases with increasing degree of initial fatty acid saturation. The only fatty acid that didn't see full conversion was stearic acid, further supporting that hydrogenation (Eq. (1a)) is not the rate limiting step in the production of heptadecane. It should be noted that, since full hydrogenation occurred in each reaction, no residual linoleic or oleic acid was detected, therefore they are not listed as products in Fig. 5. Cumulative heptadecane product yields also decreased with increasing degree of saturation, suggesting that the double bond(s) promote the rate of parallel side reactions that will deviate reactants from the desired deoxygenation pathway.Following from earlier work (Zhang et al., 2018a), methanol was added to reactor solutions to serve as a source for in situ hydrogen production in place of pressurized H2 gas addition. H2 is not only necessary for the saturation of oleic acid, but also the continuous reduction of the metal into its active metallic state. In hydrothermal water, methanol and other low molecular weight organic co-constituents can undergo aqueous phase reformation (APR) (Stekrova et al., 2018; Coronado et al., 2017). For example, each mole of methanol can react to form up to 3 mol equivalents of H2 with suitable catalyst and temperature (Eq. 4): (Coronado et al., 2017) (4) C H 3 O H ( m e t h a n o l ) + H 2 O → 3 H 2 + C O 2 In comparison, each mole of glycerol (C3H8O3) and formic acid (CH2O2) can react to form up to 7 mol and 1 mol of H2, respectively (Eqs. (5-6)): (5) C 3 H 8 O 3 ( g l y c e r o l ) + 3 H 2 O → 7 H 2 + 3 C O 2 (6) C H 2 O 2 ( f o r m i c a c i d ) → H 2 + C O 2 Fig. 6 a shows the effects of adding different hydrogen donor sources in oleic acid conversion by Ni/ZrO2. Added concentrations of individual hydrogen donors were varied according to source to keep total theoretical H2 production constant. Pressure variations resulting from differing hydrogen source loadings were taken to be negligible. Results showed that oleic acid conversion was dependent on hydrogen donor source, with heptadecane yields being greatest for methanol in comparison to glycerol and formic acid, despite the same theoretical hydrogen production potential. These discrepancies in decarboxylation potential can be related to the structures of the hydrogen donors. It has been shown that the binding of acidic functional groups, like that in formic acid, to the basic sites of amphoteric ZrO2 impedes catalytic transfer hydrogenation (CTH) responsible for hydrogenation and cyclization of butyl levulinate to valerolactone (Chia and Dumesic, 2011). Similar blockage could be inhibiting H2 donation or blocking sites necessary for reactant adsorption for reaction. Polyalcohols including sugars undergo aqueous phase reforming for H2 production similarly to methanol. However, methanol has been shown to have a greater selectivity for hydrogen when reformed over nickel (Coronado et al., 2016). Additionally, the methanol and water solvent mixture has been shown to have synergistic effects for biomass liquification, which also contains hydrogenation (Zhang et al., 2019; Zhao et al., 2020). Fig. 6b shows the effect of varying methanol concentration initially added to the reactor. In the absence of an external hydrogen source, oleic acid was still able to be hydrogenated, indicating the ability of Ni ZrO2 to promote the water gas shift reaction and produce H2 in situ. While the catalyst fully converted oleic acid at all conditions, heptadecane yield and overall product yield was lower when the methanol concentration was cut in half compared to the baseline concentration. More surprisingly, though, we also found that heptadecane yields dropped when higher methanol concentrations were used despite higher stearic acid yields. This suggests that formation of other products of APR, e.g., dissolved carbonate species, may act to inhibit stearic acid decarboxylation if present in too high of concentration. One such product could be CO, formed form either decarbonylation of the fatty acid or the reverse water gas shift reaction mentioned previously. CO has been shown to block nickel active sites, which would slow the production of heptadecane (Loe et al., 2019; Jackson et al., 1998). It is possible that some H2 donor precursors such as formic acid or glycerol could also promote the production of CO, which might limit decarboxylation efficiency. Future work is recommended to measure H2 and CO generation directly during reactions, which was precluded by the micro-reactors used in the present study.Finally, experiments were conducted to assess deactivation and potential regeneration of Ni/ZrO2 catalysts. It was already discussed that reactions appeared to stall after 20 h of reaction initiated with oleic acid, with further decarboxylation of stearic acid to heptadecane being limited between 20 and 30 h of reaction. Addition of a second spike of methanol had little effect, indicating that the stalled activity was not the result of limiting hydrogen supply (which we estimated to be in significant excess of hydrogenation requirements to begin with). These findings are consistent with previous reports of deactivation of Ni-based catalysts, where loss in activity is often attributed to corrosion, sintering, or leaching of the active metal, or surface coking of organic products (Crawford et al., 2020; Miao et al., 2016; Cheng et al., 2021; Champon et al., 2020). Analysis of dissolved Ni after the initial 20 h of reaction (Table 2) indicated minimal leaching, so other deactivation mechanisms were suspected. A separate recycling experiment compared oleic acid conversion using Ni/ZrO2 which had already reacted for 20 h with oleic acid compared to virgin catalyst. Partial deactivation was observed, where yields for heptadecane were 52% lower for the used catalyst than virgin catalyst (Fig. 7 ). Additional experiments showed that the catalyst deactivation occurs regardless of whether or not oleic acid and methanol are introduced with the catalyst to the hydrothermal media during the 20 h prior to initiating the reaction.Catalyst exposed to hydrothermal conditions for 20 h before introducing oleic acid and methanol was similarly deactivated (Fig. 7), excluding potential active site coverage by deposited organic species as the primary mode of deactivation. Catalyst pre-exposed to 20 h of hydrothermal conditions was also subjected to reactivation through re-calcination and furnace reduction as described in the initial catalyst synthesis procedure. Results showed no recovered activity for heptadecane production ( Fig. 7), suggesting that sintering may be responsible for the observed deactivation. The increase in the temperature of reduction as seen in H2-TPR data (Fig. 8 ) is observed after catalyst deactivation, increasing from 276 °C to 313 °C after undergoing 20 h reaction. This increase indicates a less H2-reactive form of Ni after extended exposure to the hydrothermal reaction environment. This may result from loss of active metal / support interaction upon hydrothermal treatment. Further, the shoulder peak observed in the fresh catalyst indicates inner-support nickel particles which have maximized support interface, often responsible for greater activity (Roh et al., 2002; Zhang et al., 2020). These results support that metal migration and sintering from the bulk to the surface occurs under hydrothermal conditions. Metal migration also displaces metals from their initial support deposition chemistry, disrupting the electronic effects offered by this interaction and leading to lower reducibility of the nickel (Colorado School of Mines, 2023). Catalyst activity recovery post-sintering has not yet been formalized in literature. Therefore, it should be prioritized in future work to synthesize catalysts in a way which limits or slows this morphological transformation in the hydrothermal reaction environment. Bimetals have been shown to prevent metal migration. However, copper did not prove useful for this purpose in similar deactivation experiments conducted here. Another option would be physical anchoring of the metals in the support matrix through ligands (Jenkins and Medlin, 2021), complex organic framework caging (Zhao et al., 2018), or surface locking through synthesis modification (Yang et al., 2019). Successful material development to circumvent this deactivation could further the competitiveness of this earth abundant catalyst against its noble metal counterparts.Co-precipitated Ni on tetragonal ZrO2 proved to produce the most efficient catalyst for decarboxylation of C18 fatty acids to heptadecane under subcritical hydrothermal conditions (300 – 370 °C). Non noble active metals were chosen for evaluation due to their price point relative to other commonly used platinum group transition metals which can cost over 99% more (Daily Metal Price: Free Metal Price Tables and Charts, 2023, BASF, 2023). Influence of copper as a non-noble bimetal on this reaction was null, contradicting recent reports. Methanol outperformed other organic liquid hydrogen sources. Catalyst and hydrogen donor loading were found to have an optimizable concentrations to encourage the reaction towards decarboxylation and away from fatty acid consuming parallel reactions. Greater heptadecane selectivity was observed when stearic acid was used as the starting reactant, emphasizing that the undesired parallel reaction pathways are more significant for the unsaturated fatty acid than saturated fatty acids. Deoxygenation over the active mono-metal catalyst scaled with temperature and reaction time up until the deactivation threshold, at which point metal sintering appears to deactivate the material. These specifications can inform optimal use of Ni ZrO2 for deoxygenation of fatty acids to liquid alkanes, however, more information regarding preventative deactivation synthesis is necessary to further advance these materials as a reliable alternative to their noble metal counterparts.Financial support for this work was provided by the National Science Foundation (NSF) through the NSF Engineering through the NSF Engineering Research Center for Reinventing the Nation's Urban Water Infrastructure (ReNUWIt; EEC-1028968) and NSF award CBET-1804513.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Financial support for this work was provided by the National Science Foundation (NSF) through the NSF Engineering through the NSF Engineering Research Center for Reinventing the Nation's Urban Water Infrastructure (ReNUWIt; EEC-1028968) and NSF award CBET-1804513. Moises Carreon, Praveen Kumar, Melodie Chen-Glasser, Ryan Richards, Galen Dennis and Matthew Posewitz (CSM) are acknowledged for assistance with analysis and materials characterization.Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.hazadv.2023.100273. Image, application 1

There is growing interest in the substitution of expensive noble metal catalysts with low-cost earth-abundant metals in applications targeting biofuels production from waste organic feedstocks. Here, nickel (Ni) catalysts supported on zirconium dioxide (ZrO2), both with and without copper (Cu) as a co-metal, were tested in hydrothermal reactions of unsaturated and saturated C18 fatty acids as models for waste oil feedstocks. In contrast to recent reports, this study showed no enhancement of nickel's activity for fatty acid conversion to alkane products when Cu was present. Ni/ZrO2 prepared by metal coprecipitation methods converted 100% of oleic acid with 25% selectivity to heptadecane after 5 h of reaction at 350 °C using methanol as a hydrogen donor source, increasing to 41% heptadecane after 20 h. Lower yields were observed with catalysts prepared by wet impregnation and using alternative hydrogen donor sources (glycerol, formic acid). Greater selectivity to heptadecane was also observed at higher temperatures (370 °C) and when the initial fatty acid had greater saturation. Longer term exposure to hydrothermal media led to metal sintering and catalyst deactivation. Findings support a path forward to the development of earth-abundant metal catalysts for the upgrading of waste organic feedstocks.

As important intermediates, alicyclic amines are widely used in chemical (anticorrosion products) and pharmaceutical industry. They are normally synthesized from the corresponding aromatic amines [1]. Precious metal-based catalysts [1–12] were mainly studied, while Co- and Ni-based catalysts were also reported [13–16]. Ru-based catalysts are applied in the reactions such as hydrogenation of aniline [1,5–7] and p-phenylenediamine [8]. They usually displayed high activity and selectivity to primary amines [1]. Nevertheless, deep understanding of ruthenium catalysts for the reactions at hand is still necessary.We have previously prepared supported Ni catalysts (60 wt%) and found that they were highly active for the hydrogenation of aromatic rings [17–19]. In particular, ethylamines were added so that the effects of -NH2 on the hydrogenation of toluene were studied [19]. We found that ethylamines inhibited the hydrogenation of toluene by restraining the adsorption of reactants on the Ni surface. In addition, the Ni-based catalysts with basic supports were less influenced by the ethylamine. Supported Co catalysts (60 wt%) were also prepared and it was found that the heat of TEA adsorption was lower on Co than on Ni [20], leading to the higher activity exhibited by Co/MgAlO than Ni/MgAlO for the hydrogenation of toluene in the presence of TEA.Previous work [19,20] demonstrated that the Ni and Co catalysts supported on MgAlO were more active than those with Al2O3 and MgO alone for the hydrogenation of toluene in the presence of TEA.In this work, 2–6 wt% Ru/MgO-Al2O3 catalysts were prepared for the hydrogenation of aromatic amines to alicyclic amines. Hydrogenation of toluene with and without triethylamine (TEA) was used as probe reaction. The heat of adsorption for TEA on Ru was found to be low, and thus the toluene adsorption was less affected by TEA on Ru than on Co and Ni. Thus, the activity was less affected by TEA on Ru than on Ni and Co for the hydrogenation of toluene. The selectivity to primary amines was also found to be high over the Ru catalysts for the hydrogenation of aromatic amines, owing to the weak adsorption of the amine group on the Ru surface.The 2–6 wt% Ru/MgAlO (MgO:Al2O3 = 7:1, w/w) were prepared by a two-step co-precipitation method similar to the one described previously [21]. Another 2 wt% Ru/MgAlO (designated as 2%Ru/MgAlO-2) as well as a 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO were prepared by the one-step co-precipitation method [21] for comparison. The detailed procedure of catalysts synthesis is described in the Supporting Information (SI).The composition and structural properties of the catalysts samples were characterized by N2 adsorption-desorption isotherms, powder X-ray diffraction (XRD), and X-ray fluorescence spectroscopy (XRF).Hydrogen temperature programmed reduction (H2-TPR), H2 titration and microcalorimetric adsorption of H2, toluene and TEA, were applied to study the surface chemical properties of catalysts. All the catalysts were reduced and passivated before characterizations were conducted. The detailed procedures are described in SI.The hydrogenation of aromatic amines and toluene was carried out in a stainless steel fixed-bed reactor with 10 mm inner diameter. The products were collected and analyzed by a GC with a capillary column and a FID detector. The turnover frequency (TOF) of toluene was calculated by dividing the number of toluene molecules converted per second by the number of surface Ru sites obtained from H2 titration. Under the high conversions used, the TOF (s−1) estimated represents an average site activity. Detailed description of the above procedures can be found in SI. The absence of mass transport effects (diffusional resistances) could be excluded for the reactions studied in this work as reported in the SI.Fig. S1 displays the N2 adsorption-desorption isotherms and pore size distributions of the reduced Ru/MgAlO catalysts, while Table 1 summarizes the information obtained. The isotherms indicated the presence of mesoporous structure (type IV with H3 hysteresis loops). The BET surface areas of the 2 wt% Ru/MgAlO, 4 wt% Ru/MgAlO and 6 wt% Ru/MgAlO were 369, 378 and 277 m2/g, respectively, corresponding to pore volumes of 1.37, 1.36 and 0.99 cm3/g, respectively. It is seen that the 2 wt% Ru/MgAlO and 4 wt% Ru/MgAlO possessed similar pore structures and surface areas, while the 6 wt% Ru/MgAlO displayed remarkably lower pore volume and surface area.The loadings of Ru measured and the H2 uptakes (metal dispersion measurements) are also given in Table 1. The loadings of Ru were close to the nominal values, indicating no loss of Ru during the preparation. When the loading of Ru increased from 2 to 4 wt%, the H2 uptake increased to 52 μmol/g. However, when the loading of Ru further increased to 6 wt%, the H2 uptake increased only by 30 μmol/g, indicating a decreasing dispersion of Ru in the 6 wt% Ru/MgAlO catalytic system. In fact, the ratio of H2 uptake to Ru loading was the highest for the 4 wt% Ru/MgAlO, which implies the highest dispersion of Ru among the series of supported Ru catalysts studied. After assuming the complete reduction of supported ruthenium, the dispersions and mean particle sizes of Ru in the catalysts were estimated and presented in Table 1.Fig. S2 shows the powder XRD patterns for the reduced 6 wt% Ru/MgAlO and MgAlO support alone. Only diffraction peaks belonging to MgO (at 44 and 62°) could be observed, indicating good dispersion of Ru particles in the 6 wt% Ru/MgAlO (Ru particles of less than 4 nm). The major peaks for Al2O3 are overlapped with those for MgO [22,23]. The peaks at 44 and 62° were more likely MgO since the mass ratio of MgO and Al2O3 in the Ru/MgAlO was 7:1. The powder XRD patterns for the 2 wt% Ru/MgAlO and 4 wt% Ru/MgAlO are not given since they were very similar to that of 6 wt% Ru/MgAlO.Fig. S3 shows the results of H2-TPR for the fresh Ru/MgAlO samples. For the 2 wt% Ru/MgAlO, a broad peak at 697 K and a narrow peak at 786 K were detected, indicating a strong interaction between Ru species and the support, where part of Ru species was difficult to be reduced. Only one broad peak was observed at 679 K for the 4 wt% Ru/MgAlO, which is due to Ru species interacting strongly with the support. This peak decreased to 663 K for the 6 wt% Ru/MgAlO, with an additional reduction peak appearing at 600 K. The latter represents weaker interactions of Ru species with the support and which led to the decreased dispersion of Ru in agreement with the H2 adsorption results. Fig. 1 shows the results of the conversion of aniline with temperature for its hydrogenation to cyclohexylamine (CHA) and the selectivity of reaction for the latter product over the 2 wt% Ru/MgAlO, 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO catalysts. CHA was the main product and dicyclohexylamine (DCHA) was the only by-product. It is observed that the conversion of aniline increased with reaction temperature. The 2 wt% Ru/MgAlO exhibited significantly higher conversion values of aniline than the 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO at the lower temperature range. The selectivity to CHA decreased significantly (deamination to DCHA) over the 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO, while this was maintained high on the 2 wt% Ru/MgAlO with increasing reaction temperature. For example, the selectivity to CHA at 453 K was 88, 39 and 40% over the 2 wt% Ru/MgAlO, 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO, respectively. Thus, the deamination reactions were significantly less active on Ru than on Ni and Co, leading to a higher selectivity to the primary alicyclic amines.The 4 wt% Ru/MgAlO, 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO catalysts were also compared for the hydrogenation of m-xylylenediamine (MXDA) to 1,3-cyclohexanedimethanamine (1,3-BAC) and 4,4-diaminodiphenyl methane (MDA) to 4,4′-diaminodicyclohexyl methane (H12MDA). The obtained results are shown in Table 2 . The by-products resulted mainly from the deamination reactions, except for the 60 wt% Co/MgAlO over which the main by-product was the 4,4′-diaminomonocyclohexyl monophenyl methane (H6MDA) for the hydrogenation of MDA. The results presented in Table 2 showed that the 4 wt% Ru/MgAlO was highly active and selective, as opposed to the other two catalytic systems for the the two reactions reported. A stability test for the hydrogenation of MXDA over the 4 wt% Ru/MgAlO is shown in Fig. S4, which indicated that the catalyst was stable for at least 240 h of reaction.To explain the high activity and selectivity of the Ru/MgAlO catalysts for the hydrogenation of aromatic amines, the hydrogenation of toluene in the presence of TEA was performed as a probe reaction. Fig. S5 compares the 2 wt% Ru/MgAlO, 60 wt% Ni/Al2O3 and 60 wt% Co/MgAlO catalysts for the hydrogenation of toluene. Methylcyclohexane was the only product; only 0.5% 4-methyl-1-cyclohexene was detected over the Ru/MgAlO at the high WHSV (weight hourly space velocity). It was found that the 2 wt% Ru/MgAlO displayed high activity for the reaction, similar to the 60 wt% Ni/Al2O3 and 60 wt% Co/MgAlO. Fig. 2 presents the effect of Ru loading on the conversion of toluene and TOF for the hydrogenation of toluene as a function of WHSV over the Ru/MgAlO catalysts. The conversion of toluene was about 1.4 times larger on the 4 wt% Ru/MgAlO than the 2 wt% Ru/MgAlO at WHSV of 80 h−1. However, the toluene conversion over the 6 wt% Ru/MgAlO is only 1.1 times larger compared to the 4 wt% Ru/MgAlO catalyst. According to the H2 uptakes, TOF values of toluene conversion over the Ru/MgAlO catalysts were calculated. The TOF increased with increasing WHSV, and a rather constant value was reached in the 80–100 h−1 range. TOF values of 1.3, 0.81 and 0.68 s−1 were estimated for the 2 wt% Ru/MgAlO, 4 wt% Ru/MgAlO and 6 wt% Ru/MgAlO, respectively. However, the average intrinsic activity of a Ru atom decreased, indicating that surface Ru atoms on the smaller particles were more coordinatively unsaturated, and thus more active.Fig. S6 and Table 3 compare the activities (conversion and TOF) of the Ru/MgAlO catalysts for the hydrogenation of toluene after using different amounts of TEA. It is apparent that the addition of TEA decreased toluene conversion and the activities further decreased with increasing amount of TEA added. However, the loading of Ru did not seem to affect the inhibitory effect of TEA.Table S2 compares the effect of TEA on the conversion and TOF of toluene for its hydrogenation over the 2 wt% Ru/MgAlO, 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO catalysts. The results showed that the 2 wt% Ru/MgAlO was much less affected by TEA than the 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO. For example, the TOF decreased by 42, 93 and 89% over the 2 wt% Ru/MgAlO, 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO, respectively, when 5 wt% TEA was added. These results agree well with those for the hydrogenation of aromatic amines to alicyclic amines over the Ru, Co and Ni catalysts.Fig. S7 (SI) compares the TEA adsorption on the reduced 2 wt% Ru/MgAlO, 60 wt% Ni/MgAlO, 60 wt% Co/MgAlO and MgAlO support, with the initial heats of 109, 177, 82 and 111 kJ/mol [19,20], respectively. It is seen that the interaction of TEA with MgAlO is fairly strong (111 kJ/mol), while the interaction of TEA with Ni is very strong (177 kJ/mol). In contrast, the interactions of TEA with Ru and Co surfaces were even weaker than with that of support (MgAlO), which might be the reason why the conversion of toluene was less affected by TEA on the Ru/MgAlO and Co/MgAlO. It should be mentioned that the lower initial heat of TEA obtained on the 60 wt% Co/MgAlO than on the 2 wt% Ru/MgAlO catalytic surface does not necessarily mean a stronger interaction of TEA with Ru than Co, since the loading of Ru in the 2 wt% Ru/MgAlO was low, leading to the similar coverages and heats for the TEA adsorption on the 2 wt% Ru/MgAlO and MgAlO surfaces.The low heat of adsorption of TEA on Ru indicates the weak interaction of -NH2 group with the Ru surface, which might account for the low selectivity to the deamination reactions (including condensation of -NH2 groups), i. e., the high selectivity to the primary amines. Fig. 3 shows the effect of surface coverage of toluene on the heat of adsorption of toluene with and without pre-adsorbed TEA on the surface of the 2 wt% Ru/MgAlO and MgAlO support alone. The initial heats of adsorption were found to be 71 and 22 kJ/mol on the clean 2 wt% Ru/MgAlO and MgAlO, respectively, indicating the fairly strong interaction of toluene with Ru and the weak adsorption of toluene on the MgAlO support. With pre-adsorbed TEA, the initial heat of toluene adsorption on the 2 wt% Ru/MgAlO decreased to 50 kJ/mol, suggesting the occupation of some surface Ru sites by pre-adsorbed TEA, and the restriction of toluene adsorption on the Ru surface. Fig. 4 shows results of the heat of adsorption of hydrogen on its surface coverage following preadsorption of TEA over the 2 wt% Ru/MgAlO catalyst. The initial heat of H2 adosorption on the 2 wt% Ru/MgAlO was found to be 69 kJ/mol. With pre-adsorbed TEA, the initial heat of adsorption (very low surface coverage) decreased to 48 kJ/mol, while the coverage was practically the same, indicating the inhibition effect of TEA on the H2 adsorption.Table S3 compares the initial heats of adsorption of toluene and H2 on the 2 wt% Ru/MgAlO, 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO without and with pre-adsorbed TEA. With pre-adsorbed TEA, the initial heat of toluene adsorption on the 2 wt% Ru/MgAlO decreased by 21 kJ/mol, significantly lower than that on the 60 wt% Ni/MgAlO (85 kJ/mol) and 60 wt% Co/MgAlO (42 kJ/mol). This result is suggested to explain well why Ru was less affected by TEA than Ni and Co for the hydrogenation of toluene. Table S3 (SI) also shows that pre-adsorbed TEA greatly decreased the initial heat of H2 adsorption over the 2 wt% Ru/MgAlO. However, the coverage of H2 did not change much. Our previous work indicated that strongly adsorbed H might be less active than the weakly adsorbed one [24]. Thus, it is suggested that the decreased activity for toluene hydrogenation over the supported Ru, Ni and Co catalytic surfaces might be mainly due to the inhibition of adsorption of toluene (rather than that of adsorption of H2) by TEA.It has been reported that the majority of active Pt metal sites for H2-SCR might be located within a ring around the Pt metal particles [25,26]. Thus, in the present work the metal-support interface might play important role. Unfortunately, the current data do not allow to identify and quantify the role played by the metal-support interface in the various catalytic systems.Highly dispersed Ru (2–6 wt%) supported on MgAlO carrier prepared by a two-step co-precipitation method were found much more active and selective than 60 wt% Ni/MgAlO and 60 wt% Co/MgAlO for the hydrogenation of aromatic amines to alicyclic amines, and for the hydrogenation of toluene in the presence of TEA (probe reaction) as well.The activity for toluene hydrogenation over the Ru/MgAlO decreased significantly upon the addition of TEA. However, the Ru/MgAlO catalysts were remarkably less affected by TEA than Ni/MgAlO and Co/MgAlO for the hydrogenation of toluene, which explains well why Ru/MgAlO catalysts were much more active than Ni/MgAlO and Co/MgAlO for the hydrogenation of aromatic amines to alicyclic amines.The pre-adsorption of TEA was found to decrease the heat of adsorption for toluene on the investigated catalysts. However, the heat of adsorption decrease was greatly less on the Ru/MgAlO than the Ni/MgAlO and Co/MgAlO. This result might account for that the Ru/MgAlO was less influenced by the presence of adsorbed TEA than the Ni/MgAlO and Co/MgAlO for toluene hydrogenation.The high selectivity to the primary amines for the hydrogenation of aromatic amines over the supported Ru catalysts can also be attributed to the low heat of adsorption due to the presence of amine group over the Ru surface. Li Zuo: Data curation, Writing – original draft. Jingxuan Cai: Methodology, Supervision, Writing – review & editing. Zhouyang Guo: Resources, Investigation. Yuchuan Fu: Resources, Visualization, Supervision. Jianyi Shen: Conceptualization, Methodology, Writing – review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Financial supports from the NSFC (21773108), NSFC-DFG (21761132006) and fundamental research funds for central universities are acknowledged. Supplementary material Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.catcom.2022.106496.

Ru/MgO-Al2O3 catalysts (2-6 wt% Ru) were prepared by a two-step co-precipitation method and were found to be highly active and selective for the hydrogenation of aromatic amines to alicyclic amines. Hydrogenation of toluene in the presence of triethylamine (TEA) was used as a probe reaction to explain the obtained results. It was found that adsorption of toluene in the presence of TEA on Ru than on Co and Ni catalysts was significantly less inhibited and thus the conversion of toluene. Ru/MgO-Al2O3 was thus found to be significantly superior to the Ni/MgO-Al2O3 and Co/MgO-Al2O3 for the hydrogenation of aromatic amines to alicyclic amines.

With the decrease of light crude oil supplies, there is rising concern about converting heavy oil into transportation fuels. As a poor-quality heavy oil, fluid catalytic cracking (FCC) slurry oil is mainly used as a blending component for producing marine fuel oil [1,2]. To improve air quality, strict emission regulations for marine fuel oil have been made compulsory by International Maritime Organization. The high sulfur and nitrogen content of FCC slurry oil inevitably affects its market share as the blending component of marine fuel oil, owing to the increasingly stringent ship emission standards. FCC slurry oil is abundant in polycyclic aromatic hydrocarbons and contains a certain amount of nitrogen, sulfur, and heavy metal compounds. Hydroprocessing is one of the most important routes to utilize the heavy oil resources effectively [3–9]. During the hydroprocessing of FCC slurry oil, the feedstock impurities such as sulfur and nitrogen can be removed by hydrotreating reactions. Meanwhile, large molecules can be transformed to value-added liquid fractions of light and medium distillates by hydrocracking reactions.The development of hydrogenation catalyst is believed to play a decisive role in the overall hydroprocessing effect [10–13]. The supported Mo-based sulfide catalysts are often used for hydroprocessing heavy feeds [11,14–17], and Ni is used as a promoter to decorate the edges of MoS2 slabs. Additionally, the support surface disperses the active metals via interaction and facilitates catalyst activation (sulfidation). However, too strong interaction between active metals and the catalyst support can lead to poor reducibility of metal species, which will make it difficult to sulfide and ultimately result in poor catalytic performance. Acidic supports such as alumina [11,15] and mixed oxides (SiO2-Al2O3 [18,19], TiO2-Al2O3 [3,20,21], B2O3-Al2O3 [22,23], etc.) has been reported for the heavy-oil upgrading process. In particular, the introduction of boron and, or phosphorus to alumina mainly affects the acidity and active metal dispersion [24–29]. Still, contradictory results have been obtained on the distribution of Mo species. Adding boron into Al2O3 slightly increased the acidity of the B2O3-Al2O3 support and promoted the hydrocracking activity of NiMo/B2O3-Al2O3 [22]. Moreover, an appropriate amount of B2O3 could improve the dispersion of active nickel and molybdenum species and promote the hydrogenation ability of the NiMo/B2O3-Al2O3 catalyst. In contrast, Morishige et al. [30] showed that boron doping weakened the interaction between γ-Al2O3 and Mo species in the Mo/γ-Al2O3 catalyst and decreased the Mo dispersion. Xiang et al. [31] indicated that phosphorus doping promoted MoS2 dispersing on the NiMo/Al2O3 catalyst surface by reducing the slab length and increasing the stacking number of MoS2 particles. Zhao et al. [26] reported that phosphorus modification enhanced the Mo sulfidation degree as well as the Mo dispersion and thus caused an increase in the denitrification rate. In addition, a decline in asphaltene conversion [11] and a slight increase in activities of hydrodesulfurization(HDS), hydrodenitrogenation(HDN), and hydrodemetallization [25] were found on NiMo/Al2O3 with phosphorus modification for the hydrocracking of vacuum residue. In these above studies, boron and, or phosphorus modification involves boric acid, borane isoproxyl [29], phosphoric acid, or ammonium dihydrogen phosphate [31] as starting materials.Boron phosphate (BPO4) has been applied as an effective catalyst or a catalyst support for some reactions such as oligomerization [32,33], dehydration [34], epoxide activation [35], and oxidative dehydrogenation [36]. However, little attention has been given to the boron phosphate-aluminum (BPO4-Al) composite as catalyst support for catalytic hydrogenation of real oils. In this study, the NiMo/BPO4-Al catalysts with different boron phosphate/aluminum molar ratios were synthesized with a complete liquid-phase method. A sol–gel route was used to prepare the catalyst precursor and then a thermal treatment in liquid phase was performed. The catalytic performance of these catalysts was evaluated with the hydroprocessing of FCC slurry oil in a batch reactor. The impact of boron phosphate as a support modifier on the catalytic performance was discussed in terms of HDS rate, HDN rate, and changes of hydrocarbon groups fractions and distillation fractions distribution of liquid oil product. X-ray diffraction (XRD), nitrogen physisorption, H2 temperature-programmed reduction (H2-TPR), temperature-programmed desorption of ammonia (NH3-TPD), X-ray photoelectron spectroscopy (XPS), and transmission electron microscopy (TEM) techniques were employed to characterize these catalysts and discuss their catalytic activities.A set of four oxidized NiMoAl catalysts with boron phosphate modification were prepared through the complete liquid-phase method, with a MoO3 loading of 24.0 wt% and a Ni/Mo molar ratio of 0.5. The preparation process was based on the description for synthesizing CuZnAl catalyst by Huang et. al. [37,38]. Appropriate amounts of isopropyl alcohol aluminum (AIP) were firstly mixed with isopropyl alcohol, and the resulting mixture was stirred at 85 °C for 3 h. Then deionized water was slowly added, and nitric acid was used to adjust the pH value of the mixture solution. After 2 h of the hydrolysis reaction of AIP, nickel nitrate, ammonium molybdate, and boron phosphate were introduced into the above mixture successively and this system was maintained at 95 °C for 6 h to get the light green thin sol. After aging, the as-made gel was dispersed in the liquid paraffin system containing span-80 and treated under the nitrogen flow at 300 °C for 4 h. The resulting four oxidized catalysts were labeled as NM-BPA(0), NM-BPA(0.04), NM-BPA(0.28), and NM-BPA(0.55), indicating the molar ratio of boron phosphate/aluminum of 0, 0.04, 0.28, and 0.55 in the BPO4-Al mixture support.Before the measurements, the prepared oxidized NM-BPA(x) catalysts were thoroughly extracted with petroleum ether to remove the surface liquid paraffin and then dried in an oven overnight. XRD patterns were collected on a X-ray diffractometer (Rigaku D/MAX-2500) with a Cu Kα radiation source at 40 kV and 100 mA. Nitrogen adsorption/desorption experiments were carried out with an Autosorb analyzer (Quantachrome) at −196 °C. Before each run, 0.20 g of sample was degassed under vacuum at 200 °C for 5 h to remove impurities. H2-TPR measurements were made using an automatic analyzer (Micromeritics AutoChem II 2920) equipped with a thermal conductivity detector. The sample was heated in a mixture of 5 % H2 in Ar and the TPR profile was obtained in the temperature range of 50–750 °C. NH3-TPD measurements were conducted on a chemisorption analyzer (Xianquan TP-5080) equipped with a mass spectrometry detector. The sample was pretreated with He at 300 °C for 30 min and then adsorbed with ammonia at 100 °C. After purging for 30 min with He, the sample was heated up to 800 °C with a heating rate of 10 °C/min. The sulfided NiMo samples were characterized with XPS (Thermo Scientific Escalab 250Xi) and TEM (JEOL JEM-2100F) techniques. Detailed experimental conditions could be found in the previous literature [39,40].The catalytic hydroprocessing experiments were performed in a high-pressure autoclave (100 mL) equipped with a magnetically driven impeller. An FCC slurry oil provided by Shaanxi Yanchang Petroleum Group Co. ltd. of China was used as the feed. In a typical test, 1.2 g of oxidized catalyst, 0.338 g of carbon disulfide, and 30 g of FCC slurry oil were placed into the reactor. After replacing the air with hydrogen completely, the autoclave was pressurized with hydrogen to 6 MPa at room temperature. The reaction conditions were as follows: reaction temperature of 400 °C, reaction pressure of 9 MPa, reaction time of 6 h, and stirring rate of 600 r/min.The sulfur and nitrogen concentrations in the feed and the collected liquid product oil were measured with an ultraviolet fluorescence sulfur analyzer (ZDS-2000A, Jiangsu XinGaoke) and a chemical luminescent nitrogen analyzer (ZDN-2000A, Jiangsu XinGaoke), respectively. The boiling range distribution of liquid oil samples was determined with a gas chromatography simulated distillation method (ASTM D2887). Saturates, aromatics, resins, and asphaltenes (SARA) compounds in the feedstock and liquid products were separated and quantified according to NB/SH/T 0509. The H/C atomic ratio of liquid products was determined by elementary analysis (Thermo Scientific FlashSmart Elemental Analyzer).The HDS rate (X S) and HDN rate (X N) were calculated by using the following formula: (1) X S = m 0 × S f e e d - m 1 × s p r o d u c t m 0 × S f e e d × 100 % (2) X N = m 0 × N f e e d - m 1 × N p r o d u c t m 0 × N f e e d × 100 % where S feed and S product denote the sulfur concentrations (μg/g) in the feedstock and liquid product; N feed and N product indicate the nitrogen concentrations (μg/g) in the feedstock and liquid product, respectively; m 0 and m 1 are the mass (g) of the feedstock and liquid product, respectively.The prepared oxidized catalysts were characterized by nitrogen physisorption, XRD, H2-TPR, and NH3-TPD. The textural properties of NM-BPA(0) and NM-BPA(0.55) samples are given in Table S1 and Figure S1. The BET surface area and total pore volume of NM-BPA(0) are 281 m2·g−1 and 0.47 cm3·g−1. The NM-BPA(0.55) sample with BPO4 modification has a lower specific surface area and a higher pore volume, which is 190 m2·g−1 and 0.53 cm3·g−1, respectively. Both two catalysts show a bimodal mesopore distribution. In the smaller pore diameter region, mesopores with a pore size of 3.7 nm can be seen in these two samples. Besides, secondary mesopores with a wide pore size distribution is centered at 5.8 nm for NM-BPA(0) and 6.9 nm for NM-BPA(0.55). Fig. 1 shows XRD patterns of NM-BPA(x) samples. As for the unmodified NiMoAl catalyst, four weak peaks at 28.2°, 38.4°, 49.3°, and 64.5° are observed, and they correspond to AlOOH (JSPDS 49–0133). In the three NM-BPA(x = 0.04, 0.28, 0.55) samples, we can observe new distinct peaks at 24.5°, 26.8°, 29.1°, 40.0°, 48.8°, 50.2°, and 63.7°, which are attributed to boron phosphate (JSPDS 34–0132). This suggests the presence of BPO4-AlOOH mixed support in these BPO4-modified NiMoAl samples. As for the above four catalysts, no nickel and molybdenum species are noticed, indicating that active metal components are well dispersed on the catalyst support.The H2-TPR profiles of oxidized NM-BPA(x) samples are presented in Fig. 2 . The unmodified NiMoAl catalyst exhibits two hydrogen consumption peaks. According to the literature [41–43], peak α at 485 °C is related to the partial reduction of polymolybdate octahedral Mo species (Mo6+ to Mo4+), whereas peak β at around 583 °C is associated with the reduction of Ni species. Note that the NM-BPA(0.55) catalyst shows two reduction peaks at 467 °C (peak α) and 569 °C (peak β), respectively. The incorporation of boron phosphate (BPO4/Al = 0.55) into unmodified NiMoAl catalyst shifts peak α to lower temperature, suggesting that the polymolybdate species existing on NM-BPA(0.55) is easier to be reduced and sulfided. Fig. 3 A displays a set of four NH3-TPD profiles for NM-BPA(x) samples. The NH3 desorption temperature is related to the strength of acid sites. Two distinct NH3 desorption peaks can be observed at the temperature of 100–350 °C and 350–800 °C, which represent weak-intermediate and strong site acidity, respectively. Similar results were reported for NiMo/γ-Al2O3 catalysts with phosphorus or boron promoters [44,45]. As shown in Fig. 3A, for the samples with BPO4 modification, both of two NH3 desorption peaks shift to a higher temperature compared to the unmodified NM-BPA(0) catalyst, implying the strengthening of weak-intermediate and strong acid sites in the catalysts owing to the addition of boron phosphate. This can result from the strong interaction between boron phosphate and surface acid sites of AlOOH support. To obtain the distribution of weak-intermediate and strong acid amounts of the catalysts, each NH3-TPD profile was cut into two parts and integrated the corresponding peak areas, and the results are presented in Fig. 3B. The total acid amount of the NM-BPA(0) sample was considered to be 100, and those of other samples were calculated on the basis of the peak area ratio of BPO4-modified catalysts to the NM-BPA(0) sample. As can be seen in Fig. 3B, both the weak-intermediate acid amount and total acid amount of NM-BPA(x) samples decrease with increasing the molar ratio (0–0.55) of boron phosphate/aluminum in the BPO4-Al mixture support. But the strong acid amount shows a different variation trend, which means rising firstly and then falling. Therefore, BPO4 modification can change the surface acidity of the NiMoAl catalyst. The acidity strength is improved but the total acid amount decreases with incorporating BPO4 into the catalyst.The sulfided NM-BPA(x) catalysts were analyzed by XPS to obtain the surface element compositions and the chemical state. According to the literature [46], the calculated surface atomic ratios of n(Ni)/n(Al) and n(Mo)/n(Al) can be used to obtain information on the dispersion of active metals on the catalyst surface. As listed in Table 1 , the NM-BPA(0) sample has a n(Ni)/n(Al) ratio of 0.04 and a n(Mo)/n(Al) ratio of 0.10, respectively. It is important to point out that both the surface atomic ratios of n(Ni)/n(Al) and n(Mo)/n(Al) increase gradually with increasing the molar ratio of boron phosphate/aluminum in NM-BPA(x) catalysts. This indicates that adding BPO4 enriches Ni and Mo species on the catalyst surface, which may improve atom utilization and thereby lead to better catalytic activity. Fig. 4 shows the measured and curve-fitted Mo 3d XPS spectra of four sulfided NM-BPA(x) catalysts. The fitting principles have been described in previous work [47]. As can be seen in Fig. 4, three Mo species with different oxidation states, i.e., Mo4+ (MoS2), Mo5+ (MoSxOy), and Mo6+ (MoO3), are present on these sulfided NM-BPA(x) samples. It is generally acknowledged that n(Mo4+)/n(Mo4+ + Mo5+ + Mo6+) can be regarded as the sulfidation degree of Mo, and the calculated values for sulfided NM-BPA(x) samples are shown in Table 1. As can be seen from Table 1, NM-BPA(0.04) and NM-BPA(0.28) have similar Mo sulfidation degree with unmodified NM-BPA(0) sample. But the Mo sulfidation degree of NM-BPA(0.55) is 64 %, which is higher than that of the unmodified catalyst (54 %). This indicates that adding an appropriate amount of BPO4 into the NiMoAl catalyst promotes the sulfidation of Mo species to active Mo4+ species, which is consistent with the TPR result mentioned above.TEM analysis has been widely used to study the morphology structure of active phase for the sulfided NiMo catalysts [24,48]. Fig. 5 displays representative TEM images of four sulfided NM-BPA(x) catalysts. The observed black stripes are the (Ni)MoS2 active phase with a typical layered structure. To understand the influence of boron phosphate modification on the active phase structure of the sulfided NiMoAl catalyst, the slab length and stacking number of (Ni)MoS2 slabs have been statistically analyzed on the basis of 400–500 slabs for each sulfided sample, and the results are shown in Table 1 and Fig. 6 . As can be seen from Table 1, distinct differences in the average slab length and stacking number of (Ni)MoS2 crystallites are found among these four sulfided NM-BPA(x) catalysts. With increasing the molar ratio (0–0.55) of boron phosphate/aluminum in the BPO4-Al mixture support, the average stacking layer number increases gradually from 1.3 to 2.1, whereas the average slab length of (Ni)MoS2 firstly decreases from 4.6 nm to 3.7 nm and then rises to 4.8 nm. It is widely accepted that, on the basis of “Rim-Edge” model of MoS2 active phase[49,50], multi-layer MoS2 slabs are more conducive to the adsorption of reactant molecules on the edge sites in comparison with single-layer (Ni)MoS2, thereby promoting the catalytic activity in hydroprocessing reactions. Meanwhile, for (Ni)MoS2 slabs with a certain stacking number, a shorter average slab length can lead to a larger dispersion of effective Mo atoms. Although NM-BPA(0), NM-BPA(0.04), and NM-BPA(0.28) have similar Mo sulfidation degree, the NM-BPA(0.28) catalyst has a higher average stacking number and smaller average slab length, indicating a higher dispersion of (Ni)MoS2 slabs.The distributions of (Ni)MoS2 stacking number and slab length of sulfided NM-BPA(x) samples are displayed in Fig. 6. As can be observed from Fig. 6A, the percentage of single-layer active phases gradually decreases, and the percentage of multi-layers (two to five layers) active phases increases when the molar ratio of BPO4/Al in NM-BPA(x) catalysts changes from 0 to 0.55. It is generally believed that two or more stacking layers of (Ni)MoS2 active phases can exhibit higher hydrogenation activity [26]. As displayed in Fig. 6B, the slab length of (Ni)MoS2 active phases in all four catalysts mostly distributes in the range of 2–6 nm. With the incorporation of boron phosphate (BPO4/Al = 0.04, 0.28), the percentage of (Ni)MoS2 slabs<4 nm increases. Although the NM-BPA(0.55) catalyst shows a similar distribution of (Ni)MoS2 slab length with the unmodified NM-BPA(0) catalyst, it has a much higher Mo sulfidation and (Ni)MoS2 stacking number, which could result in better catalytic activity.The catalytic activities of four prepared NM-BPA(x) catalysts were determined for hydroprocessing FCC slurry oil in terms of HDS, HDN, and hydrocracking. The desulfurization and denitrification results for all NM-BPA(x) catalysts are displayed in Fig. 7 . The feed FCC slurry oil contains 3047 ppm sulfur and 2212 ppm nitrogen. As shown in Fig. 7, the NM-BPA(0.04) catalyst has a higher HDS rate (77.3 %) compared to the unmodified catalyst. A decline in HDS activity can be observed with further increasing the molar ratio of boron phosphate/aluminum in the BPO4-Al mixture support. However, it is noted that the addition of boron phosphate greatly enhances the conversion of nitrogen-containing compounds. For example, the HDN rate increases from 7.6 % up to 24.2 % when the molar ratio of boron phosphate/aluminum reaches 0.55.It is known that the HDS and HDN activities of the NiMo catalyst are associated with the Mo sulfidation degree and the dispersion of active metals. The HDN reaction could proceed harder than the HDS reaction, owing to the stronger bond energy of CN bond than that of CS bond. During the HDN process, two types of reactions are involved, i.e., aromatic ring hydro-saturation and following CN bond scission. The coordinatively unsaturated sites (CUS, or “sulfur vacancies”), which are mainly located at the edge or corner sites of (Ni)MoS2 slabs, are responsible for the hydro-saturation reaction [20]. But in the HDS process, the sulfur removal of sulfur-containing compounds can occur directly on the CUS sites without associated aromatic ring hydrogenation. Thus, the competitive adsorption and hydrogenation of nitrogen compounds on the same active sites could hinder the adsorption and desulfurization of sulfur compounds.For the NM-BPA(x) catalysts, as discussed above, the incorporation of boron phosphate has a significant influence on the catalyst structures. XPS analysis evidences that the presence of boron phosphate causes the enrichment of Ni and Mo species on the catalyst surface and an increase in the Mo sulfidation degree, which favors promoting the HDN reaction. Besides, as indicated by TEM characterization, the morphology structures of active phases, i.e., the stacking layer number and slab length of (Ni)MoS2, are significantly adjusted because of the addition of boron phosphate into the NiMoAl catalyst. The dispersion of (Ni)MoS2 active phases is considerably improved with increasing the molar ratio (0–0.55) of boron phosphate/aluminum in the BPO4-Al mixture support, which results in a significant increase in the HDN activity. Although high HDN activity is often connected to a rise in the amount of intermediate strength acid centers [23,51], the variations in acid strength and acid amount caused by adding boron phosphate into NM-BPA(0) seem not directly correlated to the HDN activity, as indicated by NH3-TPD results. Additionally, due to competitive adsorption and hydro-conversion of nitrogen-containing molecules on the same active sites, the HDS rate of sulfur compounds decreases with increasing the molar ratio (0–0.55) of boron phosphate/aluminum in NM-BPA(x) samples. Similar results were also reported by Ferdous et al. [23] and Chen et al. [45]. A decrease in HDS activity with increasing boron content in NiMo/Al2O3 was observed for hydrotreating heavy gas oil [23]. Chen et al. [45] observed that the addition of boron promoted the HDN activity of NiMo/γ-Al2O3 catalyst, whereas it presented a detrimental effect on the HDS activity.To obtain the chemical composition change of the feed before and after hydroprocessing, oil samples can be separated and quantified by four different hydrocarbon groups based on solubility and adsorption characteristics, i.e., saturates, aromatics, resins, and asphaltenes [52–54]. The SARA fractions of liquid products obtained during hydroprocessing FCC slurry oil with different NM-BPA(x) catalysts are given in Table 2 . The FCC slurry oil is composed of 24.2 wt% of saturates, 44.3 wt% of aromatics, 26.0 wt% of resins, and 5.5 wt% of n-C7 asphaltenes. Using the unmodified NM-BPA(0) catalyst, the liquid product contains 31.6 wt% of saturates, 47.2 wt% of aromatics, 20.3 wt% of resins, and 0.9 wt% of n-C7 asphaltenes, indicating that the resins and asphaltenes fractions are cracked into lighter fractions. Moreover, the H/C atomic ratio increases from 1.21 to 1.29 through hydro-saturation reactions, in accordance with an increase in saturates fractions. As the molar ratio (0–0.55) of boron phosphate/aluminum in NM-BPA(x) catalysts increases, the resins fraction in liquid product falls from 20.3 wt% to 14.0 wt%, and the saturates fraction rises from 31.6 wt% to 37.8 wt%, implying that the addition of boron phosphate into the NiMoAl catalyst further upgrades the quality of liquid oil product. Although four prepared NM-BPA(x) catalysts show considerable activity in terms of n-C7 asphaltenes conversion, the NM-BPA(0.55) catalyst produces more saturates and fewer resins than other catalysts.The distillation fractions distribution of liquid oil products, as well as the variations of heavy components in SARA fractions, could also be used to evaluate the hydrocracking performance of catalysts [55–57]. As listed in Table 2, the liquid products are categorized on the basis of the boiling point as naphtha (<180 °C), middle distillate (180–350 °C), vacuum gas oil (350–500 °C), and residue (>500 °C). Among the four prepared catalysts in this study, the NM-BPA(0.55) catalyst gives the lowest residue fraction and the highest naphtha fraction. From these results, the NM-BPA(0.55) catalyst is suggested to be effective for hydrocracking heavy components into lighter oil. The introduction of boron phosphate into the NiMoAl catalyst promotes the hydrocracking activity during the hydroprocessing of FCC slurry oil.The present study showed the impact of BPO4 modification on the structure and heavy oil hydroprocessing performance of the NiMoAl catalyst. In this respect, the BPO4-AlOOH mixed support formed during the complete liquid-phase synthesis of BPO4-modified NiMoAl catalysts. At the same time, these catalysts exhibited a bimodal mesopore distribution. With the addition of BPO4, the surface acidity was strengthened, but the weak-intermediate acid amount and the total acid amount declined. For the sulfided NM-BPA(x) catalysts, both the surface atomic ratios of n(Ni)/n(Al) and n(Mo)/n(Al) increased with increasing the molar ratio of boron phosphate/aluminum, as indicated by XPS analysis. With the proper BPO4/Al molar ratio in NM-BPA(x) catalysts, the dispersion of (Ni)MoS2 active phases was enhanced by lowering the slab length and increasing the stacking layer number.It was noted that the HDN activity of the NiMoAl catalyst was greatly enhanced with increasing the molar ratio of boron phosphate/aluminum in NM-BPA(x) samples. The enrichment of Ni and Mo species on the catalyst surface, higher Mo sulfidation, and better dispersion of active phases were responsible for that significant increase in the HDN activity. However, the desulfurization rate increased first and then decreased with increasing the boron phosphate/aluminum ratio. Additionally, our work also confirmed that boron phosphate modification improved the hydrocracking activity of the NiMoAl catalyst during the hydroprocessing of FCC slurry oil. Compared to other synthesized catalysts, the liquid product obtained by using the NM-BPA(0.55) catalyst showed a considerable n-C7 asphaltenes fraction but more saturates and fewer resins. Meanwhile, lower residue and higher naphtha fraction in the liquid product were reached over the NM-BPA(0.55) catalyst. Changwei Liu: Data curation, Writing – original draft. Chunyan Tu: Methodology, Supervision, Writing – review & editing. Qi Chen: Data curation. Qian Zhang: Data curation. Wei Huang: Supervision.The authors declare no conflicts of interests.This work was supported by the National Natural Science Foundation of China (21808155) and the National Key R&D Program of China (2018YFB0604600-01).Supplementary data to this article can be found online at https://doi.org/10.1016/j.crcon.2022.07.001.The following are the Supplementary data to this article: Supplementary data 1

Mesoporous NiMoAl catalysts with boron phosphate (BPO4) modification were synthesized through the complete liquid-phase method. X-ray diffraction (XRD) analysis evidenced the presence of BPO4-AlOOH mixed support in these BPO4-modified NiMoAl samples. The total amount of acid sites declined, but the surface acidity was strengthened by adding BPO4 into the NiMoAl catalyst. It's worth noting that the incorporation of BPO4 could increase the concentrations of Ni and Mo species on the catalyst surface and greatly improve the dispersion of (Ni)MoS2 active phases, as indicated by X-ray photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM) measurements. The catalytic performance of these BPO4-modified NiMoAl catalysts was investigated with the hydroprocessing of fluid catalytic cracking (FCC) slurry oil. The nitrogen-containing compounds removal from the oil was significantly enhanced with increasing the molar ratio of boron phosphate/aluminum. The NM-BPA(0.55) catalyst exhibited the best hydrodenitrogenation (HDN) activity, highlighting the significant impact of Mo sulfidation degree and the dispersion of active metals on HDN performance. The introduction of boron phosphate could also promote the hydrocracking activity of the NiMoAl catalyst, as demonstrated by SARA analysis and simulated distillation of liquid products.

The rapid economic and societal development seen in recent decades has led to a dramatic increase in energy consumption worldwide. Fossil fuels such as coal and oil have been our major energy sources since the first industrial revolution; however, they are not clean energy sources. The combustion of fossil fuels releases a variety of pollutants, such as carbon dioxide (CO2), sulfur dioxide (SO2), oxynitride (NO x ), and other volatile organic compounds (VOCs), which cause a range of environmental problems [1–3]. Among them, the greenhouse effect, which is caused by excessive CO2 emissions, has gained wide attention because it causes climate change and endangers human health and mankind's survival [4,5]. Furthermore, human reliance on fossil fuels has led to energy shortages. Therefore, converting CO2 into fuels and/or other valuable commodity chemicals constitutes an attractive strategy for mitigating these problems [6–8].Scientists around the world have devoted enormous effort to solving the problem of CO2 reduction. CO2 reduction reactions can be classified into four categories according to the energy source: thermocatalysis, photocatalysis, electrocatalysis, and photothermal catalysis, which is a combination of thermo- and photocatalysis [9–11]. Of these, thermocatalysis has some obvious drawbacks, such as high energy cost and relatively low product selectivity, which is hard to regulate. Photocatalysis and electrocatalysis are catalytic methods that have received increasing research attention in recent years because they can be used to convert CO2 into valuable fuels under milder conditions and are more efficient than conventional thermochemical processes [9]. However, the reaction barrier of electrocatalysis is high and the reaction kinetics are sluggish, and photochemical catalysis is subject to limited utilization of solar energy [12]. Recently, scholars have attempted to combine light and heat to perform CO2 reduction by photothermal catalysis. The coupling of solar and thermal energy can effectively regulate the activity and selectivity of this reduction reaction. In addition, it offers a new way to fully utilize solar energy.Photothermal catalytic materials such as metal sulfides, metal oxides, carbon nitrides, and metal–organic frameworks (MOFs) have been widely researched. Metallic materials are commonly used to prepare photothermal catalysts, among which group VIII elements have unique activation capacities and efficient energy utilization over the entire solar spectrum, as experimentally confirmed by Meng et al. [13] They reported that group VIII elements show great promise for CO2 reduction. Among these elements, Ni is a particularly good candidate catalyst material because, in addition to its high activity, it is more abundant in nature and thus easier to obtain than noble metals. Although no CO2-to-fuel technology has been industrialized, Ni-based materials are very promising for such applications. Ni is also the first-choice catalytic material for most methanation-manufacturing plants [14]. Furthermore, when Ni is present in the form of metal nanoparticles (NPs), it acts not only as active reaction sites [15], but also as heat-collection centers to increase the local temperature of the reaction system [15,16]. In addition, the local surface plasmon resonance (LSPR) effect is no longer limited to noble metals, and that of Ni NPs has recently gained attention [16,17]. Thus, Ni-based catalysts have strong applicability in photothermal catalytic systems.Herein, the mechanism of photothermal catalysis will be introduced from the perspective of different products. Furthermore, this review provides an overview of recent advances in Ni-based catalysis for photothermal CO2 reduction (Fig. 1 ).At present, scientists still have not reached a unified opinion on the definition of photothermal CO2 catalytic reduction. The source of heat and the substrate of the reaction are the main considerations when defining photothermal catalysis. Here, we adopt the most commonly recognized classification method, and divide photothermal catalytic reactions into three categories; photo-assisted thermal catalysis, thermally assisted photocatalysis, and photothermal co-catalysis. In this section, we will briefly illustrate their differences (Fig. 2 ).Photo-assisted thermocatalysis is essentially a thermocatalytic reaction. In this mode, there are two sources of thermal energy, one is an external heat source that increases the local temperature of the catalyst, and the other is solar energy. In the former case, the elevated temperature lowers the activation energy barrier of traditional thermocatalysis and alleviates catalyst poisoning. In the latter case, light serves as an energy source to drive thermocatalysis (also known as photo-driven thermocatalysis) [18,19]. Catalysts with a broad-spectrum absorption range efficiently convert concentrated solar energy into thermal energy, which then drives the reaction. Photo-driven thermocatalysis greatly improves the utilization of solar energy, and can be carried out under milder catalytic conditions.In thermal-assisted photocatalysis, the catalyst absorbs light or external heat to increase the temperature of the system and then drives the reaction with photo-generated electrons. When catalysts (usual semiconductors) absorb adequate energy from external photons, the electrons (e−) and holes (h+) formed in the conduction and valence bands are separated and transferred to the catalyst surface, where they then participate in redox reactions [19,20]. In contrast to that for thermal catalysis, the temperature for this mode is relatively low and the catalyst has no thermocatalytic activity. The heat energy reduces the apparent activation energy of the photocatalysis, accelerates the thermal molecular movement of reactants (or intermediate substances), and promotes the mobility of charge carriers.In photothermal co-catalysis, photo- and thermocatalysis are coupled and act simultaneously to promote the catalytic process. In this mode, the sources of thermal energy can be photothermal conversion, external resistive heating, and exothermic chemical reactions. The synergistic effects of light and heat energy improve the reactivity and selectivity of the system [19].It is difficult to completely distinguish between the thermochemical and photochemical pathways in photothermal catalysis systems [18]. In photo-assisted thermocatalytic reactions, very high photothermal conversion efficiency of the catalyst is required to reduce the input of other forms of heat, and it is therefore not commonly studied in practice. At present, whether a reaction proceeds by thermal-assisted photocatalysis or photothermal co-catalysis remains controversial, and there is still a lack of accurate judgment indicators [15]. In the case of the Ni-based catalysts of interest in this review, the temperature of the catalytic system increases under light irradiation due to such activity as plasmonic localized heating and the thermal vibration of molecules. Some studies classify such reactions as photothermal co-catalysis because the elevated temperature promotes the conversion of CO2 to some extent [16,21]. However, the contribution of thermal catalysis is relatively small compared with that of photocatalysis. Thus, some researchers believe that such reactions are still essentially photocatalytic reactions or thermal-assisted photocatalytic reactions [15,20]. Regardless of the catalytic mode categorized, Ni-based catalysts do not require external heat sources during CO2 reduction.A wide variety of products, including carbon monoxide (CO), methane (CH4), and methanol (CH3OH), can be obtained because of the different reaction processes of CO2 reduction [13,22–24]. Scientists have strived to improve the selectivity for a specific product in CO2 reduction reactions to avoid the need for subsequent separation of different products. Most catalytic techniques for CO2 reduction are based on hydrogenation, such as the reverse water–gas shift (RWGS) and Sabatier reactions. In this section, we will categorize the reactions by products and review the photo/photothermal CO2 reduction mechanism.The RWGS reaction is one of the most common CO2 catalytic conversion reactions ( Eq. (1) ). (1) CO 2 + H 2 = CO + H 2 O Δ H 298 K = 41.2  ​kJ/mol The final product of the RWGS reaction is CO, a valuable product that can be used as a reducing agent to smelt metals. CO can also be transformed into other higher-value fuels such as CH3OH or certain liquid hydrocarbons through Fischer-Tropsch (FT) synthesis. Thus, some scientists classify the RWGS reaction as part of the kinetic network in CH3OH production and FT synthesis [25,26].Based on the point at which C–O bond scission occurs in the process, two possible mechanisms have been proposed for RWGS reactions: 1) a surface redox pathway, in which C–O bond scission occurs first, followed by hydrogenation of the resulting O atom to form H2O (Path 1 in Fig. 3 ); and 2) a formate-mediated pathway, in which CO2 is hydrogenated to form a carboxyl group (COOH) or converted to carbonate (CO3 2−) or bicarbonate (HCO3 −) on the surface of the catalyst and then transformed into formate (HCOO) (Path 2 in Fig. 3) [25]. In the second mechanism, CO2 hydrogenation occurs before C–O bond scission [25,27]. The type of catalyst metal plays a crucial role in determining the main reaction pathway. Dietz et al. [28] reported that metals with high affinities for oxygen (O2) promote CO2 dissociation (the surface redox pathway). As the interaction between O atoms and metals weakens, CO2 hydrogenation becomes more favorable (the formate-mediated pathway). For instance, Ni catalysts with high affinities for O2 tend to follow the surface-redox mechanism rather than the formate-mediated pathway in RWGS reactions. Conversely, metals such as Pt have poor affinity for O2, which is not conducive to lowering the CO2 dissociation barrier, and therefore the formate-mediated pathway is more advantageous.The RWGS reaction is essentially a thermal catalytic reaction that needs to be carried out under relatively stringent conditions. Its appropriate reaction temperature is 400–700 ​°C, which places high demands on energy supply [26,29]. In recent years, researchers have attempted to introduce light energy to RWGS reaction systems to develop suitable photothermal catalytic materials that enable the RWGS to proceed under a milder conditions while maintaining selectivity toward CO. For example, Jia and coworkers [30] reported Ni-NPs/N-doped CeO2 (Ni/N–CeO2) as a novel photothermal catalyst, and achieved nearly 100% selectivity for CO at around 340 ​°C due to the photothermal effect.CH4, which can be generated from CO2 through the Sabatier reaction, is widely used as fuel for heating and lighting. Despite extensive research, the mechanism of CO2 methanation is not yet fully understood. A general description is shown in Eq. (2) . (2) CO 2 + 4 H 2 = CH 4 + 2 H 2 O Δ H 298 K = - 164 kJ/mol Usually, the formation of CO is an essential step in the Sabatier reaction. According to the ways in which methyl (–CH3) groups form, three possible theories for CO2 methanation have been proposed: 1) the C atomic hydrogenation pathway, 2) the HCO hydrogenation pathway, and 3) the COH hydrogenation pathway. In the C atomic hydrogenation pathway, CO breaks down to C and O atoms, and then the C atom is further hydrogenated to –CH3 (Path 3 in Fig. 3). The HCO hydrogenation pathway starts with HCOO, which is generated as an intermediate in RWGS reactions. The first step is the dissociation of HCOO to form HCO and an O atom. In the next step, HCO is hydrogenated to CH3O and then dissociates to form –CH3 (Path 4 in Fig. 3). In the COH hydrogenation pathway, hydrogenation of COH occurs after COOH dissociation, resulting in the formation of CH3OH. CH3OH further dissociates to produce –CH3 [31] (Path 5 in Fig. 4 ).In recent years, considerable effort has been dedicated to improving the selectivity for CH4 products in CO2 reduction. Mateo and coworkers [16] reported a composite catalyst consisting of barium titanate (Ni-BTO)-supported Ni NPs, which can hydrogenate CO2 to CH4 at nearly 100% selectivity under optimal reaction conditions. Methanation catalysts in photothermal catalysis, including Ni-BTO, will be discussed in more detail in the next section.CH3OH is a basic raw material of many organic chemical processes and has significant commercial value. It is considered not only as an alternative hydrogen carrier, but also as a future energy source with a higher energy density than those of Li-ion batteries and liquefied H2 [32]. Transforming CO2 into CH3OH has become an active area of research in recent years ( Eq. (3) ). (3) CO2 + 3H2 → CH3OH ​+ ​H2O Singhal and coworkers [33] reported Ni-loaded InTaO4 that achieved a CH3OH yield of 200 ​μmol/g, which is 1.9 times higher than that of bare InTaO4, without an external heat source. However, as well as the general challenges of photothermal CO2 reduction, the CH3OH synthesis reaction suffers from relatively low product selectivity [23,32]. For example, Wu et al. [23] found out that upon introducing light irradiation to Pd/ZnO-catalyzed CO2 reduction, although the space-time yield (STY) of CH3OH was improved 1.8 fold at 250 ​°C, the CO STY was promoted more (4.2 fold). Clearly, there is still a long way to go in the development of light-driven CO2-to-CH3OH conversion.Compared with C1 products, organic compounds with two or more carbons (C2+) have higher added value because of their wider usage. According to some techno-economic analyses, currently multiple-electron transfer products, including ethanol (C2H5OH), ethylene (C2H4), and other C2+ products, have low preparation selectivity but high market prices [34].In addition to improving catalyst activity, some researchers have aimed to improve product selectivity toward C2+ hydrocarbons. Albero et al. [35] summarized the factors affecting selectivity for C2+ products. CO2-to-C2+-product conversion is a multi-electron transfer process, and materials or reaction conditions that favor e− photogeneration and storage are more favorable for C2+ product formation. Therefore, two different co-catalysts are usually used to manage the transfer of e− and h+ separately and alleviate the carrier recombination problem. Materials with good electron conductivity, such as carbon nanomaterials, also facilitate electron delocalization and prolong charge lifetime. Surface defects such as oxygen vacancies (VO) and mid-gap states can act as buffer sites of abundant local e− and act as reaction active sites. The surface plasmon resonance effect (SPR) of metal NPs can generate hot e− and promote the synthesis of C2+ products. Scientists have reported some catalysts with outstanding selectivity for C2+ products under specific conditions. For example, in 2017, Billo and coworkers [36] reported a study on black-TiO2-supported Ni-nanoclusters (Ni/TiO2[Vo]). The Ni nanoclusters and intentionally introduced VO create dual active sites for adsorption and dissociation of CO2 molecules with mixed carbon-oxygen coordination. This catalyst showed 100% selectivity toward CH3CHO under moist CO2 and a 300 ​W halogen lamp.Present knowledge of C2+ products is empirical and remains based on experimental results. Accordingly, theoretical and mechanistic understanding of the factors that influence the formation of C2+ products is poor [35]. The vast majority studies still focus on the RWGS and Sabatier reactions.Owing to the complexity of the CO2 reduction process, multiple reaction processes and various products typically coexist in such reaction systems. To more accurately understand the mechanism of photothermal CO2 reduction, infrared (IR) spectral absorption method is often used to identify the intermediate substances and products in the reaction process. In situ Fourier-transform IR (FTIR) spectroscopy is widely used in the mechanistic study of photothermal CO2 catalysis because it provides real-time information on product evolution without damaging sample structure [37]. Using in situ FTIR spectroscopy and spectroscopic analysis, Zhang et al. [15] identified intermediates such as carbonate species, formate, CO, and methyl (–CH3) in the catalytic process, which led to elucidation of the CO2 methanation pathway. Table 1 summarizes the characteristic groups and associated characteristic IR frequencies of some common substances. It should be noted that under different reaction conditions, the characteristic peaks of these substances and functional groups may slightly change due to intermolecular forces, induction effects, conjugation effects, and other factors, so the table is for reference only.Density functional theory (DFT) is used to study the electronic structure of multi-electron systems. In the field of CO2 reduction, DFT can be used to identify optimal reaction paths and determine their rate-determining steps by calculating the Gibbs free energy change (ΔG) for each reaction step [38]. For example, in many catalytic systems, the generation of adsorbed COOH is the rate-determining step, and by rational design of catalyst structure, it is possible to reduce the energy barrier for this step (or even change the rate-determining step) to facilitate CO2 reduction. Furthermore, DFT provides a mature theoretical basis for efficient catalyst design in terms of morphology and crystal plane design, which saves significant time and energy in catalyst synthesis [38,39]. For example, in a study by Yang et al. [39] Ni-MOL-100, which has abundant (100) crystal faces, was found to form stronger interactions with CO2 and be more favorable for the CO2-to-CO reduction process than Ni-MOL-010, which has (010) as the main exposed crystal face.In addition to IR spectroscopy and DFT theoretical calculations, many qualitative analytical techniques have been used for the mechanistic study of photothermal CO2 reduction. For example, temperature-programmed desorption (TPD) is used to quantitatively measure the number of active sites [40]. X-ray photoelectron spectroscopy (XPS) can be used to detect changes in the valence states of metal elements during chemical reactions [36,39]. Isotope labeling is also commonly used [15]. Thus, to realize the mechanistic study of photothermal CO2 reduction, researchers need to choose appropriate technical tools according to the actual situation.Several important parameters are involved in evaluating the photothermal CO2 reduction process, including catalytic conversion efficiency, distribution and selectivity of products, quantum yield (QY), solar energy conversion efficiency, and confirmation of the carbon source. In this section, a brief introduction of each is presented.The CO2-conversion rate is an important issue that must be considered in catalytic systems. Turnover number (TON) and turnover frequency (TOF) reflect CO2-conversion efficiency and the activity of catalytic centers. TON and TOF are typically calculated as follows ( Eqs. 4-5 ): (4) TON = n p r o d u c t n c a t a l y s t (5) TOF = n p r o d u c t n c a t a l y s t × t where n p r o d u c t and n c a t a l y s t represent the molar numbers of total or desired products and catalysts, respectively, and t is the reaction time [45,46]. The two parameters TON and TOF are widely used to evaluate the performances of catalysts, especially for metal NPs and homogeneous metal complex catalysts [46]. For instance, Chen et al. [47] prepared gigantic coordination molecules using a large number of metal ions and organic ligands, including Ni36Gd102, which exhibited good photocatalytic activity in CO2-to-CO conversion reactions. In this study, owing to the complex composition of the catalyst, the authors chose Ni2+ to represent the catalyst and calculate its TON and TOF. Based on initial CO productivity, the TON and TOF of Ni36Gd102 were calculated to be 29,700 and 1.2 s−1, respectively.As previously described, CO2 reduction can generate various products, including C1 products, such as CO and CH4, and multi-carbon products, such as C2H5OH. Accordingly, it is necessary to regulate the distribution of products in the photothermal catalysis process to improve its selectivity for specific products. Product selectivity is calculated as follows ( Eq. (6) ): (6) Product selectivity = n t a r g e t p r o d u c t s n t o t a l p r o d u c t s The C1 products are dominated by CO and CH4. Currently, 80% or even 95% selectivities for each can be achieved. There are many ways to tune product selectivity, including: 1) Selecting a suitable support. For example, under the same reaction conditions, the main product obtained using a Ni catalyst supported on TiO2 or Al2O3 is CO, while that obtained using a BTO support is CH4 [16]. 2) Adjusting the loading amount and size of the metal NPs. It has been reported that Ni NPs in the particle-size range 2–3 ​nm provide the highest CH4 yields [27]. 3) Controlling metal valence. The metal valence state is a critical factor in the conversion of CO to CH4. Under the same experimental conditions (UV/Visible light@142 ​mW/cm2), NiII and Ni0 favor the production of CO and CH4, respectively [15]. 4) Adjusting the position of metal NPs in sheet materials (discussed later) [48]. 5) Regulating the temperature of the reaction. According to Eqs. 1-2 , the Sabatier reaction is highly exothermic, while the RWGS reaction is endothermic. Thus, the low temperature region is thermodynamically favorable for CH4 production. As the temperature increases, the reaction will gradually favor the production of CO [40]. Selecting a suitable support. For example, under the same reaction conditions, the main product obtained using a Ni catalyst supported on TiO2 or Al2O3 is CO, while that obtained using a BTO support is CH4 [16].Adjusting the loading amount and size of the metal NPs. It has been reported that Ni NPs in the particle-size range 2–3 ​nm provide the highest CH4 yields [27].Controlling metal valence. The metal valence state is a critical factor in the conversion of CO to CH4. Under the same experimental conditions (UV/Visible light@142 ​mW/cm2), NiII and Ni0 favor the production of CO and CH4, respectively [15].Adjusting the position of metal NPs in sheet materials (discussed later) [48].Regulating the temperature of the reaction. According to Eqs. 1-2 , the Sabatier reaction is highly exothermic, while the RWGS reaction is endothermic. Thus, the low temperature region is thermodynamically favorable for CH4 production. As the temperature increases, the reaction will gradually favor the production of CO [40].It should be noted that, with few exceptions, most current catalysts exhibit poor selectivity for multi-carbon products under mild conditions, largely because the reaction steps and mechanisms that determine selectivity for C2+ products are not fully understood. For example, the selectivity of multi-walled carbon nanotubes supported on TiO2 for C2H5OH is 69.7% (reaction conditions: H2O/CO2, 5:1 (mol:mol), 15 ​W UV lamp @ 365 ​nm, 5 ​h) [49,50]. The selectivity of Au/TiO2 for ethane is only 27% (reaction conditions: moist CO2, Hg lamp @ 254 ​nm, 20 ​mW/cm2) [50]. Some researchers speculate that reaction conditions or materials that favor electron photogeneration and storage should be more conducive to the synthesis of C2+ products [35]. Not surprisingly, the synthesis of C2+ products requires higher electron concentrations at active sites, which makes it more challenging than the synthesis of C1 products.QY can be used to evaluate the performance of a photocatalytic or photothermal catalytic system. The overall quantum yield (OQY) and apparent quantum yield (AQY) can be calculated using ( Eqs. 7-8 ) [45,46,51]: (7) OQY ​ ( % ) = α × n p r o d u c t n p h o t o n × 100 (8) AQY ​ ( % ) = α × n p r o d u c t n p h o t o n ' × 100 where α and n p r o d u c t represent the number of electrons needed for product evolution and the number of desired products. Thus, α × n p r o d u c t is the number of electrons that participate in the reaction. For example, if the product is CH4, α is equal to 8. For the denominator, n p h o t o n and n p h o t o n ′ represent the number of absorbed and incident photons, respectively. Because only some of the incident photons can be absorbed by the catalyst, n p h o t o n is always smaller than n p h o t o n ′ , and OQY is higher than AQY [46,51].Solar energy conversion efficiency (η) is another evaluation index that should be considered in photocatalytic and photothermal catalytic systems. It can be calculated as ( Eq. (9) ) [45,51]: (9) η ( % ) = O u t p u t c h e m i c a l e n e r g y E n e r g y o i n c i d e n t s o l a r l i g h t × 100 Similar to that of general photocatalytic reactions, the efficiency of photothermal CO2 reduction largely depends on the absorption ability of the catalyst to the light source and the selectivity to the wavelength. Most current photocatalysts absorb light mainly in the ultraviolet band. For instance, Zhang and coworkers [15] investigated the effect of excitation light wavelength on the performance of Ni (10 ​wt%)–ZrO2-Reduced photocatalyst, and found that light with a wavelength lower than 320 ​nm contributed the most (79%) to CO2 conversion. Expanding the spectral utilization range of catalysts and improving the absorption and conversion ability of catalytic systems in the visible and near-IR regions are highly active areas of research [15,16,24,36]. As previously described, photothermal catalysts efficiently collect energy from the solar spectrum. Furthermore, group VIII metals show great light-harvesting abilities and can efficiently convert CO2 to CH4 [13]. Mateo et al. [16] demonstrated that Ni NPs elevate the light absorption of BTO by absorbing light in the entire visible region.In summary, it is extremely important to consider solar-energy-conversion efficiency in photothermal catalysis.Determining the source of carbon products generated during the CO2 reduction process is necessary to assess these catalytic systems. This is particularly important when using catalysts based on semiconductor carriers (especially carbon-based materials such as graphene and carbon nanotubes), where surface contaminants may play a role [45]. Thus, isotopic labeling experiments using 13CO2 under identical catalytic conditions and subsequent spectroscopic/spectrometric analyses are usually required. For example, Wu et al. [52] reported N-doped graphene quantum dots (NGQDs), which showed excellent distribution and selectivity for C2+ products. They confirmed that the carbon in the products originated from CO2 rather than the decomposition of NGQDs through isotopic 13CO2 labeling experiments. Furthermore, isotope tracing experiments can also be used to identify intermediates and reaction mechanisms during the catalytic CO2 reduction process [15].Ni is one of the most commonly used catalyst materials because of its high catalytic activity, economic feasibility, and abundance in nature [25]. Ni NPs come in various crystalline forms, among which Ni(111) has been widely studied as the most stable extended Ni surface [25,31,39,53]. Thus, the catalytic effect of Ni(111) on CO2 hydrogenation is presented here as an example.Under suitable conditions, Ni(111) can effectively catalyze RWGS reactions via the surface redox route and the formate-mediated pathway [25,31,53]. This has been confirmed from both theoretical calculations (usually DFT calculations) and experimental study. Heine et al. [53] observed the Ni(111)-catalyzed formation of CO by ambient-pressure XPS, confirming that the RWGS reaction occurred accompanied by the formation of C atoms. However, the catalytic effect of Ni(111) on CO2 methanation remains controversial. Certainly, CH4 can be formed over supported Ni catalysts. For example, Vogt et al. [54] reported the formation of CH4 over Ni catalysts supported on different metal oxides, such as TiO2, Al2O3, and ZrO2. By assessing the activity of catalysts with different supports at different temperatures, they found that the apparent activation energy from CO2 to CH4 is almost independent of the support, indicating that the reaction occurs mainly on the Ni NPs. In the case of Ni itself, most researchers support its catalytic effect on the methanation of CO2. They believe that the disproportionation step ( Eq. (10) ) is the critical process, after which atomic carbon can be further hydrogenated to CH4 [55]. (10) 2CO→C ​+ ​CO2 (g) It should be noted that Lozano-Reis et al. [31] recently reported that CH4 does not form on Ni(111) under any operating conditions. This statement was supported by accurate kinetic Monte Carlo simulations. The authors suggested that some previous DFT and microkinetic studies were based on unreliable assumptions and did not provide direct evidence for CH4 formation.Rather than using Ni alone as a catalyst, increasing research attention is being paid to combining Ni particles with specific supports or other substances, such as metal oxides and carbon-based materials, to catalyze the photothermal reduction of CO2 through the synergistic effects between substances.Most metal oxide semiconductors exhibit good photo-oxidation activity, stability, and recyclability. However, owing to their large bandgaps ( E g > 3 eV ), most metal oxide photocatalysts can only operate under UV irradiation (wavelengths shorter than 400 ​nm) and thus achieve limited use of the solar spectrum [45]. To address this issue, bandgap engineering and coloration have been proposed as steps to improve the photo/photothermal catalytic performances of metal oxides, where Ni doping with photocatalyst pre-hydrogenation is a common strategy (Fig. 4) [7,36].TiO2 was one of the first semiconductor photocatalysts to be studied and applied to water splitting and CO2 reduction. Because of its chemical stability, non-toxicity, and low cost, it has been used extensively to prepare heterogeneous photocatalysts. Although TiO2 photocatalysis research has made great progress, its inherent disadvantages, such as wide bandgap, poor light absorption, and short lifetime of photogenerated charge carriers, still limit its use [36,56]. Billo et al. [36] first introduced metallic Ni to pure TiO2, generating a disordered TiO2 surface. Owing to its altered photophysical properties, the catalyst developed a significant mid-gap state at 1.3 ​eV, which is the strongest region in the solar spectrum, significantly broadening the absorption spectrum of the catalyst (Fig. 5 a). Moreover, subsequent hydrogenation of Ni/TiO2 introduced abundant VO, which also serve as active catalytic sites. Through experimental studies as well as theoretical calculations such as DFT analysis, researchers have confirmed that the formation of Ni nanoclusters and VO in Ni/TiO2[Vo] provides more CO2 adsorption and dissociation sites. The light-harvesting ability of the catalyst is enhanced by reducing the optical bandgap and creating mid-gap states. In addition, photoluminescence (PL) and corresponding time-resolved PL measurements were performed (Fig. 5b). They revealed that hydrogenation of the Ni/TiO2 surface leads to the formation of metallic Ni0. Photoexcited electrons are transferred from TiO2 to the Ni0 surface, which inhibits the recombination of electron–hole pairs, thereby prolonging carrier lifetime and improving photocatalytic CO2 reduction efficiency. Ni/TiO2[Vo] exhibits a high product selectivity for CH3CHO, and its product yield (10 μmol/gcat), although not high, was still more than 18 times higher than the solar fuel yield of commercial TiO2 (P-25) [36].Similarly, Zhang et al. [15] pretreated Ni/ZrO2 samples under H2, showing that the valence state of the metal has a significant influence on product selectivity. As shown in Fig. 5c and d, smaller spheres/spheroids with an average particle size of 2.2 ​± ​0.8 ​nm (Fig. 5e) attached on the ZrO2 crystals can be observed, demonstrating the successful loading of Ni NPs. According to their study, the formation rate of CH4 on Ni/ZrO2 without H2 pretreatment is only 0.94% of that after pre-reduction treatment. This difference is caused by the transformation of Ni2+ into Ni0, which is consistent with other reports. Another noteworthy point is that controlled experiments to confirm the effects of light and heat on the whole catalytic process were performed. The authors found that photoexcitation is necessary for CO2 activation on the ZrO2 surface and the formation of CH4, while Ni NPs can act as heat-collecting centers to convert light energy into heat energy and increase the local temperature of the metal particles, thus providing a synergistic photothermal catalytic effect without an external heat source (Fig. 5f).Perovskites are a class of complex metal oxides with ABO3 structures. BTO, a typical perovskite material, has been shown to have photocatalytic activity for water-splitting reactions, but it is less used for photocatalytic CO2 conversion reactions. Mateo et al. [16] used a wet impregnation method to prepare Ni-BTO catalysts. The same H2/Ar reduction treatment was performed to obtain Ni NPs. During the catalytic reaction, there was no external heat source. The temperature increase was mainly due to the light radiation, the photothermal effect exhibited by the loaded Ni NPs, and the heat released by the reaction. The absorption spectrum of Ni-BTO under UV–visible radiation conditions was greatly broadened (Fig. 5g) and the CH4 yield reached 103.7 mmol/(g·h), which is higher than those of most currently reported photothermal catalysts. However, the stability of the Ni-BTO catalysts was poor. The CO2 conversion of the Ni-BTO catalyst dropped to 47% of its original value at the third use. To investigate the cause of this catalyst deactivation, the used Ni-BTO samples were subjected to X-ray photoelectron spectroscopy (XPS) analysis. The authors claimed that the reaction by-product H2O adsorbs on the active sites and reacts with Ni to form Ni(OH)2, which inhibits further methanation reactions. Thus, surface oxidation of Ni greatly reduces the catalytic performance of the reaction. To retain catalyst activity, a clean Ni surface needs to be regenerated before use. To address this issue, the catalyst was reactivated with H2 and light after each catalytic cycle, and this treatment allowed five consecutive cycles of photothermal CO2 reduction to be completed without any significant deactivation (Fig. 5h). Table 2 summarizes the performances of some recently reported Ni/metal oxide photothermal catalytic systems. It can be seen that this type of catalyst is relatively well developed in terms of selectivity for desired products.As well as metal oxides, sulfides have also garnered attention in photocatalysis research. Xu et al. [24] introduced metallic Ni into CoS2 and used the partially filled conduction band of the metal photocatalysts as an intermediate band, allowing Ni–CoS2 to satisfy the redox potential of IR-light-driven CO2 reduction. Meanwhile, the carrier CoS2 was designed to be ultrathin, thereby reducing the diffusion length of carriers and the electron–hole recombination rate. The yields of CH4 and CO in the CO2 reduction reaction catalyzed by Ni–CoS2 nanosheets were 101.8 and 37.5 μmol/(g·h), respectively, which is the highest recorded for IR-light-driven CO2 reduction among all reported single-component photocatalysts. In their experiment, a quartz tray was used instead of a liquid solvent to effectively retain the heat generated by photoinduction. No external heat source was used in their experiments. According to their report, the catalyst surface temperature increased by 26 ​°C during the CO2 reduction process, which also had a promotional effect on the catalytic process.Clearly, CO and CH4 still dominate the products of CO2 photo/photothermal catalytic reduction, and the yields of C2+ products are low. Furthermore, the photothermal effect in current metal-complex-catalyzed photothermal CO2 reductions mainly comes from light radiation and heat collection by metal NPs, rather than from an external heat source, constituting a thermally assisted photocatalytic reaction. The summary of the Ni-based catalysts presented above reveals that the following strategies are often adopted to improve the photothermal catalytic properties of Ni/metal complexes. 1) Constructing complexes. Combining Ni with a suitable support can promote the directional migration of carriers and prolong the lifetime of photogenerated e−. Ni combined with other semiconductors can also narrow the photocatalyst bandgap and improve its spectral absorption properties. The synergistic effect of Ni with supports is significant. 2) Introduction of surface defects such as VO and mid-gap states. Defects can improve the light absorption of photothermal catalysts and, in some cases, these defects can also act as reactive sites [36]. 3) Hydrogenation treatment. Hydrogenation pretreatment can adjust the valence state of Ni, which has an impact on product selectivity [15]. NiO or Ni(OH)2 may be generated in the process of CO2 reduction, decreasing the catalytic activity of the photothermal catalytic system, and the catalyst after the reaction can be treated with H2 to enhance its cyclic stability [16]. 4) Adaptation of experimental equipment to facilitate the application of the photothermal effect, such as the use of quartz disks to retain the heat generated by light induction, as described above [24]. Constructing complexes. Combining Ni with a suitable support can promote the directional migration of carriers and prolong the lifetime of photogenerated e−. Ni combined with other semiconductors can also narrow the photocatalyst bandgap and improve its spectral absorption properties. The synergistic effect of Ni with supports is significant.Introduction of surface defects such as VO and mid-gap states. Defects can improve the light absorption of photothermal catalysts and, in some cases, these defects can also act as reactive sites [36].Hydrogenation treatment. Hydrogenation pretreatment can adjust the valence state of Ni, which has an impact on product selectivity [15]. NiO or Ni(OH)2 may be generated in the process of CO2 reduction, decreasing the catalytic activity of the photothermal catalytic system, and the catalyst after the reaction can be treated with H2 to enhance its cyclic stability [16].Adaptation of experimental equipment to facilitate the application of the photothermal effect, such as the use of quartz disks to retain the heat generated by light induction, as described above [24].Because metals exist in various forms and their morphologies are very malleable, there is still much room for research on photothermal CO2 catalytic reduction using metal complexes. Thus, their application prospects are expanding.MOFs are a class of porous crystalline materials that have been developed in recent years. Their combination with Ni is advantageous for the improvement of catalysis efficiency [39]. MOFs are a class of hybrid compounds formed by assembling metal ions or metal clusters with bifunctional organic ligands. Because of their high specific surface areas and porosities, tunable pore sizes, good thermochemical stabilities, diverse topologies, and ease of functionalization, they show excellent promise in environment pollutant removal and remediation, including applications for CO2 reduction [7,60,61].During the photocatalytic reduction of CO2, photoexcited electrons in the MOFs are transferred from the highest occupied molecular orbital (HOMO) of the organic linkers to the lowest unoccupied molecular orbital (LUMO) of the metal nodes, achieving a charge-separated state that promotes the photocatalytic reaction [60]. The introduction of metal NPs to MOFs can generate hot electrons through the LSPR effect under light irradiation, resulting in local temperature increase at the metal particles and a photothermal effect, which actives the absorbed reactants and significantly enhances catalytic activity [16,17,62]. Noble-metal NPs like Pd and Ru NPs have been widely studied. Recently, more attention has been paid to transition metals such as Ni, Co, and Fe [17].Several factors influence the catalytic performance of MOFs, including: 1) Choice of Ligand. To expand the light absorption region of the catalyst and to enhance the interaction between CO2 and MOFs, polar functionalized ligands and metal ligands can be used to replace unmodified organic ligands [60,63]. For example, the ligands with polar –NH2 groups facilitate ligand-to-metal charge transfer, which alleviates photogenerated electron–hole recombination by prolonging charge separation, thereby further enhancing the efficiency of the photocatalytic reaction [60]. Furthermore, –COOH and –OH are considered to be polar groups favorable for CO2 capture [64]. 2) Choice of metal moiety. Introducing certain metals to MOF materials to lower the redox potential and promote charge transfer can further optimize the photothermal CO2 reduction process [60]. Compared with most MOF-based catalysts with single-metal active centers, Ni-MOFs exhibit high CO2-to-CO catalytic selectivity (97.7%) [65]. Furthermore, the construction of bimetallic systems by partial metal substitution opens up new directions for efficient photocatalytic CO2 reduction. Chen et al. [66] introduced Ni with high electron affinity into a Ti oxo cluster, and obtained an extended charge separation state in the Ni/Ti bimetallic MOF. This catalyst shows a larger spectral absorption range and much higher photocatalytic efficiency than those of Ti-MOFs and Ni-MOFs. 3) Surface area of the MOF catalyst. Reducing the size of MOFs catalysts and converting bulky MOFs to low-dimensional MOFs are strategies that are often employed to increase the surface areas of catalysts. As shown in Fig. 6 b–g, Yang and coworkers [39] found that ultrathin Ni-MOFs exposing abundant (010) and (100) planes both showed much higher photocatalytic CO2-to-CO activities than those of bulky Ni-MOFs (Fig. 6a). The CO yield of the (100)-rich Ni-MOFs was 11.89 ​± ​0.65 mmol/gcat under 4 ​h of photoexcitation, which was ∼4.5 times higher than that of the bulky MOFs (Fig. 6h). This is mainly due to its larger surface area and more metal sites that can bind CO2. Adjacent Ni catalytic sites also produce synergistic catalysis in the CO2 reduction process. 4) Pore properties of MOFs. High porosity means a high surface area, ensuring adequate contact between the catalytic centers and CO2 [60]. However, larger pore volumes are not always better. CO2 adsorption ability is the result of multiple factors. For example, Tran et al. [67] prepared Ni-MOF-184 and Zn-184 and compared the pore volumes and CO2 adsorption capacities of the two catalysts. They found that, although Ni-MOF has a smaller pore volume of 1.10 ​cm3/g, its CO2 uptake was 71 ​cm3/g, i.e., 1.65 times that of Zn-MOF-184, which has a larger pore volume (1.12 ​cm3/g). Choice of Ligand. To expand the light absorption region of the catalyst and to enhance the interaction between CO2 and MOFs, polar functionalized ligands and metal ligands can be used to replace unmodified organic ligands [60,63]. For example, the ligands with polar –NH2 groups facilitate ligand-to-metal charge transfer, which alleviates photogenerated electron–hole recombination by prolonging charge separation, thereby further enhancing the efficiency of the photocatalytic reaction [60]. Furthermore, –COOH and –OH are considered to be polar groups favorable for CO2 capture [64].Choice of metal moiety. Introducing certain metals to MOF materials to lower the redox potential and promote charge transfer can further optimize the photothermal CO2 reduction process [60]. Compared with most MOF-based catalysts with single-metal active centers, Ni-MOFs exhibit high CO2-to-CO catalytic selectivity (97.7%) [65]. Furthermore, the construction of bimetallic systems by partial metal substitution opens up new directions for efficient photocatalytic CO2 reduction. Chen et al. [66] introduced Ni with high electron affinity into a Ti oxo cluster, and obtained an extended charge separation state in the Ni/Ti bimetallic MOF. This catalyst shows a larger spectral absorption range and much higher photocatalytic efficiency than those of Ti-MOFs and Ni-MOFs.Surface area of the MOF catalyst. Reducing the size of MOFs catalysts and converting bulky MOFs to low-dimensional MOFs are strategies that are often employed to increase the surface areas of catalysts. As shown in Fig. 6 b–g, Yang and coworkers [39] found that ultrathin Ni-MOFs exposing abundant (010) and (100) planes both showed much higher photocatalytic CO2-to-CO activities than those of bulky Ni-MOFs (Fig. 6a). The CO yield of the (100)-rich Ni-MOFs was 11.89 ​± ​0.65 mmol/gcat under 4 ​h of photoexcitation, which was ∼4.5 times higher than that of the bulky MOFs (Fig. 6h). This is mainly due to its larger surface area and more metal sites that can bind CO2. Adjacent Ni catalytic sites also produce synergistic catalysis in the CO2 reduction process.Pore properties of MOFs. High porosity means a high surface area, ensuring adequate contact between the catalytic centers and CO2 [60]. However, larger pore volumes are not always better. CO2 adsorption ability is the result of multiple factors. For example, Tran et al. [67] prepared Ni-MOF-184 and Zn-184 and compared the pore volumes and CO2 adsorption capacities of the two catalysts. They found that, although Ni-MOF has a smaller pore volume of 1.10 ​cm3/g, its CO2 uptake was 71 ​cm3/g, i.e., 1.65 times that of Zn-MOF-184, which has a larger pore volume (1.12 ​cm3/g).In addition to pure MOFs catalysts, many MOF-derived materials have been investigated for catalytic CO2 reduction. For instance, Khan et al. [17] reported an efficient MOF-derived Ni-based catalyst for photothermal CO2-to-CH4 conversion (Fig. 6i). They first synthesized Ni-MOF-74, and then pyrolyzed this MOF material at different temperatures under continuous N2 flow to modulate the properties of the resulting carbonaceous species. Fig. 6j shows the homogeneous distribution of Ni NPs throughout the carbon matrix. Because higher pyrolysis temperatures led to a greater degree of graphitization of the catalytic material, the MOF-derived carbon-based material obtained at a pyrolysis temperature of 600 ​°C exhibited the highest CH4 production rate (448 mmol/(g·h)). This production rate is also the highest among all reported CO2 photo-methanation catalysts. The temperature and pressure changes during the reaction are shown in Fig. 6k. Under light irradiation, all the samples exhibit a progressive increase in temperature during CO2 reduction, while the pressure shows the opposite trend due to the consumption of gas reactants, confirming the occurrence of the photothermal reaction.In summary, the excellent properties of Ni-based MOF materials, such as large surface area, good structural controllability, and low photogenerated electron–hole recombination give them high application potential for photo/photothermal CO2 catalytic reduction. Ni-based MOF catalysts will undoubtedly make a greater contribution to CO2 reduction with the advent of rational structural design.Metal-free materials are widely used in aerospace, electronics, engineering machinery, and other fields owing to their low specific gravity and high strength [52,68,69]. In CO2 catalytic reduction, metal-free materials have been extensively researched owing to their abundance in nature, good electrical conductivity, environmental friendliness, and low cost [70]. This section will focus on graphene and graphitic carbon nitride (g-C3N4), which are frequently used in CO2 photo/photothermal catalytic reduction systems. Although graphene and g-C3N4 share the same 2D layered structure, they have very different electrical properties and thus play different roles in CO2 catalytic reduction systems.Graphene is a 2D material with a honeycomb lattice structure connected by sp2 hybridization [70,71]. In the internal structure of graphene, each carbon atom has four valence electrons, three of which form sp2-bonds, leaving one unbonded electron in the Pz orbital. Similar to the electronic configuration of the benzene ring, the Pz orbital of each carbon atom perpendicular to the layer plane can form π-bonds with multiple atoms throughout the layer. Such a structure enables graphene to have excellent electrical and optical properties as a zero-bandgap semiconductor. It is worth noting that graphene can act as both an acceptor and transporter of electrons, prolonging the separated state of electron–hole pairs and the lifetimes of charge carriers, thereby promoting multi-electron reactions. Furthermore, graphene can improve photostability and promote CO2 adsorption by catalysts [70]. Lin and coworkers [72] prepared a photocatalyst with a magnetic hollow structure consisting of metallic Ni NPs surrounded by few-aof graphene (Ni@GC) (Fig. 7 a–d). Ni@GC are hollow spheres with large surface areas and highly porous structures that accelerate the separation and transport of photoexcited charge carriers. Because of the synergistic contribution of Ni NPs with high electron density and graphene with porous structure, this catalyst exhibited a high CO2 adsorption capacity of ∼28 ​cm3/g for CO2 at 273 ​K (Fig. 7e and f).g-C3N4 is a typical polymeric semiconductor in which the CN atoms in the structure are sp2 hybridized to form a highly exotic π-conjugated system. The bandgap of g-C3N4 is ∼2.7 ​eV, which is sufficient to absorb blue-violet light with wavelengths less than 475 ​nm in the solar spectrum. This photoresponse range is already larger than those of most photocatalysts [70]. In the absence of a co-catalyst, g-C3N4 has a good catalytic effect on CO2 reduction. Since 2012, when researchers demonstrated that g-C3N4 could photoreduce CO2 to CO in the presence of water vapor, this material has become very actively researched in photocatalysis [73]. By using an active unsaturated edge confinement strategy, Cheng et al. [74] recently synthesized few-layer porous g-C3N4 photocatalysts with single Ni atoms as anchoring points for CO2 reduction (Fig. 7g). The detailed few-layer morphology and porous structure can be seen in the TEM and STEM images of the Ni5–CN sample (Fig. 7h and i). The graphitic π-conjugated layer structure, anisotropic structure, and defective vacancy self-modification of porous ultra-thin g-C3N4 materials contribute to the performance enhancement of this catalyst. The porous structure provides vacancy ligands for trapping single Ni atoms, realizing a high density of single-atom active sites, which has significant advantages for improving the CO2 adsorption capacity of the catalyst. Because of the strong chemisorption and carrier dynamics, CO2 is converted from the gaseous to the adsorbed state by releasing absorption heat (QP and QC) (Fig. 7 j). Upon the introduction of single-atomic-site Ni, it strongly binds to CO2 molecules as adsorption active sites with the appropriate activation energy (Ea) (Fig. 7 k). Ea is lower than the dissociation energy (Qs), so Ni rapidly binds with CO2 and promotes carrier transfer. The highly unsaturated Ni–N coordination maximizes the formation of photocatalytic sites. According to this study, the CO yield of this catalyst is 7.8 times higher than that of pure g-C3N4 under visible-light excitation.Besides the two metal-free materials mentioned above, many other carbon-containing materials also exist, such as carbon nanotubes (CNTs) and covalent organic frameworks, and they have been widely applied in CO2 photo/photothermal catalysis [35,75]. However, low surface area, unstable carrier dynamics, and other drawbacks of all-metal-free materials hinder their wider application. Thus, to further enhance their catalytic efficiencies, a series of strategies, such as doping, morphology control, and surface modification have been adopted. Combining metal-free materials with metallic Ni has also been extensively studied.1) Elemental doping. In heteroatom doping certain carbon atoms in the graphite structure are replaced with other atoms, thus changing the spin density and charge distribution of the carbon material, modulating its adsorption capacity for the reactants and expanding the light-response range of the catalyst. This method can significantly modulate the optical and electronic properties of the material. The dopants can be classified into nonmetal atoms (e.g., N, O, and S) and metal atoms (e.g., Ni, Co, and Zn) [26]. This review mainly introduces metal-atom doping [71].Doping of metal-free materials with metal atoms can create impurity levels in the semiconductor bandgap that act as electron traps when the semiconductor is excited, inhibiting the recombination of photoexcited electrons and holes [70]. For example, metal-doped CNTs exhibit better adsorption capacities for CO2 than normal CNTs and carbon nanocages. Furthermore, CO2 adsorption capacity and catalytic effect vary with the metal species. The transition metal Ni has been reported to show good performance. Xu et al. [75] conducted a density flooding theory study on the reaction process of CO2 reduction to CH3OH on the surface of Ni-doped CNTs. The results showed that the Ni-doped carbon nanotubes exhibit lower overpotential values and higher reaction energies.2) Morphology control. The sizes, shapes, and geometric features of catalytic materials greatly influence their photocatalytic performances. Morphology control can facilitate carrier migration and promote the surface reactivity of such materials [70]. For example, graphene and g-C3N4 are typical 2D materials, which have larger surface areas and extensive π-conjugation compared with those of 3D materials. Such morphology effectively promotes the flow of charge carriers. The preparation method of the catalyst plays a vital role in morphological control. For instance, solvent and thermal exfoliation are commonly used to increase specific surface area, and template methods are employed to introduce porous structures [76].In conclusion, using strategies such as elemental doping and morphological tuning can improve the light-harvesting capacities of metal-free materials and reduce photogenerated-charge-carrier recombination rates. The combination of Ni with metal-free materials such as g-C3N4 and graphene by doping or loading can make full use of their morphologies to increase their contact areas with reactants and promote light absorption. Furthermore, in some catalytic systems, Ni can act as a heat-collecting center to increase temperature and is also an active site for CO2 adsorption and conversion. Thus, the effect of the combination of Ni and metal-free materials is greater than the summed effects of the two parts.Ni exists in many other forms and can have numerous loading supports. In this section, some other Ni-based materials for CO2 photo/photothermal catalysis will be introduced.Si is the second most abundant element on earth, especially in the form of SiO2, and has been extensively studied as a carrier for photocatalysts. However, Ni/SiO2 lacks specific products selectivity [77]. Yan et al. [48] prepared siloxane nanosheets loaded with Ni NPs by employing 2D silicon surface chemistry. They used ethanol and water as dispersants for the Ni source and prepared two samples, labeled Ni@SiXNS-EtOH and Ni@SiXNS-H2O. The two samples were found to have different structures, and this structural difference affects the pathway of CO2 reduction. In Ni@SiXNS-EtOH, most Ni NPs are sandwiched between the SiXNS, and very few Ni NPs are immobilized on the surface (Fig. 8 b), which makes the formation of bridging ∗CO difficult and limits the generation of CO. Therefore, CH4 is formed mainly through the HCOO pathway (Fig. 8d). Under the reaction conditions of light irradiation at 300 ​°C, the CH4 yield reached 100 mmol/(gNi·h) and the selectivity for CH4 was ∼90%. However, in Ni@SiXNS-H2O, the Ni NPs are mainly present on the surface of the nanosheets (Fig. 8a), and the main reaction paths were: C–O bond breakage to form O and ∗CO, ∗CO transformation from the adsorbed state to the gaseous state as the main product, and further dissociation of some ∗CO to ∗C, followed by hydrogenation to form CH4 (Fig. 8c). A comparison between the two and Ni@SiO2 is given in Table 3 . This study demonstrates once again that the choice of catalyst material and preparation method significantly influences the reaction path and product selectivity for CO2 reduction.In the CO2 reduction reaction, competition with the hydrogen evolution reaction (HER) exists. Thermodynamically, proton reduction occurs more easily than CO2 reduction. Kinetically, CO2 reduction is a multi-electron reaction, while hydrogen precipitation reaction is a two-electron reaction [78]. Therefore, inhibiting hydrogen evolution during CO2 reduction is also a major challenge. Transition-metal complexes are considered good candidates for CO2 reduction because of their multiple redox states, which facilitates electron/proton transfer. Lin et al. [79] designed a simple heterogeneous photocatalytic system consisting of a Ni bipyridine complex (Ni(bpy)3Cl2) and cadmium sulfide (CdS). CdS readily catalyzes the generation of H2 on the hydrogenated surface. The introduction of Ni(bpy)3Cl2 promotes the transfer of photogenerated electrons from CdS to CO2, further improving selectivity for the CO2 reduction reaction. Under the same reaction conditions, the CO yield for the complex system containing Ni(bpy)3Cl2 was ∼6 times that for the CdS-catalyzed system.More Ni-based catalysts are listed in Table 4 [80–87]. The products of the catalytic system using H2 as the proton source are more diverse, and all are carbon-containing substances at various selectivities. In contrast, CH4 can also be used as a proton source. This type of reaction is called CO2 reduction with methane (CRM), and the products are mainly CO and H2. Selectivity for the highly utilizable product H2 can reach ∼45%. The catalysts for CRM are often prepared by constructing bimetallic alloys on suitable carriers, and Ni is one of the most promising metals here. CRM has gained much attention owing to the possibility of achieving both high fuel productivity and light-to-fuel efficiency (η) [87].This chapter briefly introduces and lists some Ni-based catalysts combined with metal complexes, metal-free materials, and MOFs. The Ni-based catalysts do not generate CO2 directly during the synthesis, but generate carbon emissions indirectly from the electrical energy consumption when using related equipment. Therefore, photothermal catalysts with good catalytic activities and long service lives are required to reduce the energy consumption and carbon emissions generated during preparation. However, photothermal CO2 catalytic reaction systems are subject to low solar energy utilization, severe photogenerated e−/h+ pair recombination, competition from the HER, and low product selectivity [36,45,77,78]. In this section, strategies to improve the catalytic performances of photothermal catalysts, such as the selection of suitable carriers or complexes, morphology modification, and elemental doping, are presented. In the practical application of Ni-based catalysts, cost and cycling stability are influencing factors. Therefore, a composite materials’ cost and availability should be considered. The cycle stability of photothermal catalysts can be enhanced by operations such as hydrogenation treatment and reasonable light illumination [16], as well as ensuring a sufficient supply of reaction feedstock (e.g., CO2, H2, or H2O).Photo/photothermal catalytic reduction of CO2 is a potentially viable strategy to solve global warming and the energy crisis. In this review, three modes of photothermal catalysis were introduced: photo-assisted thermocatalysis, thermal-assisted photocatalysis, and photothermal co-catalysis. CO2 reduction is a complex process, and parameters such as CO2 conversion, product yield, product selectivity, and QY are often used to judge the effectiveness of a particular photo/photothermal catalytic system. However, although the products of CO2 reduction as a multi-electron transfer process are very diverse, the most common are C1 products, including CO and CH4. Currently reported photothermal catalysts are still less selective for other products.Ni is a potential candidate for CO2 photo/photothermal catalysis owing to its availability, low cost, and, more importantly, its proven photocatalytic activity. When Ni is present in the form of NPs, the local temperature around the metal particles increases under light illumination owing to the surface plasmon resonance effect, achieving the effect of synergistic photothermal catalysis. Combining Ni with suitable supports or materials to take full advantage of the synergistic effect can significantly improve catalytic CO2 reduction performance. Combination with metal oxides and MOFs results in good photothermal catalytic performance, with product yields reaching the mmol/(g·h) level under optimal conditions, e.g., 103.7 and 448 mmol/(g·h) for CH4 in BTO and Ni–C-600 photothermal catalytic systems, respectively. In addition, strategies such as elemental doping, morphology control, and pore adjustment can effectively solve some common problems in photothermal CO2 reduction. Medium-sized, uniformly distributed NPs and sheet-like materials with large surface areas always lead to better CO2 reduction performance. Although photo/photothermal CO2 reduction has been extensively researched, corresponding catalyst research is still in the developmental stage, and the following challenges remain to be overcome. They are listed here as a reference for researchers to conduct follow-up studies (Fig. 9 ).First, although Ni-based catalysts play an important role in the photothermal reduction of CO2, their development is still limited by various factors, such as: how to choose a suitable support, how to achieve a stable connection between the support and Ni NPs, the reduction method of Ni precursor solution and the catalyst deactivation brought about by Ni0 oxidation during the catalytic process. Solutions to the above problems will greatly advance the application of Ni-based catalysts in the field of CO2 reduction.Second, to efficiently and rationally design Ni-based catalysts, it is essential to thoroughly understand the mechanisms of the different reaction pathways possible for CO2 reduction. Although this review presents several widely accepted reaction pathways, a fully unified mechanism supported by experimental evidence and theory remains elusive. Using traditional characterization techniques with photothermal catalytic systems, isotope tracing techniques, and theoretical calculation (e.g., DFT) to thoroughly explore these catalytic reaction pathways will provide valuable information required for the modification and development of photothermal catalysts.Third, achieving C–C coupling to obtain high-value-added C2+ products (e.g., C2H5OH and olefins) will surely emerge as a focus of future research. Current studies on the formation of C2+ products are based on experimental results, and there is a lack of systematic theoretical studies on their formation factors. Thus, improving the selectivity for C2+ products remain a challenge. Meanwhile, various in situ characterization techniques should also be fully utilized to provide directions for the rational structural design of catalysts. Furthermore, the sensitivity of C2+-product detection should be enhanced by the development of related instruments.Fourth, the practical application and large-scale production of photothermal catalysts for catalytic CO2 reduction must be considered, not just that at the laboratory stage. The photothermal catalysts currently reported show good catalytic performances under ideal conditions, such as high CO2 concentration and artificial-light excitation. However, the application of such catalysts is still limited in practical situations with low atmospheric CO2 concentrations or for industrial waste gases with complex compositions [89]. Therefore, designing cost-effective catalysts with strong adsorption and high selectivity for CO2 must also be a major endeavor. Strategies that can be used here include adjusting pore structure, expanding specific surface area, and making full use of the synergistic effect between catalyst materials.Finally, carbon capture, utilization, and storage (CCUS) technology has received widespread attention worldwide as a means to combat severe global climate change. CO2 reduction is one method of carbon utilization. It is worth considering how this process might be combined with CCUS, i.e., capturing and purifying the CO2 emitted during the production process and then carrying out the reduction operation.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the National Natural Science Foundation of China (No. 21906056), Open Project of State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology (No. ESK202104), the Science and Technology Commission of Shanghai Municipality (22ZR1418600), Shanghai Municipal Science and Technology (No. 20DZ2250400).

Converting CO2 to fuel is a promising strategy to mitigate the greenhouse effect and achieve ‘carbon neutrality’. Photothermal catalysis has been widely used for CO2 reduction because it effectively reduces the apparent activation energy of the reaction and provides milder catalytic conditions as well as higher catalytic efficiency than conventional catalytic methods. In this review, the basic principles of photothermal catalytic CO2 reduction and the factors used to evaluate photothermal catalytic conversion efficiency are introduced. Then, the common types of Ni-based catalysts and their design strategies are summarized and discussed. Among these catalysts, metal oxides have been extensively studied and developed. Accordingly, they currently achieve product yields up to the mmol/(g·h) level. Strategies such as elemental doping and morphology control are often adopted for the modification of photothermal catalysts as a means to improve catalytic performance. Finally, future trends in the field of photothermal catalytic CO2 reduction are proposed, including mechanistic studies, practical applications, and coupling with other carbon-neutral technologies.

No data was used for the research described in the article.Biopolymers are gaining traction in plastics because of their inherent biodegradability and unique characteristics for particular applications. Biopolymers may be taken from nature, biosynthesized by live organisms, or chemically synthesized from biological components. Biopolymer waste disposal is now one of the most significant worldwide issues confronting humankind and ecological balance. As the world's population continues to rise, so does the need for the polymer to meet the rising consumption levels, resulting in a severe environmental problem caused by the buildup of biopolymer waste. Biopolymer pyrolysis, a well-known technique for producing usable liquid fuels from low-value polymeric wastes, has fewer greenhouse gas emissions than other technologies, such as incineration and gasification. Bio-based polymers may be made from a variety of renewable sources. Plant-based precursors have made bio-based polymers, including lignocellulose fibers, cellulose esters, polylactic acid, and polyhydroxyalkanoates [1]. One of the abundant biopolymers is cellulose, produced from many living organisms, including plants, animals, bacteria, and certain amoebas. Organic materials/precursors with high amounts of cellulose and other fibers are chosen because they improve their mechanical intensity [2]. Food and agriculture waste are an appealing source of cellulose for industrial applications since it does not endanger the food supply and boosts the local economy [3,4].An interesting alternative to the direct treatment of cellulose is the cracking of solutions where the cellulose is dissolved in a solvent. The influence of the solvent during the thermal degradation of polymers is significant. A proper solvent medium must be used to break the plastic waste into low molecular-weight products. Solvents that may donate hydrogen, in particular, participate in the thermal breakdown of polymers, which impacts the generation and dispersion of hydrocarbons [5]. One of The polar functional groups allows plastics to be solvated by polar solvents like carbolic acid (phenol). The phenol molecule comprises two hydroxyl (−OH) groups attached to a phenyl group (−C6H5), making it an aromatic organic chemical that is volatile. Although most microorganisms are poisonous to phenol, it is often found in many industrial effluents and is frequently utilized as broad-spectrum disinfection [6,7]. Therefore, the presence of phenolic compounds in aquatic environments is unpleasant and unwanted and dangerous to animal and human health [8]. Phenolic constituents often result from the production of petrochemical by-products [9] and makeup around 38% of the unwanted pyrolysis oil ingredient [10]. Phenolic compounds can be extracted from bio-oil by a method such as the liquid-liquid extraction technique because its presence reduces the bio-oil quality and causes high viscosity, high acidity, corrosiveness, low heating value, and faulty product that harms machinery [11]. Thus, it is environmentally favorable to utilize the phenolic compounds as a cellulose solvent using an appropriate method not only for liquid fuel generation but also for carbon-free gas fuel (hydrogen) production due to hydrogen bonds.Polymer pyrolysis is a well-known process for producing valuable liquid fuels and has fewer greenhouse gas net emissions than other contemporary technologies like incineration and gasification and is a valuable method for chemical recycling, which lowers the carbon footprint of polymers. It can mitigate the adverse environmental effects of current management practices via landfilling and incineration and partially reduce carbon emissions while manufacturing virgin polymers. As we used a hydrocarbon solvent with six hydrogen atoms, hydrogen can also be produced, which significantly enhances the yield of generated H2 gas during the reaction. A few technologies have also been employed for H2 generation from bio-oil derivatives, such as dry reformation, partial oxidation, and auto-thermal reformation. Steam reforming is the most efficient and practical technique for producing hydrogen from hydrocarbons [12,13]. Compared to traditional reforming processes, the steam reforming reaction might well be conducted at significantly lower reaction temperatures, lowering the risk of catalyst carbonization and sintering and capital and operating costs. Additionally, most of the heat needed for the endothermic reforming processes is provided by the heat emitted by the exothermic carbonation reaction. Therefore, we conducted the in-situ catalytic steam reforming of phenol coupled with cellulose thermal cracking (or pyrolysis). Phenol has been used in many of previous research as a source for hydrogen production [9,14,15]. However, a significant obstacle to phenol steam reforming is the endothermic nature of the process, which has a complex of numerous side reactions, including phenol breakdown, which produces carbon dioxide, carbon monoxide, and most significantly and negatively, coke [16]. This issue can be solved by developing suitable, active, and stable nano-sized catalytic materials for the pyrolysis-catalytic steam reforming reaction.A catalyst is any chemical compound that reduces the activation energy to speed up chemical reactions like steam reforming and thermal cracking without being wasted during the reaction. Various transition metals such as nickel [17], cobalt [18], lanthanum [19], molybdenum [20] and tungsten [21] and noble metals like rhodium [22], platinum [23], ruthenium [24], palladium [25] and supports such as alumina [26], titanium [27], calcium [28] and etc have been studied to produce hydrogen from reforming reaction of various feedstock. The availability, affordability, chemical safety, and stability of a reducible metal oxide like titanium dioxide (TiO2) make it exceptional support [29,30]. However, it experiences coke formation, which has a negative impact on its long-term H2 generation sustainability [14]. The CaO may be put on the appropriate support to get around this restriction and improve stability [31]. Additionally, the CaO alone suffers from cyclic instability and severe attrition loss in the reaction due to its weak mechanical strength. The mechanical strength and cyclic stability of CaO materials can also be improved by introducing TiO2 to ensure the sustainability of H2 production. Due to their superior physicochemical characteristics, we discovered in our earlier study [31] that equal ratios of Ti and Ca showed bifunctional capabilities, had both basic and acid phases, and had a variety of impacts on the catalyst activity in the transesterification process. The carrying capabilities of both heterogeneous and homogeneous catalysts allow transition metal catalysts to be employed as hybrid catalysts [32]. Because of their low price especially in comparison to catalysts like Rh, Ru, or Pt, as well as their effectiveness for C–O, C–H, and C–C bond breakage and water gas shift reaction, which was caused by the high Lewis acidity intensity of nickel metal, nickel-based catalysts have been explored more in-depth for the removal of tar [33,34]. However, the supported mono-metallic nickel (Ni) catalyst often experiences quick deactivations brought on by the sintering of Ni nanoparticles (NPs) and the accumulation of coke [35]. Adding a second active metal to a bimetallic catalyst uses the synergy between the active metals, often transition metals, to increase coke resistance and active phase dispersion. We have illustrated that this issue of nickel can be solved by introducing another transition metal in the catalyst with great resistance to carbon [16,36]. The enhanced coking resistance of the bimetallic and trimetallic catalysts may be attributed to changing the electronic structure of the catalyst. The advantages of lanthanum (La2O3) as promoters were determined by steam reforming of the bio-aqueous oil's component [18]. La2O3 is intended to increase the distribution of active metal particles on the support and decrease the agglomeration of such materials throughout reforming. Additionally, because of the increased mobility of lattice oxygen anions, it may reduce the formation of coke [19]. La2O3 enhances the catalyst's thermal stability and modifies the acidic and basic characteristics of compounds [37]. Ni and La are examples of transition bimetallic NPs with large surface area and energy, making them effective catalysts. In a bimetallic Ni-La catalyst, the strong oxygen affinity of La promotes carbon oxidation and minimizes coking, while hydrogen overflow from Ni to La limits its oxidation. Despite the encouraging results for bimetallic catalysts [10,25,27,35], the catalyst's stability for extended periods of time in the stream during cellulose cracking at high temperatures has to be clarified. Additionally, the overall surface area and activity of these catalysts are significantly impacted by the coking and sintering of relatively large metal particles [38], particularly when it comes to the heat breaking of polymer bonds. Therefore, it is thought to be desirable to promote the bimetallic active transition metals by a little quantity of noble metal in order to profit from their higher coking resistance and stability while reducing the problem of their high cost and scarcity [39]. The hydrogen spillover mechanism, which accounts for the lowering of the reduction temperature by noble metal promotion, states that as hydrogen adsorbs and dissociates on noble metals, the adsorbed hydrogen atoms disperse on the support surface to reach non-noble metal species, boosting their reduction. Additionally, morphology and composition of the noble metals catalysts are crucial factors that affect their catalytic activity and stability. Since they have the capacity to dramatically alter the catalytical structure and effect performance, their partial application is economically feasible even at low concentrations (∼1 wt%) [40]. When Pd NPs had rough surfaces, dendritic topologies, or porous architectures, for instance, they outperformed their compact counterparts in terms of catalytic efficacy [41]. To improve the catalytic performance, the porous structure may provide a significant specific area, many exposed active sites, and an effective diffusion channel for molecules and electrons. As Pd promoters have attracted great interest, we also targeted the in situ hydrogen production reaction by employing a small amount of Pd in our previous work with remarkable increase in the catalytic activity [25,42]. Pd promoters are good substrate components for customized multimetallic catalysts for hydrogen generation provided the control mechanisms learned, and their increased thermal stability is boosted by a strong metal-support interactions.Despite the evidence disclosed in previous research regarding trimetallic catalysts, there is still a lack of investigations on developing trimetallic nanosized catalysts to crack and reform the cellulose bonds and phenol compound to liquid fuels and hydrogen gas in the scientific literature. To increase the feasibility of this study and increase the ratio of the cellulose to solvent to 2:8, which is much higher compared to our previous works [43–45], we modified the experimental setup and connected a Parr Benchtop Reactor (PBR) (see Fig. 1). This modification allows the easily liquefying and increasing of the amount of polymer and plastic waste in the reaction without causing line blockage. The novelty of the work also lies in developing an understanding of the role of the trimetallic Ni-La-Pd catalyst supported on TiCa for hydrogen production and liquid fuel generation from cellulose dissolved in phenol in the unique process conditions of the in-situ pyrolysis-catalytic steam reforming. Four catalysts were synthesized by hydrothermal treatment methods followed by conventional impregnations method and named as TiCa (ratio 1:1), N/TiCa (Ni to TiCa ratio is 1:9), NL/TiCa (Ni:0.7, La:0.3, TiCa:9), and NLP/TiCa (Ni:0.6, La:0.25, Pd:0.15, TiCa:9) nanocatalysts. The physicochemical characteristics of the fresh catalyst were examined XRD, BET, N2 adsorption-desorption isotherm, NH3-TPD, IR-Pyridine, IR-Pyrrole, H2-TPR, CO2-TPD, FTIR-KBr, TEM, EDX, HRTEM, SAED, Elemental mapping analysis, and ICP test. Catalysts were screened at 600 °C, the optimum catalyst was tested at 500–800 °C, and stability was studied for 72 h on stream. GC-TCD characterized gaseous products, and liquid fuels were also examined by GC/MS, GC-FID, and FTIR. The catalysts were analyzed by TGA-DTG, BET, N2 adsorption-desorption isotherm, TEM, FTIR-KBr, and CHNS.The nanosized TiCa support hydrothermal followed by impregnation route and followed our published works [31,44]. A particular amount of CaO and TiO2 were combined with 100 mL of deionized water for dilution purposes and with the 1:10 mass ratio and agitated for a couple of hours, separately in two separate beakers, after 5 M of sodium hydroxide (NaOH) had been dissolved and mixed for an hour. The NaOH was employed to improve the nucleation and growth rates of the NPs [46]. A 100 mL Teflon container was used to contain the fluid. Then, an autoclave made of stainless steel was firmly sealed and housed two Teflon bottles filled with these solutions. The autoclave was put into a temperature-controlled electric oven and hydrothermally treated for 48 h at 160 °C. The treated sample was filtered and washed with distilled water several times and dried at 110 °C overnight. The prepared TiCa sample was calcined in an oven (Model Ney Vulcan D-130) at 800 °C for 3 h. The synthesis steps for each catalyst are depicted in Fig. S1.The monometallic N/TiCa nanocatalyst was prepared in the same steps as the TiCa nanocatalyst. Firstly, a specific amount of nickel nitrate hexahydrate and then 5 M NaOH were dissolved in 100 mL deionized water at room temperature for an hour until a homogenous and clear solution appeared. Subsequently, the solution was transferred into an autoclave reactor equipped with a 100-mL Teflon cylinder and kept in the oven for 48 h at 160 °C. The powder containing Ni NPs was then washed via filter paper on a Buchner funnel that was sealed with a rubber bung on the top of a side arm conical flask. The side arm of the flask was connected with a vacuum pump to speed the filtration and washing process of the samples, followed by drying at 110 °C overnight and then calcination for three hours at 800 °C. The Ni NPs supported on TiCa nanosized support with a volume ratio of 1–9 were prepared via impregnation. Briefly, after gently adding the prepared TiCa material into 150 mL of a beaker filled with deionized water, the Ni powder was introduced to the solution and kept stirring on a hot plate stirrer at 90 °C until water vaporized. The slurry was dried at 110 °C overnight and then calcination for three hours at 800 °C. The same procedure was employed for synthesizing La NPs by hydrothermal treatment method and then impregnating La into N/TiCa to produce a bimetallic NL/TiCa nano catalyst. Trimetallic NLP/TiCa nanocatalyst (Ni:La:Pd:TiCa ratio is 0.6:0.25:0.15:9) was prepared in the same way as the monometallic and bimetallic ones, replacing the TiCa support by a bimetallic NL/TiCa nanocatalyst.The crystalline structure of the catalysts was characterized by X-ray diffraction (XRD) conducted on a D8 ADVANCE Bruker diffractometer equipped with Cu Kα radiation (λ = 0.154 nm, Philip), 40 kV and 30 mA. The Fourier-transform infrared (FTIR) spectra (from 4000 to 400 cm−1) were collected on a Shimadzu IR-Prestige-21 spectrometer to examine functional groups in the synthesized and used catalysts. Before measurement, the samples were diluted with potassium bromide (KBr) and pressed into pellets. The KBr pellet was prepared by mixing KBr and catalyst (1 (mg):100 (mg)), and the excellently designed combination was pressed to procedure a 13 mm diameter pellet. N2 adsorption-desorption isotherms of the fresh and used samples were obtained at − 196.1 ºC over the whole range of relative pressures using a Beckman Coulter SA3100™ instrument. Before N2 adsorption-desorption measurements, samples were degassed at 180 °C in a vacuum for 12 h. Specific surface areas (SBET) of the fresh and used samples were calculated by the Brunauer-Emmet-Teller (BET) equation, considering the range of relative pressures 0.1 < P/Po < 0.3. Barrett-Jouner-Halenda (BJH) technique was used to compute pore size and volume from the desorption branch of the isotherm, and the BET method was used to estimate surface area. The elemental composition of the catalysts was detected using inductively coupled plasma (ICP-test) on the Agilent ICPOES720. A JEOL JEM-ARM200F apparatus operating at 200 kV was used to capture the samples' transmission electron microscope (TEM) and high-resolution transmission electron microscopy (HRTEM) micrographs. The same instrument was employed for the Energy Dispersive X-Ray (EDX) elemental analysis and crystallographic experimental analysis by the selected area diffraction (SEAD) technique. Temperature-programmed reduction (TPR) scans were performed to study the catalyst's reducibility in a Micromeritics Chemisorb 2720 apparatus. Approximately 100 mg of the precursor material and a flow of 20 mL/min of pure hydrogen, with a 20 °C/min heating rate, were used for the tests. Using the Micromeritics Chemisorb 2720 apparatus, temperature-programmed desorption of ammonia (NH3-TPD) was carried out to investigate acidity. The material was pre-treated in He flow at 200 °C for 30 min before being admitted with ammonia. Following cooling to room temperature, the sample was exposed to a stream of pure NH3 (20 mL/min) for 30 min. The sample was purged in flowing He (20 mL/min), and the temperature of the catalytic sample was then raised to 900 °C (T = 20 °C/min), removing the physically adsorbed ammonia. The surface acidity and the evaluation of the catalysts' protonic and Lewis acid sites and the supports were also investigated by means of an FTIR spectroscopic study of adsorbed pyridine as a probe molecule. Pyridine (2 Torr) was first adsorbed for 30 min at 423 K, then released for the same amount of time at 500 °C. The analysis was done using a Cary 640 FTIR spectrometer (Agilent, Selangor, Malaysia) with CaF2 windows and a stainless steel cell that can withstand heat. The pelletized material was activated for one hour at 400 °C before pyridine adsorption. After that, the sample was heated to 150 °C while being exposed to pyridine (4 Torr), and the spectra were then gathered at room temperature. To investigate catalysts' basicity, temperature-programmed desorption of carbon monoxide (CO2-TPD) experiments were conducted using the same apparatus and procedures of NH3-TPD analysis except with the replacement of NH3 by CO2 flow. The fundamental characteristics of the catalyst were further characterized using pyrrole-probed IR spectroscopy. On an Agilent Cary 640 FTIR spectrometer with a high-temperature stainless steel cell and CaF2 windows, in situ FTIR was used to accomplish the experiments. All samples underwent a 1-hour activation period at 500 °C before the measurements. The activated catalyst was then outgassed at room temperature for 15 min after being exposed to 4 Torr of pyrrole for 15 min. Three scans were used to capture each spectrum at ambient temperature with an aspect ratio of 8 cm−1.The amount and type of coke formation on the catalysts after being used in the in-situ pyrolysis-catalytic steam reforming conditions were determined by thermogravimetric analysis (TGA) and derivative thermogravimetry (DTG), TEM, and CHNS (carbon, hydrogen, nitrogen, and sulfur) elemental analyzer. The TEM images of the used catalyst were conducted with a JEOL JEM-1011 microscope that functioned at 80 kV. TEM specimens were equipped by dispersing the catalyst powder in acetone with sonication and dropping it onto an ultrathin carbon-coated copper grid. The TGA analysis was performed using the Shimadzu TG-50 instrument in a nitrogen flow at a heating rate of 20 °C/min.The experimental setup for the in-situ pyrolysis-catalytic steam reforming reactions is mainly comprised of two reactors, and the setup is shown schematically in Fig. 1. The first is a Parr Benchtop Reactor (PBR) equipped with a stir-shaft, a pressure gauge, an autoclave body, a sampling tube, a safety valve, a heating jacket, and a thermocouple thermometer. The PBR is installed to homogenize the liquid phase of the high volume of cellulose in the phenol (2:8) at 70 °C and pressurize the solution with the N2 gas into the fixed bed reactor. The exit line of the PBR is swathed with glass fiber heating tape to vaporize the liquid before entering into the second reactor that is responsible for catalytic testing. The water line before the reactor was also preheated to 200 °C so that the water could first vaporize before being mixed with the gas phase of the cellulose and phenol. A vertical tube reactor with an inner diameter of 8 mm and a length of 300 mm was used for the catalytic testing, and it was situated within a furnace with a heating zone. Before the reaction, 0.2 g of catalyst powder was in-situ reduced at 600 °C for 1 h in pure hydrogen 30 mL/min flow after being fixed in the reactor's center using layers of quartz cotton at atmospheric pressure. After reduction, the reactor was purge-gassed with pure nitrogen for a time to clear out any excess reducing gas. A mass-flow controller system was used in each test to regulate the feeding stream. The thermocouples were positioned in the middle of the inflow area of the fixed bed reactor and were used to monitor the pressure, flow rates, and temperature continually. After the reaction, condensable molecules were liquefied by a glass coil heat exchanger equipped with a chiller at 10 °C. The gas reactor effluent was analyzed online employing a GC-TCD (Agilent 6890 N), and the liquid product was analyzed using a GC-FID (HP 5890 Series II) equipped with a 0.53 mm x 30 m CP-Wax capillary column and GC/MS (Agilent 7890B). Each run was repeated at least six times to ensure accuracy and reproducibility. The result analyses, such as phenol conversion (based on a calibration curve from GC-FID results), and produced gas composition in yield, were calculated following our previous research [25] and as shown in Eqs. (1), (2), (3), and (4). (1) Phenol conversion ( % ) = [ Phenol ] in − [ Phenol ] out [ Phenol ] in × 100 (2) H 2 yield ( % ) = moles of H 2 obtained moles of H 2 stoichiometric × 100 (3) CO yield % = moles of CO obtained moles of CO stoichiometric × 100 (4) CO 2 yield % = moles of CO 2 obtained moles of CO 2 stoichiometric × 100 The quantity of each chemical that must react for the reaction to be fully catalyzed is known as the stoichiometric moles. So, for example, we have Eq. 5's representation of the balancing steam reforming equation. (5) C 6 H 5 OH + 11 H 2 O ↔ 6 C O 2 + 14 H 2 Δ H o = 463.65 kJ / mol (6) C 6 H 5 OH + 5 H 2 O ⟶ 6 CO + 8 H 2 Δ H o = 710.91 kJ / mol (7) CO + H 2 O ↔ C O 2 + H 2 Δ H o = − 41.15 kJ / mol The structural properties of calcined catalysts determined from nitrogen adsorption isotherms at − 196.1 ºC are shown in Table 1, while the adsorption-desorption isotherms curves of nitrogen at − 196.1 ºC and pore size distribution profiles of the fresh samples are presented in Fig. 2. Table 1 shows the textural properties of fresh catalysts, which were specific surface area, total pore volume, and average pore diameters. The BET surface area of the TiCa catalyst is 6.26 m2/g. After the nickel material, the surface area increased to 11.89 m2/g, the pore volume also rose from 0.0374 to 0.0803 cm3g−1, and the average pore size reduced from 23.89 to 27.02 nm. After introducing lanthanum, the surface area increased to 18.01 m2/g, and pore volume and average pore diameters were reduced to 0.0685 cm3/g and 15.22 nm. The addition of transition metals significantly increased the surface areas, most probably due to the increasing metal and support interactions. When palladium was added, the surface area increased significantly (to 28.17 m2/g), and the pore volume rose (to 0.0995 cm3/g). It is important to note that the textural characteristics of the catalysts enhanced somewhat with the addition of the noble metals, but just a modest improvement was seen for the transition materials. The increase in the surface area and the total pore volume could also be obtained through these ways. The deposition of metal nanoparticles on the external surface leads to new adsorptive sites, which increase the adsorption of N2 on the surface. TiCa mixed with metal nanoparticles could form a porous coordinated complex composition wherein metal NPs could be inserted between TiCa large particles ( in the form of a sandwich structure). The low surface area of bare TiCa calcined at the same temperature confirmed this statement. The larger pore volume and surface area of the NLP/TiCa nanocatalyst compare to TiCa, N/TiCa, and NL/TiCa are beneficial for mass transfer, which often results in the high catalytic activity of the catalysts.Various pore morphologies have often been linked to the geometries of hysteresis loops. As presented in Fig. 2, the N2 adsorption-desorption isotherms of TiCa support belong to Type III (without a hysteresis loop), while the type of hysteresis loops for N/TiCa, NL/TiCa, and NLP/TiCa catalysts are Type H4 (with a significant increase in the adsorbed amount at P/Po>0.7) according to IUPAC classification [47,48]. This shows that it has a micro/mesoporous structure with a variety and abundance of mesopores. Adsorption was somewhat constrained at high P/Po, which may have been brought on by the presence of non-rigid aggregates of plate-like particles or collections of slit-shaped pores [49]. According to the adsorption isotherms, monolayer adsorption forms primarily at low relative pressure, but at high relative pressure, mesopore adsorption causes the production of many layers up to capillary condensation, which results in a significant rise in adsorption volume. Finally, the isotherm reaches a plateau, and the adsorption terminates in the mesopores. Only the TiCa and N/TiCa catalysts (approximately 11 nm and 30 nm, respectively) exhibit big pores, according to Fig. 2 of the BJH pore size distribution, but the pore diameters of the NL/TiCa and NLP/TiCa catalysts are between 5 and 9 nm. Fig. 2 shows that the bottom portion of the hysteresis loop area for this isotherm (up to 25 cm3.g−1 (STP)) overlaps the same region of the isotherm obtained for the TiCa nanocatalyst when the acquired isotherm is modified upward by ∼2.5 cm3.g−1 (STP). In contrast, the top portion of the hysteresis loop area (above 60 cm3.g−1 (STP) for this isotherm) overlaps the same region of the isotherm for the N/TiCa catalyst when the isotherm obtained following the introduction of nickel is modified higher by 3.3 cm3.g−1 (STP). Also, the amount of adsorbed nitrogen at higher relative pressures (P/Po) decreased with La doping, indicating a decrease in the mesoporous and improved in the specific surface area for the NL/TiCa nanocatalyst. This result shows that La has filled pores where capillary condensation occurs at intermediate relative pressures and that La has not interfered with the capillary condensation processes happening inside pores filling in either the higher or lower parts of the hysteresis loop. Such effects can also be attributed to the partial loading of pores and the formation of La crystallites on the external surface of N/TiCa particles. Compared to other samples, the NLP/Ti sample had the most N2 uptake in the 0.6–0.9 (P/Po) range, which suggests a larger mesopore volume. Increasing Pd and La modifiers loading seems to narrow the pore size distribution. The existence of Pd causes to block the pores of the catalysts, which leads to a decrease in the internal surface area. The deposition of Pd NPs on the external surface of the catalyst results in the generation of new adsorptive sites. Based on these results, we can tentatively presume that the catalyst surface area increased after loading Pd metal, which is considered beneficial for the in-situ pyrolysis-catalytic steam reforming conditions of cellulose dissolved in phenol. Fig. 3(a) and Table 1 show the XRD pattern and quantitative data of total crystal sizes, which were obtained through the analysis of the structure of crystalline materials and the identification of the crystalline phases present in a material to reveal chemical composition information based on their diffraction pattern. Diffraction data and JCPDS (Joint Committee on Powder Diffraction Standards) were analyzed using the X′pert Highscore software. The XRD curves of all catalysts showed characteristics peaks at 2θ angles of 23.05°, 34.33°, 37.32°, 47.45°, 53.82°, 59.28°, 69.84°, and 79.45° that were signed by red hearts corresponding to 101, 210, 102, 202, 103, 042, 242, and 161 diffractions of orthorhombic phase, which are in parallel with the standard JCPDS card number 96–231–0619 for Ca(TiO3) alloy and 92.8 nm of crystal size. The XRD pattern obtained for the TiCa catalyst shows individual peak characteristics of crystallized Ca(TiO3) and equals 93.4 nm of crystal size at 2θ angles of 68.84° and corresponding to 402 crystal structure. The two diffraction peaks at 50.81° and 72.34° for the TiCa catalyst (marked with green trefoil shapes) are ascribed to 211 and 123 monoclinic structural phase of Baddeleyite (Ti4O8; JCPDS 96–901–5356). The intensity of the characteristic peak of the N/TiCa catalyst is weaker than TiCa; probably, it may be due to the entry of Ni into the lattice of TiCa, which could cause a formation of a new solid solution structure. The prepared samples all displayed the characteristic diffraction peaks of Ti6O11 and were marked with blue diamonds (JCPDS 96–152–1096) with four prominent diffraction peaks (133.3 nm of crystal size) appeared at 30.29°, 32.17°, 64.09° and 67.34°, which could be ascribed to the 114, 206, 423, and 609 crystal planes of monoclinic Ti6O11, respectively. The diffraction peak with 2θ° values of 39.86 and marked with a purple triangle corresponding to La3Ni2O6.84 crystal plane of 312 and 43.6 nm of crystal size, confirming La3Ni2O6.84 orthorhombic structure (JCPDS 96–153–2218). After introducing La and Pd materials, multiple peaks were detected at 23.2°, 25.28°, 31.07°, 40.71°, 55.06°, and 75.04°. Clearly visible LaPd5 alloy (marked by blue circles with 26.6 nm of crystal size) was confirmed by the diffraction peaks of hexagonal at 40.71°, with the corresponding 002 crystal facets (JCPDS 96–152–2598). Similar to NL/TiCa catalyst, peaks were again observed at 75.04° with green circles, which is attributed to the metallic LaNi5 alloy with hexagonal phase structure of 211 and 259.6 nm of crystal structure (JCPDS 96–153–7852). The additional peak observed at 23.2° and marked with blue stars with a crystal size of 73.2 nm represents the 101 crystalline planes of La2.32O12Ti4 orthorhombic structure (JCPDS 96–412–4542). The anatase phase structure of TiO2 was seen at 25.258° (101) and ascribed by a red star with 89.3 nm crystal size and a JCPDS of 96–900–8215. The (102) La plane diffraction peak (with green star) observed for NLP/TiCa appeared at 31.07° with 58.9 nm (JCPDS 96–900–8526). Peaks for Ti6O11 crystal (marked with blue hearts) were detected at 55.06° and 62.75°, ascribed to the 514 and 517 crystal planes of monoclinic structures (175.8 nm, JCPDS 96–152–1096), respectively. For the NLP/TiCa and NL/TiCa samples, the diffraction peak intensity of the Ca(TiO3) alloy crystal phase is enhanced. When calcination at high temperatures, the spinel NLP/TiCa and NL/TiCa might completely decompose into Ca and Ti alloy. The excellent crystallinity and the largest surface area of NLP/TiCa nanocatalyst could be exhibited in the presents of LaPd5, LaNi5, La2.32O12Ti4, and Ca(TiO3) alloys resulting from high reducibility and metal support interaction as depicted in Fig. 3(b). It is expected that the NLP/TiCa can perform an excellent catalytic activity that can be directly associated with the catalyst's physical characteristics, such as crystallinity and surface area.The catalytic characteristics of transition and noble metal NPs deposited on TiCa may sometimes also be described in terms of the chemical contact, even though this impact of chemical interaction on catalytic activity is more often seen in Ni [25] and TiCa [31]. As in the case of the catalysts created by the deposition of Ni, La, and Pd on TiCa, strong bonding of deposited metals with the TiCa may promote efficient active center formation. In this regard, we conducted the H2-TPR analysis to study the redox properties of as-prepared catalysts, and the quantitative and profile results are depicted in Table 1 and Fig. 3(b), respectively. H2-TPR profiles of TiCa reveal that all the titanium-based oxides possess three reduction peaks at 268 °C, 411 °C, and 556 °C. These results differ from the H2-TPR profiles of our previous research [31], most probably because we used organic ash as the source of calcium material. The low-temperature peak at about 268 °C was assigned to the reduction of surface oxygen species. TPR profile at 411 °C correlated with a partial reduction of TiO2. The TPR curves for each sample show peaks over 500 °C, which are attributed to CaO reduction with significant TiO2 interaction, leading to the creation of Ca(TiO3) alloy, as demonstrated by XRD analysis. For the Ca sample, the peak seen at around 556 °C was connected to the process of CaCO3, which is created by CaO carbonation, and decomposing. The peak at 556 °C for a mixed TiCa support might be due to a decrease in the oxygen covering the surface of CaO. With a wide shoulder up to 654 °C, the N/TiCa catalyst begins to reduce into Ni° species at 390 °C, showing a low degree of reducibility and strong metal-support interaction with the TiCa. The reduction shoulder at 390 °C is associated with the reduction of nickel oxide (NiO + H2 → Ni° + H2O), which has poor interaction with the TiCa. This might be the reason for the H2 consumption decreasing from 12.09 mmol/g to 9.53 mmol/g when Ni is introduced to the TiCa support. The second peak at 654 °C belongs to the reduction of the nickel aluminate (Ni2+ → Ni°) due to the significant interaction between the nickel and the TiCa. The two identified reduction processes have been labeled Hw (weakly adsorbed hydrogen) and Hs since they are characteristics of transition metals (strongly adsorbed hydrogen). For particles between 0.9 and 2.2 nm, Sayari et al. [50] 's correlation of a greater quantity of Hw is in excellent accord with the current results for N/TiCa, NL/TiCa, and NLP/TiCa samples. This can be due to the non-dissociative nature of the H2 adsorption/desorption. N/TiCa, NL/TiCa, and NLP/TiCa samples, in contrast, have a high concentration of Hs species and metallic particles larger than 2.2 nm. This implies that Pd and La may facilitate H2's dissociative adsorption. The former peak (418 °C) is ascribed to the reduction of La2+ to metallic Lanthanum (La°), and the latter peak at 677 °C is accredited to the reduction of Ni2+ to metallic nickel (Ni°), and the 677 °C peak consumes more H2 than the 418 °C peak. Briefly, the low H2 consumption peaks at 176 °C and 314 °C for the NLP/TiCa curve, corresponding to the total reduction of Pd2+ to Pd°. The addition of Pd retards the reducibility of the NL/TiCa, as it is demonstrated by the move of the maximum of the peak from 677 °C to 705 °C, therefore, that a Pd*La, La*Ni, La*Ti and Ca*Ti interaction exist as proven by XRD result for LaPd5, LaNi5, La2.32O12Ti4, and Ca(TiO3) alloys, repectively.The study of form, which includes shape, size, and structure, is known as morphology. Morphology is significant for studying nanostructured materials because, in this context, the form determines the physical and chemical characteristics. The morphologies of fresh catalyst were analyzed by TEM, HRTEM, and SEAD, as shown in Fig. 4, and the elemental composition of materials was identified by EDX and elemental mapping analysis, as shown in Fig. 5. The Gatan microscopy suite software version 2.11 was employed to analyze the materials' TEM and HRTEM images and lattice d-spacing. The morphology of the prepared samples is verified by TEM analysis. As revealed in Fig. 4(a), the precursor TiCa nanoparticles synthesized by the hydrothermal method are interconnected and overlapped, nano-sized irregular structures, and exhibit two different shapes of particles. The overlapping and interconnection of Ti and Ca elements might be because of the formation of TiCa alloys as confirmed by XRD analysis (Ca(TiO3) alloy) and discussed in the reducibility study. Ti is illustrated mainly in spherical nanoparticles with approximate sizes of 70 nm, while Ca particles are in irregular cubic and rectangular shapes with an average diameter of 300 nm with a lattice d-spacing of 0.195 and 0.242 nm, respectively. The HRTEM images also show lattice edges with a spacing of about 0.195 nm, matching the 101 crystal plane of anatase-type TiO2 in the catalyst. As shown in the representative TEM images in Fig. 4(b) and (c), mesoporous nickel spheres with almost similar sizes and uniform spherical morphology were successfully synthesized. In the HRTEM image of Fig. 4(g) and (h), La has noticeable lattice spaces of 0.279 nm belonging to the (312) plane of La3Ni2O6.84 orthorhombic structure, as confirmed by XRD. The lattice edges with spacing at 0.211 nm can be assigned to the (002) plane of LaPd5 alloy, and the lattice edges value of 0.157 nm corresponds with the (211) plane of LaNi5 alloy. The SEAD pattern further confirms this structure (Fig. 4(i) and Fig. 4(j)).The elemental composition and the presence of Ni in N/TiCa (Fig. 5(a)) and Ni, La, and Pd in NLP/TiCa (Fig. 5(b)) were confirmed from the EDX spectrum, which shows the presence of all the expected elements without having any external impurities. As seen in Fig. 5(a), metallic Ni particles are in close proximity to an amorphous TiCa phase. The EDX mapping of NLP/TiCa nanocatalysts (Fig. 5(b)) indicates despite its complexities with the existence of trace elements such as Pd, La, and Ni, it is not difficult to note that the dispersion effect of Ni, La, and Pd elements on the NLP/TiCa catalyst is better and more uniform. Pd metal particles have small particle sizes, do not aggregate and are typically evenly scattered. Compared with the N/TiCa monometallic catalyst, areas in which the colors of trimetallic catalyst mixed indicate the interface of the active metal due to their overlapping EDX signals. The elemental line scanning analysis of the NLP/TiCa further verified the strong interaction of metal support and the formation of alloys, which agrees with the XRD, TPR, and TEM analysis.The catalytic activity of the catalyst surface and its resistance to carbon deposition in reforming and cracking processes are significantly influenced by its acid-base characteristics. The distribution of weak, intermediate, and strong basic sites and the total basicity of materials significantly impact the adsorption and dissociation capacity of the phenol and polymer molecules. This could speed up the removal of carbonaceous deposition from the catalyst surface, improving the catalytic performance and stability. Thus, CO2-TPD and pyrrole probed IR spectroscopy was carried out on fresh catalysts to understand the influence of transition and noble metals NPs content on the basicity of TiCa catalyst. The results of CO2 adsorption capacity and accessibility data of CO2 uptake for the fresh catalysts are shown in Fig. 6. By measuring and fitting the CO2 desorption peak, as shown in Table 1, it is possible to determine the catalyst's CO2 desorption quantity. Fig. 6(a) shows that all catalysts display a broad desorption peak at various temperatures, demonstrating the presence of several types of basic sites in the catalysts, including weak basic sites (100–230 °C), moderate basic sites (230–500 °C), and strong basic sites (above 500 °C) [51]. While the C atom in CO2 is the electron-deficient core and the CO2 molecules have vacant orbitals at low energy levels, TiCa may readily shed its outside electrons due to its relatively low initial ionization energy [52]. Many basic sites and adsorbed O2 on the surface were attributed to the diminished catalysts' capacity to absorb CO2. The medium desorption peak at 260 °C for the N/TiCa nanocatalyst is because of the under-coordinated O2−. The NL/TiCa, and NLP/TiCa nanocatalysts detected small shoulders at around 169, 206, and 219 °C are assigned to surface –OH [53] indicating the presence of small weak basic sites resulting in less desorption and activation of CO2 in their structure. Those desorption peaks corresponded to the interaction of CO2 with weakly basic hydroxyl groups on NL/TiCa, and NLP/TiCa nanocatalysts. Furthermore, adding La and Pd materials produced higher peaks in regions 637 °C and 850 °C; thus, it is indicated that adding La and Pd materials leads to the higher CO2 adsorption capacity of the catalysts. The peak position given to the strong basic increased further after Pd loading, indicating its strongest binding affinity to CO2 that enhanced the catalyst’s basicity. This inference was made because Pd and La interactions with the N/TiCa nanocatalysts are compatible with the observed increase in the quantity of CO2 desorbed. Since CO2 molecules may be converted into reactive CO2 δ- species as a result of the transition and noble metals utilized in this study, it is possible to efficiently boost CO2 molecule absorption by the N/TiCa, NL/TiCa, and NLP/TiCa nanocatalysts. As increased basicity may decrease the production of coke [54] during the reaction, it is anticipated that the change of basicity with La and Pd concentration in the catalysts may have some impact on the catalytic activity. The H2-TPD, in which NLP/TiCa exhibits substantial H2 adsorption, and the CO2-TPD exhibit strong correlations.The catalyst's basic sites engage with the pyrrole's N − H group via the development of an H−bond interaction, which the stretching can see in the N − H group. Fig. 6(b) shows the measurement of adsorbed pyrrole on the calcined catalysts using N − H stretching and the wide peak in the range of 3200–3700 cm−1. Additionally, it has been shown that pyrrole chemisorbs dissociatively over powerful basic metal oxides and that deprotonation of pyrrole on the most powerful basic oxygens results in the generation of pyrrolate anions that are stabilized by surface cations [55]. In N/TiCa and NLP/TiCa nanocatalysts, a strong peak was found at ∼3544 cm−1, indicative of stretching vibrations between the H atom of pyrrole and the (Ni−O− and Pd−O−) group of basic oxygen present in the catalysts' framework. Interestingly, the NLP/TiCa catalyst had a larger peak intensity than the other catalysts, indicating that the catalyst was constructed with many basic sites. The development of intra- and inter-particle porosity was one potential factor that increased the basicity of the NLP/TiCa catalyst [56]. The peak above 3544 cm−1 also represented the N − H group of pyrrole molecules in the environment, and the N − H band (physisorbed pyrrole in a liquid-like state) interacts with the π-system of nearby pyrrole molecules.Metal-acid bifunctional compositions are often used in industrial catalysts for cracking and reforming processes, where metallic sites catalyze hydrogenation/dehydrogenation reactions, and acidic sites catalyze cracking. One of the crucial variables affecting the catalytic performance in the n-situ pyrolysis-catalytic steam reforming processes is the catalyst's acidity. The findings of NH3-TPD's analysis of surface acidity in terms of the quantity and strength of acid sites are shown in Fig. 7(a) and Table 1. All samples have strong acid sites because they showed the presence of peaks above 400 °C. The highest quantity and strength of acid sites for the TiCa sample proposes that the superior acidity of the TiCa nano-catalyst may be due to the Brønsted acid sites on the catalyst [31]. A higher acidity may cause to produce better activities, as it is known to break C−C and CO− binding, but it may also favor higher coke formation during the reforming reactions. Therefore, the bare TiCa might face catalytic deactivation by coke deposition. As can be seen in the quantitative data, N/TiCa and NL/TiCa had basically similar desorption of NH3, indicating that each sample's total acid site density was the same. Ni and La were added to the TiCa structure, which reduced the number of acid sites while simultaneously shifting the high-temperature peak to lower temperatures. With Ni and La in porous TiCa support, Ni and La and support atoms are bonded to form Lewis acid sites due to the different electronegativity between the transition metals and the TiCa atoms. Thus, the acidic sites in N/TiCa and NL/TiCa may be formed by the sites containing electron holes in porous TiCa. These two catalysts had mild acid and basic properties compared to the bare TiCa with the strongest acidity and NLP/TiCa with the strongest basicity. The low acido-basicity properties of N/TiCa and NL/TiCa might also be attributed to the low amount of alkali, which was insufficient to form strong interactions between the alkali and the TiCa support during the preparation of catalysts. Since the –OH groups were lost when Pd2+ ions interacted with the support surface after the addition of Pd metal, it is possible that the catalyst's acidic quantity was dramatically reduced with the insertion of Pd atoms [57]. Consistent with the results of Pd-modified NL/TiCa, the introduction of Pd leads to catalysts of higher basicity and lower acid site having bifunctional properties, which are significant in the in-situ catalytic steam reforming of phenol coupled with thermal cracking (or pyrolysis) of cellulose.The distribution and nature of acidic sites and their effect on selectivity must also be considered. As a consequence, the acidity of the catalysts was further investigated using FTIR spectroscopy with adsorbed pyridine as a probe molecule, and the results are shown in Fig. 7(b). For coordinatively bound (Lewis acid sites, "L") and protonated bonded (Brønsted acid sites, "B") pyridines, the bands at 1540 and 1440 cm–1 were used as measures, respectively [58]. Higher wavenumbers also demonstrated better surface adhesion, increasing Brønsted acid's strength. It is noteworthy that the TiCa has many Brønsted and Lewis acid sites compared to other catalysts with transition and noble metals. All samples had bands at around 1540 cm−1 caused by pyridine adsorbed on Brønsted acid sites; however, the N/TiCa and NL/TiCa catalysts did not show any bands at 1440 cm−1 that were caused by the pyridinium ion (PyH+), which is responsible for Lewis acid sites. This does not imply that there are no Lewis acid sites; it only indicates that the Lewis acid sites were insufficiently powerful to create the pyridinium ion. However, it is simple to induce carbon deposition and deactivate the catalyst at the Brønsted acid and strong acid sites [59]. The Pd NPs inserted into an NL/TiCa have jointly disturbed the catalyst's Brønsted and Lewis acid phases. As Pd addition increases, the catalysts' pyridine adsorption peaks become less intense, indicating a weaker acid. To conclude, reducing the strength of Brønsted and Lewis acid phases is vital to prevent the undesired reaction, especially for the reactions that deal with polymer molecule thermal cracking. Therefore, it is expected that NLP/TiCa will show high catalytic activity and selectivity in the in-situ catalytic steam reforming of phenol coupled with the pyrolysis of cellulose.The spectroscopic method known as FTIR is highly effective and widely used for identifying functional groups in compounds and complex substances. In this work, KBr pellets are utilized as a carrier for the sample in the IR spectrum since they are optically transparent to light in the IR measurement range. We choose the KBr pellet approach for FTIR spectroscopy because it is straightforward and enables us to run the whole mid-IR region down to 400 wavenumbers without running split mulls [60]. All catalysts were subjected to FTIR-KBr spectroscopy, and the spectra in the 4000–400 cm–1 region are shown in Fig. 8. The TiCa sample exhibited an absorption peak at 957 cm–1, ascribed to C–O symmetrical structure ether groups. These lines are connected to chromophore vibrations (CC stretches and hydrogen out-of-plane vibrations, respectively), so it is more likely that exchangeable protons will influence them in the chromophore than exchangeable protons in other regions of the molecules. This peak shifted to 833 cm–1 after adding Ni, La, and Pd metals. The stretching vibrations of metal oxide in octahedral group complex Ni(III)–O2−, La(III)–O2−, and Pd(III)–O2− tetrahedral group complex formation is proved by the bands at 594 cm–1. The FTIR spectra at 1165 cm−1 and 1774 cm−1 for the N/TiCa, NL/TiCa, and NLP/TiCa nanocatalysts correspond to the stretching and bending vibration of C−H, C−O−H, C−O−C bonds and –CO absorption vibration, respectively. There is a highly characteristic weak band at 2376 cm−1 for all samples that correspond to vibrations of the ʋ(C−H) mode related to CH2 −CO groups; while the peak at 2932 cm−1 indicates the existence of C−H vibrations in methyl (–CH3) and methylene (−CH2 −) group. The ν(OH) band around 3642 cm−1, which is assigned to non-hydrogen-bonded water molecules coadsorbed with CO, exhibits a gradual decrease by adding Ni, La, and Pd metals. Besides, FTIR bands observed in the hydroxyl region at 3765 cm−1 are ascribed to the vibration of OH– groups adsorbed along the support surface and correspond to terminal hydroxyl groups. The biggest bands centered at 1443 cm−1 can be attributed to the surface complex of –CH2 bending (methylene group), while the 1463 cm−1 peak can correspond to the presence of metal carbonates (stretching vibration of CC). Notably, the intensity of FTIR peaks ascribed to the methylene groups and metal carbonates of the NLP/TiCa, NL/TiCa, and N/TiCa catalysts were more intense than the TiCa catalyst. This suggests that the differences in the surface methylene groups are caused by the Ni, La, and Pd crystal structure.The catalyst's performance on hydrogen generation and its stability in the system have been investigated using catalytic activity parameters regarding phenol conversion, product yield, temperature (for the best catalyst), and time on stream performance. The screening of reduced catalysts was performed at 600 °C. The activity was investigated and repeated for six cycles with an experimental duration lasting 60 min for each run; the results are illustrated in Fig. 9. Analysis of produced gas composition shows that the most significant changes in concentration occur for H2 yield, CO2 yield, and phenol conversion. It was confirmed that the addition of transmission and noble metals did not significantly affect CO yield but slightly decreased. At TiCa nanocatalyst, the phenol conversion and H2 yields were 34.3% and 40.5%, respectively. Phenol conversion was increased to 49.3% and 70.2%, and H2 yield enhanced to 45% and 57.2% for the N/TiCa and NL/TiCa, respectively. The highest conversions of phenol at 82.6% and H2 yield at 82.2% were achieved with NLP/TiCa catalyst. The increased catalytic activity under the NLP/TiCa nanocatalyst can be attributed to the increased surface area, metal support interaction, basicity, and active metal distribution of the synthesized catalyst as characterized by BET surface area, CO2-TPD, H2-TPR, and EDX. A reactant's contact with a higher surface area affects the number of collisions and the reaction rate. The NLP/TiCa nano catalyst has more porosity and is more active than those other samples because it has more surface area to form the active sites, leading to greater activities in this work. As a result of the large dispersion of active sites and the accessibility of reactants to active sites, we can assert that the high BET surface area and pore volume of Pd-containing catalysts increase the selectivity to hydrogen. Furthermore, the maximum hydrogen consumption of the NLP/TiCa nanocatalyst in the H2-TPR process also proves the total reduction of Pd2+ to Pd°. The addition of Pd in the NL/TiCa increased the reducibility properties of the catalyst; therefore, a Ca*Ti, Pd*La, La*Ti, and La*Ni interaction exist as proven by XRD result for LaPd5, LaNi5, La2.32O12Ti4, and Ca(TiO3) alloys. From the particle size distribution of the active metal in the EDX images, it can also be found that the NLP/TiCa nanocatalyst has a smaller particle size and a more uniform distribution. Therefore, it shows the most excellent catalytic activity in the reaction. According to the CO2-TPD and Pyrrole adsorption FTIR spectra findings in Fig. 6, the NLP/TiCa nanocatalyst's more excellent catalytic activity may be caused by the nature of Pd and the existence of a significant number of basic sites in the structure [61]. These findings suggest that the catalyst's basicity facilitates the considerable adsorption of CO2 molecules and increases H2 production. For instance, Pizzolitto et al. [62] observed that the NiLa/ZrO2 catalyst's basicity increase significantly reduced the dehydration of ethanol. We also found the same positive effect of basic catalyst sites on the catalytic pyrolysis steam reforming reaction PET-phenol for H2 generation [25,27,35,45]. Less crystal size of NLP/TiCa nanocatalyst (as seen in XRD analysis) plays a substantial part in minimizing the coke production and deposition and enhancing the catalyst lifetime throughout the in-situ pyrolysis-catalytic steam reforming reaction process. Given its good catalytic performance for phenol steam reforming (in terms of phenol conversion and H2 selectivity), the NLP/TiCa nanocatalyst was selected for further evaluation based on temperature effect and deactivation check studies.The reforming temperature significantly influences the concentration of reforming products; the results are shown in Fig. 10. Conversely, using only 500 °C resulted in 67.6% of phenol conversion and 59.2% of H2 yield. Due to the accelerated phenol steam reforming reaction ( C 6 H 5 OH + 11 H 2 O ↔ 6 C O 2 + 14 H 2 ) and water gas shift reaction ( CO + H 2 O ↔ C O 2 + H 2 ), an increase in temperature regulates the enhancement of phenol conversion into gaseous products and H2 yield. Higher gas yields compared to 500 °C were seen along with the high conversion at the maximum temperature, which is anticipated given that the gas-forming cracking processes are thermally regulated. Almost complete conversion of phenol (98.7%) and maximum H2 yield (99.6%) were achieved at 800 °C. At the same time, the CO2 yield is enhanced from 16.7% at 500 °C to 27.4% at 800 °C, while the CO yield decreases from 18.9% to 8.8% due to improving reaction conditions for water gas shift reaction. In addition, it was also observed that the increase in acid sites, decrease of the surface area, and metal dispersion of the TiCa nanocatalyst were sufficient to justify the low phenol conversion and H2 yield. This made it feasible to conclude that the major pathway for the deactivation of the TiCa catalyst and likely has an impact on how well the TiCa catalyst performs in terms of producing H2 is the development of amorphous coke. This statement is verified in the characterization of used catalysts.The phenol conversion and H2 production over the NLP/TiCa nano catalyst were evaluated at 600 °C for 72 h of time-on-steam, and the stability results are displayed in Fig. 11. With a slight decrease (∼4% loss in yield), the H2 yield was maintained constantly at approximately all reaction times. During the reaction, the CO yield decreased from 14.9% to 8.6%, while the CO2 yield fraction increased from 19.7% to 23.7%. After 32 h, it was noticeable that phenol conversion was about 74% and remained almost stable (with a negligible decrease) for the rest of the time (72.3% at 72 h). This is almost in agreement with the total basic sites, Pd and La dispersion, high metal-support interactions, and different alloys determined by CO2-TPD, H2-TPR, and XRD, respectively.To understand the component presented in the produced liquid during the thermal reaction of cellulose and steam reforming reaction of phenol, the liquid product at 600 °C was analyzed by GCMS technique, and the identified compounds are summarized in Table 2. The chemical components in the liquid product are analyzed in accordance with the thermal breakdown of the feedstock's cellulose and phenol. Due to their volatility and complicated structure, several chemicals were undetected. The aromatics and alkanes occupied mostly the liquid product, mainly from cellulose decomposition. The liquid was mainly composed of C2-C12 aliphatic hydrocarbon, C6-C9 aromatic hydrocarbon, C13-C18 aliphatic hydrocarbon, and C19-C72 aromatic hydrocarbon. At a retention time of ∼59 min, the liquid was composed of a C70 aliphatic hydrocarbon for the NL/TiCa nanocatalyst. It is evident that despite the variable catalytic activity of different catalysts, the relative product profile remains largely the same, with 1-propanol, ethanol, toluene, and phenol (incoverted reactant) being the major product for all the catalysts except 2,4-dimethyl-benzo[h]quinoline, and hexadecane components for the N/TiCa catalyst. The aromatic byproducts of the cracking process, which resulted from breaking the main chain in cellulose and dehydration of the –OH bond on its alkyl chains, may also be the source of phenolic compounds. The chemicals contain oxygen such as C8H10O5PW+, C40H48O4, C29H23NO, C4H7OH, C14H42O7Si7, C16H32, C17H34O2, C23H39NO2S, C28H43NO2, C28H45NO2, C40H73NO5Si4, and C21H46O2Si produced for the TiCa nanocatalyst could lead to the instability of catalyst. The oxygen compounds in produced fuel decreased with the addition of transition and noble metals NPs deposited on TiCa, indicating that Ni, La, and Pd metals promoted deoxygenation, and more oxygen compounds were decomposed into low molecular-weight substances. The occurrence of 2,4-dimethyl-benzo[h]quinoline is the unique structure of petroleum triaromatic azaarenes that could be more firmly established after the identification of individual compounds. It has been possible to radiolabel exosomes, hydrogels, and other biological materials since 1982 with the help of the generated hexadecane chemical. This substance is also helpful for positron emission tomography. Catalytic cracking activity of the cellulose and phenol under NLP-TiCa catalyst produced unique compounds such as (E)− 2-bromobutyloxychalcone, beta,epsilon-Carotene-3,3′,8,19-tetrol, 7,8-dihydro-, 1,5-benzodiazocin-6(1 H)-one, 8,10-bis(dodecylsulfonyl)− 2,3,4,5-tetrahydro-5-methyl-, 2,4,6,8,10,12-Tridecahexaenoic acid, 13-(3-chloro-4-methoxyphenyl)-, 2-decyl-3-methoxy-5-pentylphenyl ester and Lanost-9(11)-en-18-oic acid, 23-(acetyloxy)− 3-[(4-bromobenzoyl)oxy]− 20-hydroxy-,.gamma.-lactone, (3.beta.,20.xi.). The Pd metal induced stronger cracking activity than other catalysts by significantly increasing the proportion of aromatics while significantly decreasing the oxygenated products (including phenols). Furthermore, it has been shown that the H2 yield and phenol conversion during the catalytic process seem to have increased in the presence of the NLP/TiCa catalyst (see Fig. 9). The liquid products were further analyzed using the FTIR technique.The FTIR method was used to identify the functional groups present in the liquid fuel, and the FTIR curves with the intensities of each band are displayed in Fig. 12 (a) and Fig. 12 (b). The absorption band present in the wave number range 632 cm−1 can be linked to the OH– vibration [63]. At band detection 879 cm−1 (HNO3), the carbonate bending mode can be assigned to the out-of-plane deformation band [64]. There is a substantial boost in intensity of the bands at about 1033 cm−1 wavelengths, signifying the chemical functional groups of –CH2– bending vibration of the aliphatic hydrocarbon [65]. Andrea et al. [66] stated that the bands at 1033 cm–1 could also be attributed to the C–H in-plane deformation vibration of 1,4-disubstituted or 1,2,4-trisubstituted benzene rings. This peak might also correspond to the C–O stretching and C–O bending of the C–O–H carbohydrates [67]. The band at 1404 cm−1 signified carboxylate CO stretching weakened, demonstrating that the non-conjugate CO structure in lignin has been decomposed [68]. This peak may also be ascribed to the aliphatic C−H deformation of CH2 and CH3 bending and C−OH deformation of COOH, COO− symmetric stretch [69]. The characteristic absorption peak at 2345 cm−1 corresponds to CO2 based on its vibrational occurrence and reflection as a 1,3,5-triamino-2,4,6-trinitrobenzene thermal decomposition product [70] and the O–C–O anti-symmetric stretching mode [71]. The vibration peaks of −CH3 (νas(CH3)) in 2,5-dimethylfuran were observed at 2962 cm–1 [72,73], and symmetric and asymmetric stretching vibration of N − H (ν(NH) bound) [74,75] and the presence of urethane groups [76,77] were observed at 3333 cm–1.Characterization of the spent catalyst is critical for stabilizing stable catalysts against coke formation and long-term usage. The transient deactivation of the catalyst caused by the buildup of carbonaceous deposits (coke) during catalysis affects throughput, necessitates regeneration procedures, and results in a partial permanent loss of catalytic efficiency. This part analyzed the spent catalysts by TGA-DTG, CHNS, BET FTIR-KBr, and TEM. By observing the weight change that occurs while a sample is heated at a consistent rate, the TGA analytical method may be used to evaluate a material's thermal stability and the percentage of volatile components. The rates at which these volatile components are removed in %/min are determined by DTG, and results are shown in Fig. 12 (c) and Fig. 12 (d) and Table 3. The TGA curves of the used catalysts may be roughly separated into three sections, as observed. Weight loss in the first stage, denoted by WL1, is evident in the temperature range of 25–200 °C and results from removing adsorbed water and unreacted molecules. The chemical adsorbed on the sample's surface or found in the mesopores is responsible for the WL1 [78]. The second part (WL2), located in the medium temperature range (200–600 °C) shows an increasing tendency in weight percent and can be attributed to the burning of deposited coke. Here as “coke,” we considered the carbon deposited on the catalyst and all the condensed hydrocarbons determined by the material balance. The inadequate breakdown of nitrate during the creation of metal oxides should also contribute to the WL2. The WL2 area is ascribed to the overlaps of metal oxidation and removal of the amorphous carbon, except for the weight increase of the TiCa catalyst, which may be driven by the oxidation of the metallic Ti and Ca [10,16]. Amorphous carbon burns at temperatures lower than 400 °C [79–81]. Amorphous carbon may be readily removed by oxidizing at low temperatures, but it often encircles the catalyst's active metal particles and renders it inactive. The third phase (WL3) above about 600 °C comprised the decomposition of remaining residues and heavy carbonaceous species, most probably by graphitic coke. The shell structure is composed of graphitic carbon with some defect sites, as confirmed by TEM analysis (Fig. 15). The weight loss between 80 and 120 °C is ascribed to the loss of surface hydroxyls and physically or chemically linked water from all samples, which was supported by their DTG endothermic peaks [82]. The first derivative of TGA curves (DTG) in Fig. 12 (c) presented the highest weight reduction rate for all catalysts with higher intensities for catalysts with metals occurred at below 100 °C, whereas it was significantly higher for TiCa at ∼750 °C min and N/TiCa at ∼450 °C. Primary pyrolysis processes occur on cellulose at low-temperature ranges. In this stage, monomeric phenols undergo side-chain reactions, ether bond cleavage, and evaporation. Methoxy group bonds are broken down, and aromatic rings are broken down and condensed during the secondary pyrolysis events, which take place above 400 °C [83,84]. However, DTG curves show that the peaks of TiCa and N/TiCa became more considerable than that of Pd and La, indicating that the introduction of Pd and La into the catalyst can considerably affect the reforming and pyrolysis reaction behavior of phenol and cellulose. The catalyst sintering and carbon deposition behaviors are low in NLP/TiCa and NL/TiCa, whereas the TiCa catalyst deactivation is caused by carbon deposition rather than the metal sintering, but N/TiCa displays severe sintering and carbon formation performances. Generally speaking, the total weight losses of TiCa, N/TiCa, NL/TiCa, and NLP/TiCa were 72.31%, 61.36%, 61.36%, and 51.38%, respectively; which are in line with the carbon content from CHNS analysis (13.8%, 9.6%, 4.1%, and 3.8%, respectively). Most of the carbon combustion happens at the WL2 region, indicating an exothermic process consistent with oxidation. The breakdown of CO groups on the surface of carbon-based catalysts may also be responsible for the weight loss over 700 °C [85]. The carbon deposition side reaction of the TiCa and N/TiCa catalysts is also related to the higher acidity of those samples, as described in NH3-TPD and Pyridine-FTIR spectra in Fig. 7.The spent catalysts were also examined by BET surface area and N2 adsorption-desorption isotherm, and the results are depicted in Table 3 and Fig. 13. As displayed in Table 3, the surface area of the spent catalysts employed in this work was considerably decreased compared to that of the fresh samples, most probably because of thermal sintering and carbon deposition. The SBET of spent TiCa, N/TiCa, NL/TiCa, and NLP/TiCa nanocatalysts were 0.131, 1.39, 3.89, and 10.6 m2/g, respectively. However, the highest surface area remains for the NLP/TiCa nanocatalyst with more porosity and active sites than those other samples. This sample with Pd NPs showed the most increased catalytic activity (Fig. 9) and lowest carbon deposition (). These results also indicate that the catalysts with Pd NPs are more competitive for recovering the textural characteristics of the spent samples. The results revealed that the H4 type isotherms loop remained unchanged for all catalysts with transition and noble metals NPs with pore size distribution in the range of 2–25 nm after the reaction, demonstrating the collapse of the mesoporous framework was not pronounced. These results indicated that adding Pd is beneficial to keeping the catalyst's activity and improving the catalyst's anti-coking performance.Furthermore, a qualitative analysis of coke structures on the spent catalysts was conducted using FTIR spectroscopy. The two key bond vibration areas in the FTIR spectra of complete coke deposited on TiCa, N/TiCa, NL/TiCa, and NLP/TiCa nano-catalysts are shown in Fig. 14. While the vibrations in the area of 1650–1350 cm−1 belong to aromatics and specific bending modes of aliphatics, the vibrations in the 3200–2700 cm−1 essentially correspond to olefins (asymmetric and symmetric stretching) and monocyclic aromatics (olefins). Specifically, the IR bands at 586 cm−1, 864 cm–1, and 957 cm–1 are designated to functional groups of amide VI species, =CH bending out of the plane, and C–O symmetrical structure in aliphatic nature of coke, respectively. The single shoulder for the N/TiCa nanocatalyst at around 1265 cm−1 is ascribed to the −C–O single bond vibration of −C–OH group. The parent peaks at 1443 cm−1, which can approve the presence of skeleton vibration of the pyrrole ring and CC stretching [86] clearly indicating it is strengthened in this trend TiCa<Ni/TiCa<NL/TiCa<NLP/TiCA. Thus, pyrrole ring and CC stretching are assigned to the catalyst with higher weight loss and lower catalytic performance. These peaks might also correspond to the asymmetric stretch vibration of CO3 2- molecules [87]. The characteristic peaks at 1635 cm−1 correspond to the adsorbed H2O molecules at the spent catalysts, whereas the 1805 cm−1 single bans for the NLP/TiCa sample are ascribed to the C= Ο stretching modes. The FTIR peaks at 2399 cm−1, 2924 cm−1, and 3449 cm−1 wavenumbers for the spent catalysts with transition and noble metals were equivalent to the linearly adsorbed OCO species [88], asymmetric stretching vibration of C−H [89], at ascribed to the O–H stretching vibration [90,91], respectively. However, the intensities of those peaks were increased in the spent catalysts with transition and noble metals, which is evidence for decreasing monomeric phenols via successive C–O bond cleavage.TEM images of the spent NLP/TiCa sample after 72 h time on stream are revealed in Fig. 15. After the in-situ pyrolysis-catalytic steam reforming reaction, the basic morphologies of the NLP/TiCa catalyst are retained, agreeing with the preferable performance and stability during the reaction; along with some smaller microparticles placed on the surface of the larger cuboid particles. The substances may identify Coke deposits with non-uniformly shaped particles shown in the photos. The coke deposits' disorganized structure matches that of amorphous coke exactly. The deposited carbon appears as the core-shell structure on the catalyst particles can be ascribed to the encapsulated graphitic carbon with a shell of 15 nm. The two factors responsible for this carbon production are carbon development on the surface of nickel and lanthanum catalysts and carbon diffusion via transition metal particle surfaces. This sort of carbon causes transition metal active sites to have weaker coke covering, which results in active metallic particles still being in contact with the gas flow and delaying the catalyst's deactivation [92].This research revealed the synergistic influence of Pd, Ni, and TiCa support for hydrogen and valuable liquid generation from cellulose bio-polymer wastes dissolved in phenol. To increase this study's feasibility and the cellulose ratio to solvent ratio, we modified the experimental setup and added a Parr Benchtop Reactor. This modification allows the quickly liquefying and rising of the amount of polymer and plastic waste in the reaction without causing line blockage. Four catalysts were prepared through hydrothermal and impregnation techniques: TiCa, N/TiCa, NL/TiCa, and NLP/TiCa nanocatalysts. The physicochemical characteristics of the fresh catalysts were examined by XRD, BET, N2 adsorption-desorption isotherm, NH3-TPD, IR-Pyridine, IR-Pyrrole, H2-TPR, CO2-TPD, FTIR-KBr, TEM, EDX, HRTEM, SAED, EDX, and ICP test and the used catalysts were characterized by TGA-DTG, BET, TEM, FTIR-KBr and CHNS. The best catalytic interaction between Pd, Ni, La, and TiCa was observed for the trimetallic NLP/TiCa nanocatalyst, which showed the highest catalytic activity at 600 °C compared to monometallic N/TiCa and bimetallic NL/TiCa catalysts, with high stability for 72 h of time on stream, almost complete phenol conversion (98.7%), and 99.6% of H2 yield at 800 °C. The increased catalytic performance under the NLP/TiCa nanocatalyst can be ascribed to the increased surface area, metal support interaction, basicity, and active metal distribution of the synthesized catalyst as analyzed by BET surface area, H2-TPR, CO2-TPD, and EDX. The addition of Pd in the NL/TiCa increased the reducibility properties of the catalyst; therefore, a Ca*Ti, Pd*La, La*Ti, and La*Ni interaction exist as proven by XRD result for LaPd5, LaNi5, La2.32O12Ti4, and Ca(TiO3) alloys. Carbon deposition was not considered the only reason for the low performance for TiCa and N/TiCa nanocatalysts. The catalyst sintering and carbon deposition behaviors are low in NLP/TiCa and NL/TiCa, whereas the TiCa catalyst deactivation is caused by carbon deposition rather than the metal sintering, but N/TiCa displays severe sintering and carbon formation performances. The carbon deposition side reaction of the TiCa and N/TiCa catalysts is also related to the higher acidity of those samples, as described in NH3-TPD and Pyridine-FTIR spectra. The material balance also determined all the condensed hydrocarbons. The deposited coke appears as either large amorphous and graphitic with catalyst particles or as a rather uniform carbon coating that covers the catalyst particles. The liquid product compounds produced mainly from the thermal cracking of cellulose in almost all samples were ethanol, toluene, phenol, 2,4-dimethyl-benzo[h]quinoline, and hexadecane. The current study will offer an industrial basis for synthesizing effective trimetallic nanosized catalysts and help in the operative utilization of cellulose biopolymer waste and phenolic compounds for liquid fuel and hydrogen gas production.W. Nabgan: Writing, Experimental and characterization, T.A. Tuan Abdullah: Supervision and Methodology, M. Ikram: Writing and characterization, A.H.K. Owgi: Writing & English editing, A.H. Hatta: Writing tasks of the experimental part, M. Alhassan: Writing tasks of the reaction and stability parts, F.F.A. Aziz: Figures development and English editing, A.A. Jalil: English editing and characterization, T.V. Tran: Writing & English revision, R. Djellabi: Writing & English editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The primary author, Walid Nabgan, is grateful for the support from Universitat Rovira i Virgili under the Maria Zambrano Programme (Reference number: 2021URV-MZ-10). The authors are also grateful for the support given by Universiti Teknologi Malaysia (UTM) allocation budget in Pusat Pengurusan Makmal Universiti (PPMU) laboratory.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jece.2023.109311. Supplementary material .

Hydrogen and liquid fuel production from biopolymer waste, such as cellulose dissolved in phenol, was investigated using in-situ pyrolysis-catalytic steam reforming conditions. Developing a sustainable method for the thermal cracking of such biopolymers still faces difficulties due to the catalyst stability primarily impacted by coke deposition. The key to the proposed method is improving a highly active and stable catalytic reforming process in which trimetallic Ni-La-Pd supported on TiCa acts as a primary reforming catalyst. Catalysts were prepared by hydrothermal, and impregnation techniques, and the physicochemical characteristics of the fresh and spent materials were examined. The results showed that the NLP/TiCa catalysts performed effectively due to their comparatively high surface area, strong basicity, evenly distributed Pd particles, and appropriate redox and desorption characteristics. The addition of Pd retards the reducibility of the NL/TiCa; therefore, a Pd*La, La*Ni, La*Ti, and Ca*Ti interaction exist. Almost complete conversion of phenol (98.7%) and maximum H2 yield (99.6%) were achieved at 800 °C for the NLP/TiCa. These findings give an insight into industrial-scale development. They have significant potential for enhancing the generation of hydrogen and liquid products from phenol and cellulose waste, such as propanol, ethanol, toluene, etc.

Heterogeneous catalysis systems have created attention due to availability of number of catalysts or the use of diverse substrates such as silica, sepiolites, zeolites, and polymers for metal complex immobilization [1,2] associated with numerous advantages including effective recovery of the catalysts by filtering, reusability, robustness, enhanced air and moisture stability, environmental compatibility, low cost, non-toxicity, non-corrosiveness etc. [3–9]. The ability to alter the shape and local environment at the catalytic site in accordance with demands of a specific reaction is the most significant characteristic of porous material for which they exhibit enormous potential for catalytic applications. Zeolites are crystalline structure with pore containing aluminosilicate, which is commonly applied in many demands such as separations, catalysis, ion exchange, and adsorption because of their tunable properties such as high surface area and narrow pore size distribution, acidity, thermal and structural stability with specific functionality [10–12]. In comparison to traditional micrometer-sized zeolites, nanocrystalline zeolites have greater outside surface areas and well adsorption property, which causes diffusion of reactant and product molecules easily during a reaction. Na–Y zeolite have a unique structure of Faujasite-type framework (FAU) with three-dimensional channel and pore size of 7.4 ​Å that can support organic and inorganic compounds to form a hybrid material [13–15]. It is a synthetic zeolite carries negative charge in the framework and sodium cation neutralizes the aluminosilicate framework structure [16]. In the era of catalysis, transition metal complexes play a significant role [17] due to their adaptable coordination environment, and they have a wide range of applications in several areas including optical, electrical, magnetic materials, and catalysts [18–20]. In addition to enhancing the catalytic performance of the support, incorporation of transition metal complex ions in a silica surface also results in high thermal stability of the hybrid material, consistency and availability of the active sites, ease of separation, recyclability, and an environmentally friendly pathway for chemical reactions, among other benefits [21].There are many reported works on catalysis using Ni complex on functionalized porous materials, F. Havasi et al. described the use of Ni complex attached functionalized MCM-41 for the synthesis of 2,3-dihydroquninazolin-4(1H)-ones [22]. L. Tahmasbi et al. reported Co(II) and Ni(II) complex immobilized mesoporous silica and enzymes supported metal complex for the investigation of antibacterial activity against gram-positive and gram-negative [23]. 2D and 3D porous clay nanostructure were also synthesized and Ni-β-diimine organometallic complex was applied in oxidation of ethylene oligomerization [24]. There are various reports on Co(II) and Ni(II) complex anchored functionalized mesoporous materials for the application of oxidation of olefins with different epoxides, sulfides, ethylene oligomerization, epoxidation of styrene, olefins etc [25]. S. Sain and their team reported work on synthesis of zeolite and grafted a metal complex Pd2+, Ni2+ for the synthesis of novel catalyst polycyclic hetrocycle for the application of suzuki—miyaura cross coupling reaction with good yield [26]. Nickel (II), Copper (II) and Oxovanadium (IV), complexes) anchored on the pore of Y-zeolite for the application of oxidation reaction with different reactant like styrene, cyclohexane and methyl phenyl sulfide with H2O2 oxidizing agent and author noticed that with maximum conversion achieved of methyl sulphide and high selectivity sulfoxide [27]. Ni (II), Fe (III), Cr (III), Bi (III), and Zn (II) complexes were tethered on cavity of Y zeolite and prepared catalyst is used in the application of oxidation of phenol with oxidizing agent H2O2 with reasonable selectivity of the product [28] and β-diimine Ni complex anchored on SBA-15 was also used for the application of ethylene oligomerization with 98% selectivity of the product [29]. Ni complex incorporated in acid functionalized Mesoporous moiety for photocatalytic degradation of methylene blue under visible light are found to show 92% degradation of organic pollutants (MB) in 15 ​min [30].Epoxides are tiny molecules that react with various nucleophiles, including alcohols, amines, water, and others [31–37]. In presence of amines, the bonds of epoxides are opened for the synthesis of β-amino alcohols. However, they can undergo regioselective ring opening of oxiranes with creation of a large variety of attractive intermediates for the synthesis of significant biological impulsive substances and synthetic products with chiral auxiliaries [38]. β-amino alcohols have established themselves as important components of medical, pharmaceutical chemistry and biological treatment [39–43]. Numerous attempts have been includingmade by researchers to improve the electrophilic nature of epoxides through metallic coordination or H bonding [44]. For this aim, several techniques have been utilized include the employment of transition metal halides, alumina, metal amides, metal triflates, alkali metal perchlorates, silica under high pressure, and montmorillonite clay [45–52] etc.Other reports are also available on heterogeneous catalysis of epoxide ring opening reaction. M. Mohammadikish et al. prepared MOF NH2-MIL-88B(Fe) (Fe-MOF) and post-synthetic modification was done by sulphuric acid. The prepared catalyst was used in epoxide ring opening reaction with 100% conversion of the reactant [53]. S.S. Mortazvai and their team have reported synthesis of multi-functional acid functionalized porous support anchoring ionic liquid, and the prepared heterogeneous catalyst was used in epoxide ring opening reaction of styrene oxide [54]. M. Magre et al. prepared magnesium based catalysts Mg-Bu2 and Mg(NTf2)2 which were applied in ring opening of a range of epoxides with excellent yield in secondary and tertiary alcohols [55]. N. Deshpande et al. synthesized a series of heterogeneous Lewis acid catalysts for epoxide ring opening reaction [56]. K. N. Tayade reported a work on zirconium triflate grafted mesoporous SBA-15 (ZrTf/S) and applied in epoxide ring opening reaction of amines and alcohol under ambient condition [57]. Mesoporous TiO2–Fe2O3 with strong catalytic activity was used, according to Roy et al. to prepare amino alcohols and benzimidazole derivatives [58].This work was motivated by the extensive use of organic group-modified materials and transition metal complexes immobilized on the surface of porous materials as catalysts in important industrial reactions. Our groups have carried out thorough study on inorganic/organic hybrid catalysts like acid/base functionalized, mono, bi and tetranuclear Zn, Ni based complex anchored mesoporous/layered materials [21,30,59–62] on epoxide ring opening of propylene oxide and cyclohexene oxide with piperidine, morpholine, aniline, etc. Using a well-established synthesis protocol, tetranuclear Ni complex has been used to anchor onto the surface of thiol-functionalized Na–Y zeolite. Different physiochemical techniques were used to characterize the prepared hybrid materials to verify their shape, crystallinity, phase purity, and functionality in the material. The catalytic activity of tetranuclear Ni complex immobilized thiol functionalized Na–Y zeolite (Na–Y–S–Ni4) was investigated for ring breaking of cyclohexene oxide. Maximum reactant conversion and product selectivity were achieved by fine-tuning of the reaction parameters.Na–Y zeolites was obtained from Sud-Chemie-India Ltd. The received Na–Y was calcined at 540 ​°C for 5 ​h to remove moisture content.For the modification process the calcined Na–Y zeolite and the silane reagent 3-mercaptopropyltrimethoxysilane (MPTES) were mixed in 30 ​ml of anhydrous dichloromethane. The salination was accomplish for one day at room temperature under stirring condition. The final product was rinsed multiple times with anhydrous dichloromethane and dehydrated at 100 ​°C [63].The tetranuclear Ni complex was synthesized as per the past reported technique [30].An appropriate amount of thiol-modified materials was activated. After that, thiol modified material and metal complex [Ni4(LH)4]·CH3CN (I) (LH3 = (E)-2-(hydroxymethyl)-6-(((2-hydroxyphenyl)imino)methyl)phenol)) was mixed in presence of anhydrous toluene solvent at 110 ​°C for 18 ​h. The hybrid catalyst was separated, washed thoroughly and dehydrated at 80 ​°C and depicted in Scheme 1 [21].The catalytic performance was evaluated in a round bottle flask equipped with a condenser under stirring. Na–Y zeolite as a catalyst had been modified with a thiol organic group (Na–Y–SH) and Ni4 complex was heated to 80 ​°C for an hour before the reaction. Over Na–Y–S–Ni4 materials, various electrophilic ring breaking reactions of cyclohexene oxide with amines were carried out. Gas chromatography (Scientific Thermo Fisher Trace 1310, TG 5 column) was used to analyze the reaction's end product. Multiple reactions were carried out in an effort to improve the reaction conditions for maximizing conversion and selectivity. Pyrrolidine and piperidine were used as nucleophiles for ring opening of cyclohexene oxide shown in Scheme 2 , over Na–Y–S–Ni4. The optimized reaction conditions are as follows: 20 ​mg catalyst, 6 ​h reaction time and 70 ​°C for the pyrrolidine system and 30 ​mg catalyst, 6 ​h reaction time and 80 ​°C for the piperidine system.The prepared material was analyzed by different analytical techniques such as scanning electron microscope, powder XRD, N2 sorption, nuclear mass spectroscopy, and thermogravimetric analysis. The morphology and elemental indexing of elements present in the material were analyzed by using Zeiss Ultra-55 machine. Crystallinity of the materials was examined by powder XRD by using empyrean equipment with graphite monochromatized CuKα radiation at 40 ​kV and 30 ​mA. The surface area, pore volume and diameter were determined by equipped Autosorb iQ S/N: 1050013927 Station: 1 ​at 77 ​K. Before analysis, the samples were degassed at 400 ​°C for 16 ​h with flowing N2. The MAS NMR spectra of 29Si and 13C were measured on a Varian 600 ​PS solid NMR spectrometer with a spinning frequency of 7 ​kHz, a 90° pulse length of 5.6 ​μs, and a cycle delay time of 5 ​s. Hexamethyl benzene and 3-(trimethylsilyl) propionic-2,2,3,3-d4 acid sodium salt was implemented as references for the 13C and 29Si chemical shifts. Using a Hitachi STA 7200 equipment, thermogravimetric analysis (TGA) was measured to examine the thermal constancy. At room temperature, IR spectra of modified materials were recorded using an FT-IR spectrometer (PerkinElmer, Spectrum Two) with a resolution of 4 ​cm−1. For elemental analysis, ICP-OES was carried out utilizing the Seiko SPS7000 instrument.X-ray diffraction pattern of Na–Y zeolite, Na-Y-SH zeolite and Na–Y–S–Ni4 was logged up to 2θ value of 10–50° and it is found to be well matched from the literature [64] and depicted in Fig. 1 . The XRD of Pure Ni4 complex was compared with Na–Y–S–Ni4 (Fig. 1 B) and two peaks at 11.2 and 25.6 2θ value are matching with each other. However, there are clear overlapping of the peaks of Na–Y zeolite with the pure complex. Very less amount of metal complex was incorporated to solid support as well as it is anticipated that the guest moieties are homogeneously dispersed so it is difficult to get distinct peaks for complex in the NaY–S–Ni4 material in X-ray diffraction patterns. The successful immobilization of Ni4 was confirmed by ICP analysis and SEM EDX, FT-IR etc. Peak intensity decreased when the Na–Y zeolite was functionalized with the thiol group and Ni complex, this could be because the metal complex and thiol functionality were tethered to the support's channel. Fig. 2 (A and B) illustrates the SEM image of the catalyst. Based on the findings of the SEM study, it is apparent that Na–Y zeolite appears as granules with a cuboid-shaped surface. After the modification of Na–Y zeolite, the morphology sustain.The average particle size of the materials is found to be 0.5 ​μm approximately. The EDX analysis of a particular area of Na–Y–S–Ni4 is depicted in Fig. 2(C) and the occurrence of C, O, Na, Al, Si and Ni signals in EDX mapping confirm existence of Ni complex in the modified Na–Y zeolite.To determine the functional group present in the material and bond formation between the metal complex to the surface of a solid support, FT-IR was performed. FT-IR spectra Na–Y, Na–Y–S–Ni4 and pure Ni complex were shown in Fig. 3 (A). The vibrational bands at 3423 ​cm−1 and 1620 ​cm−1 are responsible for O–H stretching of silanol groups' hydrogen bonds [65]. Two strong bands at 998 ​cm−1 and 788 ​cm−1 were noted as responsible for symmetric and asymmetric stretching vibration of the Si–O–Si, [66]. Presence of peak at 1375 ​cm−1 which is refer to –CH2 comes from the propyl chain that confirms the successful functionalization of thiol group in the material [67]. Another two peaks were observed at 737 ​cm−1 and 855 ​cm−1 corresponding to Ni–O and Ni–OH [68,69] respectively. The above bands confirm the existence of Ni complex in the hybrid catalyst.To determine the thermal stability of the materials, thermogravimetric analysis was examined. The TGA of Na–Y, Na–Y–S–Ni4 and Ni complex are depicted in Fig. 3(B). In pure sample Na–Y only one stage of the weight (4.5%) was observed at 348 ​K which is related to humidity. There are two stages of the weight loss observed in Na–Y zeolite, the first stage of weight loss at 336 ​K was responsible for the moisture present in the material and the second stage of weight loss at 425 ​K was responsible for the functional group/metal complex present in the material. In case of pure Ni complex, steady weight loss was observed above 700 ​K. The overall weight loss Ni complex incorporated Na–Y zeolite is 26% and Ni complex is 70%. From the analysis it is clear that the thermal stability of the pure Ni-complex enhances after incorporation in porous supportIn order to determine the textural and structural features of Na–Y nonstructural zeolite, N2 adsorption-desorption analysis was evaluated. N2 adsorption-desorption isotherms of Na–Y zeolite and Na–Y–S–Ni4 are shown in Fig. 4 (A) which exhibits type I isotherm that demonstrates the preservation of a microporous structure of Na–Y zeolite. After the modification, pore volume decreases due to the immobilization of the organic group and metal complex into the channel which confirms the successful functionalization and anchoring of the metal complex into the functionalized surface. The surface area, pore volume and pore diameter of modified and unmodified Na–Y zeolite are given in Table 1 and pore size distributions curve given in Fig. 4(B).The ICP-OES method was used to determine the metal content, which is summarized in Table 1. The Si/Ni ratio in Na–Y–S–Ni4 is found to be 160, indicating that the tetranuclear Ni complex has been immobilized in the functionalized Na–Y zeolite. Fig. 5 (A) displays the 29Si MAS NMR spectroscopy of Ni complex anchored thiol-functionalized Na–Y zeolite materials. In 29Si MAS NMR of Na–Y–S–Ni4, different silaonic environments like Q4 (4Si, 0Al), Q3 (3Si, 1Al), Q2 (2Si, 2Al) and Q1 (1Si, 3Al) chemical shifts at -105, -100, -94 and -89 ​ppm respectively were detected [70]. In order to approve the effective functionalization and metal complex incorporation in Na–Y zeolite 13C CP MAS NMR was also examined and shown in Fig. 5(B). Peaks at 15, 21 and 45 ​ppm are assigned for C1, C2 and C3 carbon groups from the propyl chain attached in MPTES [71] indicating of successful anchoring of thiol group. Peaks in the range of 120–200 ​ppm attributed for the phenyl benzene ring attached with the ligand was also present in the NMR spectra [61].After the confirmation of the presence of organic functionality and Ni4 complex, the catalytic activity of the prepared materials was checked in epoxide ring breaking reaction with various nucleophiles. Pyrrolidine and piperidine are used as a nucleophile for ring opening of cyclohexene oxide. The catalytic activity of the Na–Y–S–Ni4 is explained in Fig. 6 . Fig. 6 llustrates that the metal complex-anchored materials are more effective than the thiol-functionalized materials and pure Na–Y zeolite in the epoxide ring breaking reaction with 20 ​mg of catalyst for the pyrrolidine system and 30 ​mg of catalyst for the piperidine system. Catalyst amount, time and temperature are important parameters for optimizing the reaction system. In comparing the activities of unmodified Na–Y zeolite, thiol-functionalized Na–Y zeolite and Na–Y–S–Ni4, it was found that the conversion of pyrrolidine and piperidine increased dramatically and exhibited 89% and 90% respectively. In contrast, the conversion of pyrrolidine and piperidine in blank reactions was only 12% and 24%, and 24% and 33% with thiol-functionalized respectively. There was a clear increase in catalytic activity after Ni complex immobilization with thiol-functionalized Na–Y zeolite, suggesting that Ni complex had outstanding catalytic characteristics. Electrophilic activation, or the sturdy electron deficiency of the Ni+2 ion delocalizing the negatively charged oxygen atom in the intermediate state, intensely boosts the catalytic activity of Na–Y–S–Ni4, making the hybrid material highly efficient.The catalyst dosage was changed from 10 to 20 ​mg to establish the catalytic system using various nucleophiles at temperature of 70 ​°C for 6 ​h reaction period in the pyrrolidine system and 20–30 ​mg at 80 ​°C for 6 ​h reaction period in the piperidine system to achieve the highest conversion. As the quantity of catalyst increases, the reactant's conversion also rises, as illustrated in Fig. 7 . The conversion of pyrrolidine and piperidine were shown in Fig. 7(A and B). From 10 to 20 ​mg of catalyst, pyrrolidine conversion improved from 50 to 89% and from 20 to 30 ​mg of catalyst, piperidine conversion improved from 76 to 90%. However, after increase in catalyst amount from 20 to 25 ​mg in pyrrolidine reaction system 10% decrease in conversion of reactant was observed and increase in catalyst amount from 30 to 35 ​mg in piperdine reaction system 3% decrease in conversion was observed.After doing all the reactions with different catalyst amounts in both pyrrolidine and piperidine systems it is noticed that 20 ​mg catalyst is an optimized catalyst for pyrrolidine system and 30 ​mg catalyst for piperidine system as this amount of catalyst gives maximum conversion of reactant.Reaction kinetics is a key determinant of catalytic behavior for a reaction, and the kinetics is investigated for all the reactions. The reaction time was varied from 6 to 10 ​h in both pyrrolidine at 20 ​mg of catalyst, 70 ​°C and piperidine at 30 ​mg of catalyst and 80 ​°C, as shown in Fig. 8 (A and B). Epoxide ring breaking of cyclohexene oxide with pyrrolidine for 6 ​h, 89% conversion was observed, in 8 ​h reaction time, 75% and in 10 ​h reaction time, 70% conversion was observed. The conversion of pyrrolidine falls down as reaction time increases.During epoxide ring breaking of cyclohexene oxide with piperidine for 6 ​h, 90% conversion was noticed, while in 8 ​h 94% and in 10 ​h 70% conversion was observed. With the rise in reaction time from 6 to 8 ​h conversion also increased but with further increase in reaction time from 8 to 10 ​h, decrease in conversion of piperidine was observed. After performing the reaction in both the systems, 6 ​h time is optimized to get maximum conversion.The reaction temperature was varied from 60 to 75 ​°C for pyrrolidine and 70 to 100 ​°C for piperidine taking 20 ​mg and 30 ​mg of catalyst and 6 ​h reaction time is shown in Fig. 9 . At low-temperature, the reaction's conversion is also low; after an increase in temperature, conversion of the reactant also increases and further increase in temperature 90 to 100 ​°C, conversion of the reactant decreases. At 60 ​°C, 74% pyrrolidine conversion was observed, at 65 ​°C, 70 ​°C and 75 ​°C, 79%, 89% and 80% conversion of pyrrolidine was observed. After varying the reaction temperature 70 ​°C was considered as optimized reaction temperature for this reaction system.Similarly, with piperidine, at 70 ​°C, 83%, 80 ​°C 90% and 90 ​°C 99% conversion of piperidine was observed. The optimized reaction temperature for piperidine system is taken as 80 ​°C because a reasonable conversion was achieved (90%) at that temperature. Reaction at higher temperature was avoided at the cost of conversion.This reaction involves ring opening of cyclohexene oxide using Na–Y–S–Ni4 catalyst. In the first step, the bond breaking of the cyclohexene oxide takes place due to high ring tension. In the subsequent step, the catalyst got attached to oxygen to form an intermediate complex. Later, the lone pair of the nucleophile attacks the carbocation center of the intermediate complex leading to the formation of bond between them. Intermolecular proton transfer takes place from nitrogen to oxygen to form product leaving the catalyst free. The reaction mechanism between cyclohexene oxide with pyrrolidine and piperidine are shown in Scheme 3 .The reusability of the prepared Na–Y–S–Ni4 catalyst was also investigated as displayed in Fig. 10 . The catalytic activity of the material was persistent even after 4th run and no drastic change in reactivity was observed.FT-IR and powder XRD, was also performed to check the structural integrity of the spent catalysts after the reaction, and the findings are displayed in Fig. 11 (A and B). The well resolved peaks were found in XRD after reusing the catalyst. The peak intensity has significantly decreased after repeated use of the catalyst may be due to vigorous washing and activating at high temperature. However, all the signature peaks were present even after repeated use. FT-IR analysis also shows that the functionality of the catalyst was restored, including characteristic peaks of the parent materials.In this work, the synthesis of tetranuclear Ni complex and grafting of the complex in thiol functionalized Na–Y zeolite is described. Powder XRD, FT-IR, SEM-EDX, N2 sorption analysis, thermogravimetric analysis (TGA) 13C CP and 29Si MAS NMR confirm the successful grafting of Ni complex. No change in morphology was observed after modification with the preservation of crystallinity, phase purity. The existence of functional groups and Ni-complex in Na–Y zeolite was confirmed by FT-IR and NMR analyses reveals the successful immobilization of the material. Finally, the modified catalyst was used for the synthesis of β-amino alcohol by epoxide ring opening of cyclohexene oxide with pyrrolidine and piperidine. The designed hybrid catalyst has found to be extremely active and truly heterogeneous in nature and can be used up to 4 times without losing much of activity.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Madhu Pandey acknowledges financial support to SHODH (Scheme of Developing High quality research). The authors acknowledge Dr. Ashish Unnarkat of Pandit Deendayal Energy University, Gandhinagar and Dr. Dheeraj Kumar Singh of Institute of Infrastructure Technology Research and Management (IITRAM), for assistance in TG analyses and FT-IR, respectively.

In this present work, synthesized tetranuclear Ni complex is anchored on thiol functionalized Na–Y zeolite. Structure, phase integrity, shape, thermal stability and the existence of functional groups are identified by using different analytical techniques like powder XRD, N2 adsorption-desorption, Fourier transformed infrared spectroscopy (FT-IR), Nuclear magnetic resolution (29Si MAS NMR and 13C CP MAS NMR), scanning electron microscope (SEM-EDX), thermogravimetric analysis etc. The prepared hybrid catalyst Na–Y–S–Ni4 shows remarkable catalytic property by promoting the formation of high-yield products like β-amino alcohol by electrophilic ring opening reaction. 89% and 90% conversion of pyrrolidine and piperidine with high selectivity of the desired product are observed in ring breaking of cyclohexene oxide.

Dyes are mainly used in the plastics, textile, paper, rubber, wood, and food industries [1]. The presence of colorant materials in streams is not only aesthetically unpleasant but can also cause problems for the environment by elevating toxicity, oxygen demand levels, and lagging the photosynthetic phenomena by reducing light penetration [2,3]. Dyes belonging to the groups based on anthraquinone, azo, and triphenylmethane are the main pollutants released from industrial effluents [4,5]. Most of the dye laden-effluents are produced by the textile industry, where 10–15% of the dye is wasted during the production process [4,6]. One of the dyes released from the textile is phenol red. Its solubility in water and ethanol is 0.77 g/L and 2.9 g/L, respectively [7]. Phenol red dye is water-soluble and can be utilized as a pH indicator in various laboratories [8,9]. This dye is used in industries such as pharmaceuticals, textiles, and chemicals, and the wastewater discharged from these industries must be properly treated [9].Various physical, chemical, and biological methods have been tested to eliminate dyes. Physical and chemical methods can be used on a small scale [10]. Advanced oxidation methods (AOMs) have been considered by researchers for cleaning effluents containing resistant organic compounds [11,12]. AOMs are based on the production of active species for degrading pollutants using solar, chemical, or other forms of energy [13,14]. The photocatalytic oxidation process is a superficial phenomenon and light reactions happen on the catalyst surface like TiO2 [11,15]. Titanium dioxide (TiO2) has so far been considered comparable to other semiconductors for many reasons, including its cheapness, availability, safety, and good reactivity [11,16]. Also, it can degrade organic compounds and as a result, the pollutants are converted to H2O and CO2 [16]. Nickel is also a useful photocatalyst that has low cost and good stability in alkaline solutions [17]. However, it is difficult to use titanium dioxide and nickel alone because their separating from the solution is a time-consuming and costly process. Therefore, the photocatalyst should be stabilized on a support surface. Many supporting surfaces like activated carbon [18], graphene oxide [19], silicate [20], and glass spheres [21] have been used for titanium dioxide and nickel oxide.The FSM-16 powder has outstanding features like high thermal stability, high specific surface area, and high pollutant sorption capacity, which makes it a suitable option for photocatalyst support. Although FSM-16 has been used to support photocatalysts to decompose some pollutants [22,23], its application as support of titanium dioxide and nickel oxide in dye removal is not yet clear.Accordingly, the main aim of this study was to evaluate the photocatalytic activity of TiO2-FSM-16 and Ni-FSM-16 in phenol red degradation. Phenol red decomposition using TiO2-FSM-16 and Ni-FSM-16 was examined in the presence of UV rays and the effect of operating parameters was studied. The characteristics of photocatalysts, reaction kinetics, catalyst reusability, leaching of the photocatalyst components, and synergy of process components have also been investigated. In another part of the work, the dye removal efficiency of TiO2-FSM-16 and Ni-FSM-16 was compared with TiO2-carbon and Ni-carbon.Acetyl trimethylammonium bromide (C19H42BrN, purity >99.9%), hydrochloric acid (HCl, 37 wt%), sodium hydroxide (NaOH, purity >99.9%), silicon dioxide (SiO2, purity >99.8%), ammonium titanyl oxalate monohydrate ((NH4)2TiO(C2O4)2·H2O, purity: 98%), ethanol (C2H6O, purity ≥99.9%), nickel nitrate (Ni(NO3)2, purity> 99.8%) were purchased from Merck Company. Phenol red (C19H14O5S, purity: 98%), tert-butanol alcohol (C₄H₁₀O, purity ≥99.5%), p-benzoquinone (C6H4(=O), purity >98%), silver nitrate (AgNO3, purity ≥99%), and ammonium oxalate ((NH4)2C2O4·H2O, purity ≥99.99%) were provided from Sigma-Aldrich Company. The stock solution of the phenol red dye was prepared daily in doubly distilled water (DDW). All working solutions were prepared by diluting the stock with DDW.Morphology of FSM-16, TiO2-FSM-16, and Ni-FSM-16 was done using the TeScan-Mira III Scanning Field Emission Microscope (FE-SEM, Czech Republic). Nitrogen absorption-desorption test was performed by BET instrument, Micrometric model ASAP2020 (USA). The XRD pattern was taken by the PANalytical manufacturer, Xpert-pro model (USA) at 2θ equal to 0.5–60° and CuKα of 1.5406 nm. Phenol red content was determined by UV–vis spectrophotometer, PerkinElmer, Lambda 25 (USA) at λ max 423 nm. An example of a calibration curve to determine the concentration of phenol red is presented in Fig. S1 in the Supplementary Information. The amount of dye mineralization was measured by the TOC device (Shimadzu, model VCSH, Japan).First, 3.8 g NaOH was added to 100 mL of DDW and 6 g of SiO2 was poured into the solution to make the SiO2/Na2O ratio equal to 2. It was then stirred for 3 h under normal conditions at ambient temperature (27 °C) to obtain a uniform solution. To remove the available solvent, the sample was placed in a rotary evaporator balloon. The evaporator temperature was set to 100 °C and the speed was 30 rpm. Approximately 60–70 mL of water was evaporated and resulted in a jelly-like liquid of sodium silicate. Finally, the sample was poured into a porcelain plate and transferred to a furnace to complete the synthesis and calcination process of layered sodium silicate. The furnace was programmed with a temperature rate of 2 °C/min at 700 °C for 6 h. The calcined sample was layered sodium silicate or Kanemite. This material was crushed in a mortar and 4 g of it was poured into 40 mL of DDW and agitated for 3 h at 27 °C. After the preparation of Kanemite paste, which was used as a source of silica in the synthesis of regular mesopores of FSM-16. Through a mechanism, the Kanemite silicate plates were folded and cross-linked to form a three-dimensional structure. To make a wet Kanemite paste, the resulting solution was filtered. The Kanemite was dispersed in 88 mL of n-hexadecyl trimethyl ammonium bromide solution (0.125 mol/L) and stirred for 3 h at 70 °C. In this step, pH should be adjusted between 11.5 and 12.5. After that, the pH was decreased to 8.5 using 2 mol/L HCl. The suspension solution was stirred for 3 h at 70 °C. The produced material was then washed at room temperature with a liter of distilled water and air-dried. The synthesized meso-cavity was placed in the Soxule device for washing. Washing was done with 20 mL of ethanol and 0.5 mL of HCl per gram of material. The meso-cavity material was placed in a Soxule device for 72 h. The material was then dried for 24 h at ambient temperature. Finally, it was placed in the furnace (950 °C, 9 h) to remove the surfactant [24].To deposit titanium dioxide on FSM-16, a solution with a concentration of 0.1 M of titanium oxalate ammonium salt was prepared. Then, 2 g of FSM-16 was added to 25 mL of titanium solution. The sample was stirred at room temperature for 24 h. After filtering and washing the samples with DDW and air-drying, the samples were calcined at 450 °C for 12 h [25].To make the Ni-FSM-16 catalyst, initially, 1 g of FSM-16 mesoporous silica was dissolved in 20 mL of normal hexane and stirred at room temperature for 2 h (Solution A). The amount of 0.5675 g of nickel nitrate was added to another vessel with 1.14 mL of DDW (Solution B). After 2 h, Solution B was added dropwise to Solution A and the new mixture was stirred for 3 h. It was then air-dried for 24 h. The solids were subjected to argon gas for calcination at 550 °C for 5 h. The gradient for increasing and decreasing the temperature at this stage was 2.5 °C/min and the flow rate of argon gas injection into the furnace was 2 L/h [26].To make TiO2-carbon and Ni-carbon, the same methods as TiO2-FSM-16 and Ni-FSM-16 were done. It should be noted that activated carbon in this study was purchased from Merck and according to the supplier's information, it had an active surface of 810 m2/g.Four factors were analyzed by the response surface method. The effect of four experimental variables, namely TiO2-FSM-16 and Ni-FSM-16 quantity, solution pH, UV irradiation time, and dye concentration on the photocatalytic removal of phenol red was explored. The variables and their upper and lower levels are listed in Table S1 in the Supplementary Information.A total of 30 tests were performed, including 16 experiments at factorial points, 8 experiments at axial points, and 6 replications at central points.The efficiency of dye removal was calculated using the following equation: (1) D y e r e m o v a l ( % ) = ( [ D y e ] i n i t i a l − [ D y e ] r e s i d u a l ) [ D y e ] i n i t i a l x 100 The photocatalysts (TiO2-FSM-16 and Ni-FSM-16) were recovered after being used in dye removal. The photocatalysts were separated using a centrifuge, washed (with water and ethanol), dehydrated (at 105 °C), and then reused for dye degradation. The reusability tests were done 7 times.Compounds of tert-butanol (500 mM), p-benzoquinone (10 mM), silver nitrate (10 mM), and ammonium oxalate (50 mM) were used as scavengers of active species and radicals in the process of phenol red removal [27]. This test was performed at optimal conditions (pH: 3, dye concentration: 20 mg/L, catalyst dose: 2 g/L, time: 120 min).The micrographs (FESEM) of the FSM-16 are presented in Fig. 1 A, B. Based on the FESEM graphs, nanoparticles with an approximate size of 200 nm were observed. In the study conducted by Hashemi et al., the FSM-16 nanoparticles have a spherical shape with a size of 100 nm [28]. The FESEM images of TiO2-FSM-16 are shown in Fig. 1C,D. As this figure shows, the photocatalyst particles are non-uniform and spherical. TiO2-FSM-16 particles have a similar shape to those reported in other works [22]. This particle size distribution with spherical morphology corresponded to van der Waals forces [29]. To decrease the surface energy, the primary particles tend to condense, with the formation of spherical masses, in the minimum surface-to-volume ratio, the minimum free surface energy can be obtained. The TiO2-FSM-16 composite powder is composed of nanoscale particles, which indicates that the prepared powder has a large specific surface area and volume. Therefore, the TiO2-FSM-16 composite can provide suitable active sites for photocatalysis. Fig. 1C,D shows that TiO2 particles are dispersed on the surface of SiO2 nanoparticles and have good stability.The shape and size of crystals in Ni-FSM-16 are slightly different from TiO2-FSM-16 (Fig. 1E,F). The Ni-FSM-16 particles have a small size and are aggregates of spherical microcrystal particles. Silicate plates are a combination of tens to hundreds of hexagonal cavities composed of Ni-FSM-16 catalyst particles. Adhesion between particles may be due to the small magnetism or the polymer between them [30,31].The XRD image of FSM-16, TiO2-FSM-16, and Ni-FSM-16 is shown in Fig. 2 . A wideband with an angle equal to 20–30° can be seen in the FSM-16 spectrum, revealing that the material is amorphous and has no specific crystalline peaks. The disappearance of the higher-order diffraction peaks indicates that the hexagonal arrangement of the channels in the TiO2-FSM-16 and Ni-FSM-16 mesoporous materials is slightly irregular. Therefore, it can be concluded that the addition of metal in the mesoporous structure of FSM-16 has a negative effect on the crystallinity of the material. This is because the use of metal ores weakens the self-assembly process and produces a less regular meso-cavity structure. In the XRD pattern of TiO2-FSM-16 and Ni-FSM-16, in addition to the four distinct peaks of the FSM-16 nanoparticles, the characteristic peaks of TiO2 and NiO are also seen, indicating no noticeable change in the crystal structure of the photocatalysts [32]. The low-angle XRD image of FSM-16 showed that the diffraction plate has a regular hexagonal structure [33]. An anatase-like reference pattern (JCPDS # 21–1272) was seen in TiO2-FSM-16 [34]. Five peaks at 37, 43, 48, 52, and 58 in Ni-FSM-16 photocatalyst are related to Miller indices of (111), (200), (220), (311), and (222), respectively, which confirms that the photocatalyst contains NiO [35–37]. Based on the literature [38], a main reflection at 2theta of 43.5°, corresponds to the (200) plane of cubic NiO (PDF-2, 01–071–1179).The active area of the photocatalysts was computed using Brunauer-Emmett-Teller (BET) technique and the pore diameter and pore volume were calculated by the BJH technique (Fig. 3 a–c). For the FSM-16 sample, the obtained BET was 1099.08 m2/g, and the absorption isotherm is almost synchronous with the desorption branch, indicating the mesoporous structure of the sample. Fig. 3b and c shows a similar pattern of BET for TiO2-FSM-16 (844.93 m2/g) and Ni-FSM-16 (718.63 m2/g), showing that the mesoporous feature of FSM-16 did not alter after the composition with Ni and TiO2 [39].The specific surface area of Ni-FSM-16 and TiO2-FSM-16 was lower than that of FSM-16, which is linked to the occupation of pores by Ni and TiO2 (Fig. 3). However, the level of synthesized photocatalyst (844.93 m2/g) is much bigger than the values stated for other photocatalysts in the literature. For example, Fatima and Supia prepared the TiO2-MCM-41 photocatalyst with a surface of 400.7 m2/g [40]. Sugiyama et al. have incorporated Ni into FSM-16 and MCM-41 structures to produce Ni-FSM-16 and Ni-MCM-41 and reported that the active surface of photocatalysts was lower than that of the base material [41].ANOVA was utilized to determine the relationship between the removal rate and the variables. It is noteworthy that the judgment was based on the F-values and P-values for phenol red removal. The F-values of the model for TiO2-FSM-16 and Ni-FSM-16 were obtained at 16.77 and 12.16, respectively, indicating that the models are significant. The P-value was <0.05 indicating the significance of the model expression. In this case, the ‘A’ variable in TiO2-FSM-16 and ‘A, B, and D’ in Ni-FSM-16 were significant. In addition, the validity of the design with values of P and high value of correlation coefficients as well as the non-significance of “Lack of Fit” is confirmed [42]. P < 0.0001 reveals the importance of the model and the interaction of variables on phenol red elimination. The pH factor played an important role in phenol red removal using both photocatalysts (TiO2-FSM-16 and Ni-FSM-16) with the biggest F-value (60.56 and 133.57, respectively) compared to other variables (Table S2 and S3). To evaluate the quality of the models, the figures for the predicted response vs the actual values and the normal data distribution diagrams are shown in Fig. S2. The normal diagram of the residues is provided in Fig. S2a,c. Such a graph is very useful for optimizing complex systems such as multivariate optimization. The points are in a straight line and no deviations are seen in the distribution of data, indicating a suitable correlation and distribution between the values [43]. Since the specific trend is not related to variance changes (decrease or increase), the variance is fixed, which shows the scatter of points against the given values (Fig. S2b,d).Investigation of the effect of the desired catalysts (TiO2-FSM-16 and Ni-FSM-16) on phenol red degradation efficiency under ultraviolet radiation is depicted in Fig. S3a. In this figure, the positive and negative effects and the magnitude of the effect of each variable on the response are identified. The sharpness of the slope in a factor indicates that it is a vital variable in the reaction [5]. Oppositely, a relatively flat line shows the insensitivity of the response to alteration in the given variable. As can be seen in Fig. S3a, the TiO2-FSM-16 curve shows the time and dye concentration of the slow curvature, showing that these factors have a small effect on the response. Accordingly, the significant sloping curvature at pH indicates that the phenol red removal was sensitive to this variable. Fig. S3b shows that the curvature time and dye concentration have a slow slope, indicating that the mentioned factors have a small impact on the response.In Fig. 4 a–d, the response surface (3D surfaces) and contour diagrams (2D contours) are depicted as a function of pH and photocatalyst dose. For both generated materials (TiO2-FSM-16 and Ni-FSM-16), while the irradiation time (120 min) and the phenol red concentration (20 mg/L) were constant, the amount of degradation decreases with increasing pH. In other words, for both catalysts, the highest efficiency occurred at pH 3. At acidic pHs, the phenol red molecule has the zwitterion form with two functional groups (sulfate with a negative charge and ketone with a positive charge) with an additional proton. The FSM-16-TiO2 surface also has protonated functional groups at low pH. The dye molecule has positive and negative charges at low pH. Therefore, the maximum efficiency at the acidic pH of 3 is due to the electrostatic attraction between protonated groups and zwitterion negative charges [44]. If the pH of the solution is higher than the pKa of phenol red (about 7.9), the proton of the ketone group is lost and the molecule takes on a negative charge [45]. The higher the pH, the more negative the catalyst level becomes (pHzpc of both photocatalysts was about 5.4). In such a situation, the dye molecule is rejected from the surface of the catalysts, and as a result, the removal efficiency decreases. Kumar et al. [46] reported that at acidic pHs, the surface of titanium dioxide becomes positive, leading to higher dye removal.With the increasing load of TiO2-FSM-16, a slight change in removal efficiency was observed. This has been linked to the tendency of particles to accumulate, which reduces the BET area of the catalyst and leads to less production of active radicals [47,48]. For the Ni-FSM-16, increasing the photocatalyst mass did not play a role in increasing the efficiency and the graph was smooth. Higher content of TiO2-FSM-16 than Ni-FSM-16 can lead to more radical production and thus higher efficiency [25]. Similar observations have been reported by researchers for degrading cephalexin using NiS and NiS-support Fe3O4@PPY photocatalysts [49].The effect of treatment time and photocatalyst dose on the phenol red elimination is depicted in Fig. 5 a–d. As shown in Fig. 5, increasing the dose of TiO2-FSM-16 photocatalyst has resulted in a slow increase in the removal efficiency, while with increasing Ni-FSM-16 the efficiency has remained almost constant. With more time of exposure of TiO2 nanoparticles to light, the dye removal efficiency increases. Longer exposure to light implies the production of more hydroxyl radicals, which are responsible for the oxidation of the phenol red dye molecule [50].With increasing filtration time, the amount of dye removal by the TiO2-FSM-16 photocatalyst increased due to the sufficient opportunity for radicals to attack the dye. But with increasing time, the removal efficiency by Ni-FSM-16 decreases, which is probably due to the desorption of absorbed dye from the catalyst surface. So, it can be confirmed that the Ni-FSM-16 photocatalyst may not have been able to oxidize the dye but rather remove it by the adsorption process.A recyclable photocatalyst would be economically and environmentally beneficial. The photocatalyst recovery results are shown in Fig. 6 a. Based on Fig. 6a, the reusability behavior of the two produced materials was different. The TiO2-FSM-16 photocatalyst has higher usability and good strength. TiO2-FSM-16 catalyst could be reused up to 3 times, while Ni-FSM-16 had up to 2 times good removal efficiency. The reduction in the efficiency of the recycled catalyst can be due to the leakage of its effective components. In these two photocatalysts, nickel and titanium are important components. The results of Fig. 6b show that these elements have leaked from the photocatalyst. Another significant point is that the amount of nickel leakage is lower than the drinking water standard (100 μg/L). Among the two prepared catalysts, the amount of leakage in TiO2-FSM-16 was lower, which indicates its stability and, as a result, its higher efficiency.In Fig. 7 , the effect of system components on the removal efficiency of phenol red dye is compared with the whole system. As depicted in this figure, the efficiency of the system components and even UV along with nickel and titanium dioxides were much lower than the whole system. FSM-16 has been reported to have a 51% removal of phenol red, possibly due to the adsorption mechanism. The high surface of the FSM-16 material has also provided a suitable platform for dye adsorption. The synergy factor for the TiO2-FSM-16/UV was calculated based on the following formulas [51]: (2) S y n e r g y f a c t o r T i O 2 − F S M − 16 = k o b s T i O 2 − F S M − 16 k o b s T i O 2 + k o b s F S M − 16 (3) S y n e r g y f a c t o r T i O 2 − F S M − 16 / U V = k o b s T i O 2 − F S M − 16 / U V k o b s T i O 2 + k o b s F S M − 16 + k o b s U V The synergistic factor for the TiO2-FSM-16 and TiO2-FSM-16/UV processes were computed at 1.55 and 2.12, respectively. These findings reveal the effective interaction of UV, TiO2, and FSM-16 components with each other that lead to elevated degradation of phenol red dye.Components like hydroxyl radicals (OH•), superoxide (O2 −•), photogenerated holes (hϑ+), and electrons (e−) can be effective in the photocatalytic dye removal process [52–54]. According to scientific papers [43,47], the compounds of tert-butanol, p-benzoquinone, silver nitrate, and ammonium oxalate are effective in abducting the active components of hydroxyl radicals (OH•), superoxide (O2 −•), electrons, and photogenerated holes (hϑ+), respectively. To clarify the decomposition mechanism, the effect of these scavengers on dye removal efficiency was explored and the results are drawn in Fig. 8 a. According to Fig. 8a, it is clear that the amount of dye removal has decreased with the addition of scavengers. The most severe decrease in phenol red removal efficiency was for ammonium oxalate, followed by silver nitrate, p-benzoquinone, and tert-butanol. This indicates that photogenerated holes (hϑ+) were the most effective species in the photocatalytic process of phenol red removal. Similar results have been stated for the catalytic elimination of p-aminophenol and methylene blue dye by TiO2/RGO catalyst [47] and Brilliant green by MgFe2O4 catalyst [43].The kinetics of the phenol red degradation process was evaluated under optimal conditions (see Fig. 8b). The first-order equation was utilized to assess the kinetic behavior of phenol red catalytic decontamination: (4) Ln [ f i n a l d y e c o n c . ] [ i n i t i a l d y e c o n c . ] = − k . t If the graph Ln [ f i n a l d y e c o n c . ] [ i n i t i a l d y e c o n c . ] vs time is drawn, a line is attained, and its slope is the reaction rate constant (k, min−1). As shown in Fig. 8b, the data follow first-order kinetics (R2 > 0.96). The phenol red photodegradation rate constant using TiO2-FSM-16 and Ni-FSM-16 was calculated at 0.028 and 0.018 min−1, respectively. The reaction rate with TiO2-FSM-16 photocatalyst was about 1.5 times Ni-FSM-16. The kinetics of dye removal by various photocatalysts have followed the first-order model [43,52,55].The amount of dye mineralization was investigated by two photocatalysts. The results showed that under optimal conditions (pH: 3, dye concentration: 20 mg/L, photocatalyst dose: 2 g/L, time: 120 min), the amount of dye mineralization by TiO2-FSM-16 and Ni-FSM-16 is 46% and 35%, respectively. It shows the effectiveness of the TiO2-FSM-16 photocatalyst in dye removal compared to Ni-FSM-16. The results of this study are in the range of values reported for acid orange removal using BiVO4/TiO2 in the presence of H2O2 and FeSO4 [56].Several methods have been reported so far for removing dyes. Among these methods, researchers emphasize the oxidation and adsorption of dyes. Table 1 compares the methods mentioned in scientific texts for dye removal with our method. As can be seen, concerning the conditions of the tests, the photocatalysts in this paper are among the satisfactory catalysts for dye removal.As it is clear in Table 1, the laboratory conditions of previously published work are different from our work, and an exact comparison cannot be made. Therefore, activated carbon as the most famous base for the catalyst was made under the same conditions as the catalysts of TiO2-FSM-16 and Ni-FSM-16 of this research and was investigated in the removal of the target pollutant. As shown in this table, the removal efficiency of carbon-based photocatalysts (BET: 810 m2/g) is lower than that of FSM-based ones. These differences are probably due to the lower surface area that carbon provided for the reaction compared to FSM.In this study, TiO2-FSM-16 and Ni-FSM-16 photocatalysts were produced and used to remove phenol red dye in the presence of UV light. The TiO2-FSM-16 and Ni-FSM-16 photocatalyst had a BET area of 844.93 m2/g and 718.63 m2/g, respectively. Maximum phenol red removal using TiO2-FSM-16 and Ni-FSM-16 was obtained at 84% and 66%, respectively under the conditions of pH 3, dye concentration of 20 mg/L, catalyst dose of 2 g/L, and irradiation time of 120 min. The photocatalyst activity of FSM-16 was increased after the corporation with TiO2 and Ni. The synergistic factor of TiO2-FSM-16 and TiO2-FSM-16/UV processes were found to be 1.55 and 2.12, respectively. The TiO2-FSM-16 photocatalyst with phenol red dye removal of 87% (TOC removal: 46%) had better activity than Ni-FSM-16 with 76% removal (TOC removal: 35%). Both photocatalysts, especially TiO2-FSM-16, had good capabilities in removing phenol red dye.Seyed Mohamadsadegh Mousavi: Conceived and designed the experiments; Performed the experiments.Seyed Hamed Meraji: Analyzed and interpreted the data; Wrote the paper.Ali Mohammad Sanati: Contributed reagents, materials, analysis tools or data; Performed the experiments.Bahman Ramavandi: Analyzed and interpreted the data; Wrote the paper.This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.Data will be made available on request.Supplementary content related to this article has been published online at [URL].The authors declare no conflict of interest.We are grateful to Bushehr University of Medical Sciences for access to laboratory equipment.The following is the supplementary data related to this article: Supplementary information-R1 Supplementary information-R1 Supplementary data related to this article can be found at https://doi.org/10.1016/j.heliyon.2023.e14488.

In this study, the performance of Ni-FSM-16 and TiO2-FSM-16 photocatalysts in phenol red removal was explored. The XRD, FE-SEM, and BET tests were used to characterize the catalysts. All experiments were performed at ambient temperature and under UV (20 W). The parameters including dye concentration (20–80 mg/L), photocatalyst concentration (0–8 g/L), UV exposure duration, and contact time (0–160 min) were optimized using RSM software. BET values of Ni-FSM-16 and TiO2-FSM-16 were 718.63 m2/g and 844.93 m2/g, respectively. TiO2-FSM-16 showed better performance in dye removal than Ni-FSM-16. At pH 3, the maximum dye removal by TiO2-FSM-16/UV and Ni-FSM-16/UV was obtained 87% and 64%, respectively. The positive hole species had the main role in photocatalytic phenol red removal. The reusability study was done for up to 7 cycles, but the catalysts can be reused effectively for up to 3 cycles. The synergistic factor for the TiO2-FSM-16 and TiO2-FSM-16/UV processes were calculated to be 1.55 and 2.12, respectively. The dye removal efficiency by TiO2-carbon and Ni-carbon was slightly lower than those obtained by the FSM-16 ones. The TiO2-FSM-16 and Ni-FSM-16 catalysts had a suitable surface and acceptable efficiency in phenol red removal.

Data will be made available on request.CH4 reforming with CO2 (also known as dry reforming of methane, or DRM) has attracted increasing interest in building a sustainable carbon-neutral society, as it can convert two main greenhouse gases into higher value fuels and chemicals while alleviating the negative effects of greenhouse gas emissions [1,2]. Moreover, since CO2 and CH4 are the main components of biogas, this process is also appealing for the efficient utilization of biogas without CO2 separation. The main product of this process - syngas (CO + H2) is an important feedstock for making chemicals and fuels (e.g., Fischer-Tropsch process and methanol synthesis) [3].The thermal-catalytic route has received the most attention for DRM research, and its feasibility has been demonstrated in small-scale industrial cases [4]. However, this reaction is highly endothermic because of the highly stable reactants (CH4 and CO2), and thus high temperatures (700–900 °C) are typically required to achieve reasonable performance. These harsh reaction conditions result in not only high energy input and operational costs, but also coke deposition and the deactivation of catalysts [5]. Therefore, it is of great interest to explore new processes for DRM under mild conditions.Non-thermal plasma (NTP) technology is an emerging approach for conventional chemical reactions (e.g., DRM) due to its unique non-equilibrium property [6,7]. NTP can be generated by gas discharges under mild conditions such as atmospheric pressure and near room temperature. NTP contains a large number of reactive species, including highly energetic electrons, excited ions, molecules, atoms and free radicals, which are the primary driving forces for thermodynamically unfavorable chemical reactions [8–11]. NTP-enabled processes can start-up and shut-down very quickly, compared to the long-time heating-up and cooling-down processes in thermal catalytic systems. This allows NTP processes to be easily powered by intermittent renewable energy sources (e.g., wind and solar energy) [12], reducing fossil fuel consumption and greenhouse gas emissions. In addition, a synergistic effect may be generated when catalysts are introduced into the NTP process under suitable conditions [13]. In this case, the catalysts are activated at low temperatures and stabilized by collisions between energetic electrons and other reactive species, thereby promoting reaction performance. These factors may provide a cost advantage for plasma-catalytic DRM over thermal catalytic route.Different NTP systems have been investigated for plasma chemical reactions, including dielectric barrier discharge (DBD), spark discharge, gliding arc, corona discharge and microwave discharge [14]. DBD is the most commonly used NTP in plasma chemical processes due to its simple structure and ease of plasma-catalyst coupling [15]. DBD coupled with catalysis has been extensively explored in a number of areas, including volatile organic compound (VOC) removal, CO2 and CH4 conversion, tar reforming, NH3 synthesis, etc [2,16–25]. Extensive research has been conducted on the plasma-catalytic DRM using DBDs, evaluating the influence of process parameters, catalysts, and reactor configurations on reaction performance. Nevertheless, the energy efficiency of this process is still insufficient for commercialization [15].The catalytic material is critical in the plasma-catalytic reforming process. Various packing materials with both non-catalytic and catalytic properties have been used for plasma-catalytic DRM, including glass beads, zeolites, metal oxides, supported catalysts with both transition and noble metals, as well as catalysts with perovskites and spinel structure [15,26–28]. Among these packing materials, Ni/γ-Al2O3 is the most commonly used catalyst in the plasma-catalytic DRM [13,29–31]. Recently, modified Ni catalysts by promoting the active metal and the support, as well as other supported catalysts (e.g., Cu/γ-Al2O3, Pt/γ-Al2O3, Ag/γ-Al2O3 and Au/γ-Al2O3) have also been investigated to enhance gas conversion, product selectivity and process energy efficiency [32–43]. According to previous research, the most important factors influencing the plasm-catalytic reforming performance are catalyst properties, discharge characteristics when coupling catalysts and the influence of plasma on catalysts. It is critical to connect these variables to reforming performance, especially gas conversion and energy efficiency, as well as the selectivity of some specific products. Although significant progress has been made in this field through both the experimental and simulation approaches [40,43,44], there is still much room to further investigate the relationship between reaction performance and plasma-catalyst coupling for DRM.Herein, we developed a coaxial DBD reactor with a water electrode for plasma-catalytic DRM into syngas, hydrocarbons and oxygenates. Based on previous research [13,36,42,45], the most common supported catalyst (Ni/γ-Al2O3) and other two noble metal catalysts (Ag/γ-Al2O3 and Pt/γ-Al2O3) were selected for this work. The properties of these catalytic materials were examined using a variety of characterization techniques. Electrical signals were used to investigate the discharge characteristics in plasma catalysis when using different catalysts. The reaction performance was evaluated using the conversion of CO2 and CH4, the selectivity of gaseous and liquid products, and energy efficiency for gas conversion and product formation. The relationship between catalyst properties, discharge characteristics and reaction performance was also discussed.The supported catalysts with different active metals (Ni, Ag and Pt) were prepared using the modified impregnation method with plasma treatment. The metal loading in these catalysts was 10 wt%, 1 wt% and 1 wt%, respectively. The metal precursors Ni(NO3)2∙6H2O, AgNO3 and H2PtCl6 were dissolved in deionized water, respectively, followed by adding the support (γ-Al2O3 beads with a diameter of 1 mm). The resulting mixtures were kept at room temperature for 12 h. After that, the solutions were then evaporated to dry at 80 °C, before being dried overnight at 110 °C. The obtained samples were treated with Ar/H2 DBD for 40 min. For plasma treatment, the input power and total gas flow rate were 40 W and 50 ml/min, respectively, with an Ar/H2 molar ratio of 4:1. The above prepared catalysts were denoted as Ni/γ-Al2O3, Ag/γ-Al2O3 and Pt/γ-Al2O3.The N2 adsorption-desorption isotherm experiments of γ-Al2O3 and the supported catalysts were carried out using a BELSORP MAX instrument. The specific surface area (S BET) and the total pore volumes (V p) were obtained at a relative pressure (P/P 0) of 0.99, while the average pore diameter (D p) was determined by the BJH method. The X-ray diffraction (XRD) profiles were collected by a Rigaku Smartlab diffractometer with a Cu-Kα radiation source. Transmission electron microscopy (TEM) analysis was carried out using a FEI Tecnai G2 F20 electron microscope. The CO2 temperature-programmed desorption (CO2-TPD) was performed using a Belcat instrument. Each sample was treated with He flow at 200 °C for 1 h prior to testing, followed by decreasing to 50 °C. The pure CO2 flow was then introduced for 1 h after the catalyst surface had been saturated. The CO2-TPD profiles were collected in a temperature range of 50–700 °C. Following the evaluation of catalytic performance, the used catalysts were analyzed by thermogravimetric analysis (TGA) in a simultaneous thermal analyzer (NETZSCH STA 449F3). Each sample was heated from 30 °C to 700 °C at 10 °C/min in a synthetic air flow. Fig. 1 shows the schematic diagram of the experimental setup. The main part of the DBD reactor contained two cylindrical quartz tubes and a stainless-steel rod. These two quartz tubes formed a casing tube with water circulating inside. A stainless-steel needle was inserted into the casing tube through a hole sealed with a rubber gasket. The needle made contact with the circulating water, allowing it to function as a low-voltage electrode. A reference capacitor (7800 pF) was connected between the low-voltage electrode and ground. The discharge length was the same as the length of the casing tube (5 cm). The flow rate of the circulating water was kept constant at 12 L/min using a water pump, and its temperature was controlled at 0 °C by a water-ice mixture. The inner quartz tube (8 mm i.d. × 10 mm o.d.) served as a dielectric layer. A stainless-steel rod (4 mm o.d.) was used as a high-voltage electrode, and was set coaxially with the inner quartz tube. Therefore, the discharge gap was fixed at 2 mm in this configuration. The discharge gap was fully packed with the catalyst particles. The DBD reactor was powered by a custom-built AC power source with a discharge frequency of 0–20 kHz and a peak-to-peak voltage of 0–30 kV. The input power of the plasma system can be controlled by changing the applied voltage and was monitored by a power meter. In this work, the input power was maintained at 70 W, and the discharge frequency was fixed at 10 kHz. The signals of the applied voltage, the voltage across the reference capacitor and the total current were sampled by a Tektronix digital oscilloscope (TDS-2014B) with a Tektronix high-voltage probe (P6015A), a Pintech differential probe (N1070A) and a Pearson current coil monitor (6585). The number of spikes in the current signals was calculated using the reported method [46,47]. The lifetime and magnitude of the current spikes were determined by the peak value and the full width at half maximum of these spikes [48]. Lissajous figures were used to determine discharge power, effective capacitance, and charge characteristics under different conditions [30,49]. The temperatures in the plasma region were measured by a thermal infrared camera (Fotric 223 s). These temperatures were lower than 70 °C under the experimental conditions in this work. The details of temperature measurement can be found in the Supplementary information (Fig. S1 and S2).A mixture of CH4 and CO2 was used as the reactants, and their individual flow rates were controlled by two mass flow controllers (Seven Star D17–09). The flow rate of each gas was 25 ml/min, and they were thoroughly mixed to produce a homogeneous gas mixture before entering into the reactor. After the reaction, the gas stream firstly flowed through a U-shaped tube in a cold trap containing a water-ice mixture to condense the liquid products. These liquid products were dissolved in dichloromethane (CH2Cl2) and were qualitatively and quantitatively analyzed by a GC-MS (GC 7820A and MSD 5975C) and a GC (GC 7820) from Agilent Technologies. The gas products from the U-shaped tube were analyzed online with a PANNA GC (A60). A digital soap film flow meter was used to measure the total flow rate after the reaction. The flow rate of each gas component after the plasma reaction was calculated based on this total gas flow rate and the concentration of each gas component was determined by GC analysis. Then, based on the following equations, the conversion (C) of reactants, the selectivity (S) of main gaseous and liquid products, the carbon balance (CB) as well as the energy efficiency (EE) for gas conversion and product formation can be determined. In these equations, q is the flow rate of each gas component. The subscripts in and out account for the inlet and outlet of the reactor, respectively. P refers to the discharge power of the plasma-catalytic reforming process under different conditions. CxHyOz stands for oxygenates and it represents gaseous hydrocarbons when z equals 0. (1) C CO 2 ( % ) = q CO 2 , in ( mol / s ) − q CO 2 , out ( mol / s ) q CO 2 , in ( mol / s ) × 100 (2) C CH 4 ( % ) = q CH 4 , in ( mol / s ) − q CH 4 , out ( mol / s ) q CH 4 , in ( mol / s ) × 100 (3) S H 2 ( % ) = q H 2 , out ( mol / s ) 2 × ( q CH 4 , in − q CH 4 , out ) ( mol / s ) × 100 (4) S CO % = q CO , out mol / s q CH 4 , in − q CH 4 , out mol / s + q CO 2 , in − q CO 2 , out mol / s × 100 (5) S C x H y O z % = x × q C x H y O z , out mol / s q CH 4 , in − q CH 4 , out mol / s + q CO 2 , in − q CO 2 , out mol / s × 100 (6) CB ( % ) = S CO + ∑ S C x H y O z (7) EE reactant ( mmol / kJ ) = ( q CH 4 , in − q CH 4 , out ) ( mol / s ) + ( q CO 2 , in − q CO 2 , out ) ( mol / s ) P ( W ) (8) EE H 2 ( mmol / kJ ) = q H 2 , out , ( mol / s ) P ( W ) (9) EE Syngas ( mmol / kJ ) = q H 2 , out ( mol / s ) + q CO , out ( mol / s ) P ( W ) (10) EE C x H y O z ( mmol / kJ ) = ∑ q C x H y O z , ( mol / s ) P ( W ) Fig. 2 illustrates the structural properties of the catalytic materials before and after reaction, including γ-Al2O3 and the supported catalysts. As shown in Fig. 2(a), the fresh γ-Al2O3 has a large specific surface area of 288.7 m2/g, a pore volume of 0.378 m3/g and an average pore diameter of 5.574 nm. The impregnation of the γ-Al2O3 beads with active metals (Ni, Ag, or Pt) reduced these three parameters to varying degrees, which can be ascribed to pore blocking in γ-Al2O3 by the penetration of these active metals [45,50]. Among the supported catalysts, these three parameters of Ag/γ-Al2O3 and Pt/γ-Al2O3 were higher than those of Ni/γ-Al2O3, possibly due to lower active metal loadings. Following the plasma reforming process, the specific surface area, pore volume and average pore diameter of the γ-Al2O3 beads decreased noticeably, which could be attributed to carbon deposition formed on the γ-Al2O3 beads during the plasma reforming reaction. However, after the plasma reaction, the textual parameters of these supported catalysts were barely changed, indicating that their properties can be kept relatively stable and that metal sintering may not occur during the plasma reaction process.The diffraction peaks of γ-Al2O3 phase were observed for all the supported catalysts in the XRD patterns of the catalytic materials before and after reaction (Fig. 2(b)). The diffraction peaks assigned to the active metals were detected on fresh samples after the active metals were loaded. The diffraction peaks at 44.4°, 51.6° and 76.1° in the Ni/γ-Al2O3 catalyst correspond to the (111), (200) and (220) phases of the metallic Ni in the enlarged XRD profile images of the fresh catalysts (see Fig. S4 in the Supplementary information). For the Ag/γ-Al2O3 catalyst, the diffraction peaks at 64.5° and 77.6° are assigned to the (220) and (311) phases of metallic Ag. The observed diffraction peaks in the Pt/γ-Al2O3 catalyst (46.4° and 81.3°) are attributed to the (200) and (311) phases of metallic Pt [51–53]. These peaks broadened with low intensity, suggesting that the active metallic species are highly dispersed on the catalyst support. This phenomenon may benefit from the catalyst preparation method used in this work. Instead of thermal calcination and reduction at high temperatures, the metal precursors of these supported catalysts were decomposed and the corresponding metallic oxides were reduced in a subsequent treatment by the Ar/H2 DBD, which could remove un-desired templates from catalysts and provide strong collision by highly reactive species (e.g., energetic electrons) under mild conditions [54]. This can enhance metal dispersion on the catalyst and achieve higher reaction activity by speeding up nucleation and slowing down crystal growth [55]. The TEM images of the fresh catalysts (see Fig. S5 in the Supplementary information) confirm the formation of crystalline and spherical nanoparticles of the active metals. The average diameter of Ni, Ag and Pt nanoparticles in the fresh catalysts was 4.4 nm, 3.0 nm and 2.6 nm, respectively. The existence of Ni, Ag and Pt was demonstrated by the selected area electron diffraction (SAED) and the energy-dispersive X-ray spectrum (EDX) (Fig. S5 and S6 in the Supplementary information). The SAED patterns confirm the crystalline structure of these metallic nanoparticles while the rings in the patterns can be assigned to the corresponding refection plane of each metallic phase in the fresh catalyst, as shown in their XRD patterns. The (111) and (200) reflection planes of the metallic Ag, as well as the (111) and (220) reflection planes of the metallic Pt, were observed in their SAED patterns but not detected in their XRD profiles. This phenomenon might be due to the weak diffraction peaks of these metallic phases, which overlap with the γ-Al2O3 peak. After the plasma reforming process, the diffraction peaks of γ-Al2O3 especially located at 2θ = 38.2°, 49.2°, 67.1° were intensified, which revealed that the crystallite size was enhanced during the plasma reforming process. No obvious changes were observed in the diffraction peaks for the supported catalysts before and after the reaction, indicating that stable properties were obtained. The increase in crystallite size of γ-Al2O3 after the reaction would decrease its specific surface area. These results are in line with those of the textual characterizations.The basic nature of the catalyst is a crucial influencing factor for plasma-catalytic reforming performance. The CO2-TPD was used to investigate the properties of the basic sites of these catalytic materials, which can be divided into three types based on the desorption temperature of CO2, namely low, medium, and high basic sites within the desorption temperature ranges of 80–140 °C, 160–240 °C and over 300 °C, respectively [56]. The area under their corresponding CO2-TPD curves can be used to calculate the number of these basic sites [45]. Catalysts with more strong basic sites can absorb more CO2 and thus promote gas conversion in the plasma-catalytic DRM [57]. Fig. 3 illustrates the basic nature of γ-Al2O3 and the supported catalysts. The γ-Al2O3 support exhibited two CO2 desorption peaks: the first one is in the range of 50–180 °C and peaked at 112.5 °C, which can be assigned to the low basic sites; the second one featured a broader range (240 °C to 640 °C) with a peak value of 484.1 °C, indicating the existence of medium and high basic sites. Compared with γ-Al2O3, the peaks assigned to the low basic sites were weakened, while those corresponding to the medium and high basic sites showed different trends in the supported catalysts. For the Ni/γ-Al2O3 catalyst, the medium and high basic sites are reflected by three more intensified peaks (312.2 °C, 458.9 °C and 563.1 °C) within the similar temperature range as γ-Al2O3. For the Ag/γ-Al2O3 and Pt/γ-Al2O3 catalysts, the peaks for medium and high basic sites were shifted to higher temperature ranges, with the highest peak temperatures of 600.5 °C and 629.0 °C, respectively. These findings suggest that loading the active metals, especially Ag and Pt enhanced the medium and high basic sites. In addition, the high specific surface areas obtained when Ag and Pt were loaded onto the γ-Al2O3 support may provide more strong basic sites for CO2 adsorption and shift the CO2-TPD peaks to higher temperatures. The basicity of the support and catalysts decreased in the following order, according to the areas under CO2-TPD curves of basic sites with different strengths: Ag/γ-Al2O3 > Ni/γ-Al2O3 > Pt/γ-Al2O3 > γ-Al2O3.Since the performance of the plasma-catalytic DRM reaction is strongly dependent on the mutual interactions between plasma and catalysts, it is important to understand how the packing materials affect the physical properties of DBD plasma and link them to the performance of the plasma-catalytic DRM. Fig. 4 shows the electrical signals of DBD under different packing conditions. Clearly, both the voltage and current signals are quasi-sinusoidal under all packing conditions, with numerous spikes in the current signals every half-cycle. Compared to the electrical signals of the DBD without packing (Fig. S7 in the Supplementary information), the number and magnitude of the current spikes in the DBD with packing were significantly reduced. This finding can be attributed to a change in discharge mode from the typical filamentary discharge of the DBD without packing to a combination of weak filamentary discharge and surface discharge in the DBD with packing [49]. We found that loading the active metals (Ni, Ag and Pt) onto the γ-Al2O3 support considerably enhanced the density and intensity of the current spikes. Kim et al. reported a similar phenomenon when they compared the voltage-current characteristics of packed-bed DBD for benzene degradation using zeolite MS-13X and supported catalysts Ag/MS-13X as packing materials [58].To evaluate the current properties more qualitatively, the average values of the number, lifetime and magnitude of the current spikes in each cycle were estimated, as shown in Fig. 5(a). These characteristic parameters are the indicators of the formation of discharge channels and the generation of energetic electrons for plasma chemical reactions [47], and they were all enhanced when using the supported catalysts. The largest number and longest lifetime of the current spikes were achieved in the presence of Pt/γ-Al2O3, whereas using the Ni/γ-Al2O3 catalyst produced the highest average magnitude of current spikes. The current properties were also reflected by the Lissajous figures (see Fig. 5(b)). Loading the active metals on γ-Al2O3 changed the shape of the Lissajous figure from near oval to a parallelogram, following the order of γ-Al2O3, the supported Ag, Ni and Pt catalysts, indicating changes in the discharge characteristics under different packing conditions. In addition, the transferred charge every half cycle, effective capacitance and discharge power were determined using the Lissajous figures (Fig. 5(c) and (d)). These parameters have a strong correlation with the production of reactive species, the spatial distribution of discharge channels and the energy dissipated in plasma reactions [27,59,60]. The transferred charge per half cycle increased from 338.7 nC (γ-Al2O3) to 448.4 nC (Ni/γ-Al2O3), 415.9 nC (Ag/γ-Al2O3) and 509.1 nC (Pt/γ-Al2O3). The variation trends in effective capacitance were similar to those in transferred charge. Loading active metals also increased discharge power, with the highest discharge power of around 45 W (at the same input power of 70 W) achieved in the presence of Ni/γ-Al2O3 and Pt/γ-Al2O3, which correspond to the changing order of the Lissajous figures. Based on the physical properties listed above, the plasma-catalytic systems packed with Ni/γ-Al2O3 and Pt/γ-Al2O3 maybe more capable of generating discharge channels and reactive species, as well as dissipating power into the plasma reforming reactions, than the Ag/γ-Al2O3 catalyst and the γ-Al2O3 support. Fig. 6 depicts the performances of different catalysts. Using the supported catalysts clearly improved the conversions of CO2 and CH4. The highest conversion (21.4 % for CO2 and 27.6 % for CH4) was achieved when the Ag/γ-Al2O3 catalyst was coupled with DBD, which was 25.1 % and 24.9 % higher than when the γ-Al2O3 support was used. The promotion effect of packing supported catalysts on plasma reforming performance was also reported in the previous work. For example, Wang et al. investigated the plasma-catalytic CO2 reforming of CH4 in a DBD reactor and found that the conversions of CO2 and CH4 were enhanced by 6.6 % and 11.0 %, respectively, when packing supported Pt/γ-Al2O3 catalyst [42]. Andersen et al. evaluated the activity of γ-Al2O3 and different γ-Al2O3 supported catalysts in the plasma CO2 reforming of CH4 [36]. Compared to the reforming using γ-Al2O3, the conversion of CH4 was increased by 20.0 % (from 27.1 % to 32.4 %) in the presence of Pt/γ-Al2O3, while the CO2 conversion was almost unchanged (from 21.7 % to 22.0 %). However, the conversions using Ag/γ-Al2O3 were similar to those using γ-Al2O3 in their work [36]. The results of this work are quite different from this finding, which can be explained as follows. The reaction performance of plasma-catalytic DRM is influenced not only by catalytic materials (e.g., textual properties and basic nature), but also by processing parameters that control plasma properties (e.g., electric field, formation of reactive species, electron energy), which influences the plasma-catalyst interaction and thus the reaction. In the plasma-catalytic reforming process, all of these effects are coupled and interact, resulting in different reaction performances when the reaction conditions are changed [2]. The order of CO2 conversion was consistent with the basicity of these supported catalysts (Fig. 3), indicating that the basic sites on the catalyst surface are the main contributors to activating and converting CO2. The catalyst with higher basicity has a greater CO2 adsorption capacity, promoting CO2 conversion while producing O to enhance CH4 conversion. However, CH4 conversion did not follow the same order as CO2 conversion, implying that the textual properties of catalysts and the physical properties of plasma discharge were more important in CH4 conversion. The highest CH4 conversion was achieved with the combined effect of these factors when using Ag/γ-Al2O3, followed by the other two supported catalysts (Ni/γ-Al2O3 and Pt/γ-Al2O3) and the γ-Al2O3 support. Fig. 7 shows the variations in the selectivities of main products, including gaseous and liquid compounds. Similar to gas conversion, loading the active metals enhanced H2 selectivity. The highest H2 selectivity (around 34.5 %) was achieved when packing the supported noble metal catalysts (Ag/γ-Al2O3 and Pt/γ-Al2O3), which was 32.7 % higher than that obtained in the presence of γ-Al2O3. Apart from hydrogen, the main carbon-containing gaseous products include CO, C2H6 and C3H6/C3H8, with minor amounts of C2H4, C2H2 and C4H8/C4H10. The selectivity of CO followed the order of the basic nature of these catalysts, with Ag/γ-Al2O3 having the highest CO selectivity at 61.1%. This finding is explained by the fact that CO was mainly generated by the conversion of CO2, whereas the CO2 conversion was significantly influenced by the basic nature of the catalysts. A similar phenomenon was reported by Zeng et al. [33]. For gaseous hydrocarbons, C2H6 showed the highest selectivity (above 10 %) regardless of the catalytic materials, followed by C3H6/C3H8, C2H2, C2H4 and C4H8/C4H10. When compared to the other catalytic materials, coupling the Pt/γ-Al2O3 catalyst with DBD produced the highest selectivity for all of these gaseous hydrocarbons. The total selectivity of these carbon-containing gaseous products was around 80 % in the presence of supported catalysts, which was significantly higher than when the γ-Al2O3 support was packed (63.6 %).The support and catalysts significantly influence the distribution of the liquid products. Methanol was the main oxygenate with the highest selectivity (8.0 %) obtained when using the γ-Al2O3 support. The other two oxygenates were acetic acid and ethanol, and their selectivity was enhanced by loading the active metals onto the γ-Al2O3 support. The production of these oxygenates as the main liquid products was also reported in the plasma-catalytic DRM process in previous work. For example, Li et al. reported a total selectivity (∼40 %) of liquid products (mainly methanol and acetic acid) when using Co/SiO2 and Fe/SiO2 in a DBD plasma conversion of CO2 and CH4 [37]. Wang et al. achieved the highest selectivity of acetic acid (40.2 %), a major liquid product in the plasma-catalytic conversion of CO2 and CH4 using Cu/γ-Al2O3 [42]. Andersen and his co-workers reported the highest methanol selectivity of 3.6% in the plasma-catalytic DRM over a similar Cu/γ-Al2O3 catalyst [36]. Li et al. reported a total selectivity of more than 30 % for liquid products in the plasma-catalytic DRM using structured Ni-based catalysts [44]. In this work, we also detected formaldehyde as one major liquid product, but it was only formed in the presence of Pt/γ-Al2O3, which was consistent with the findings of Wang et al. [42]. The total selectivity of the liquid products with different catalytic materials was in the following order: Ni/γ-Al2O3 (14.1 %) > Pt/γ-Al2O3 (13.4 %) > γ-Al2O3 (11.5 %) > Ag/γ-Al2O3 (11.3 %). The basic and acidic nature of the catalyst has been found to be critical in influencing the distribution of liquid products [35,40], and acidic sites have been reported to promote the formation of oxygenates [35]. Coupling the Ag/γ-Al2O3 catalyst resulted in the lowest liquid selectivity, which was understandable given that Ag/γ-Al2O3 had the highest basicity of these catalysts. The total selectivity of these liquid products, however, did not strictly follow the order of catalyst basicity, as the structure, surface composition, and reducibility of catalysts all had a significant impact on the distribution of the produced liquid products [44].The gas conversion and product selectivity were stable in the plasma reforming over supported catalysts for 180 min ( Fig. 8). For Ni/γ-Al2O3, the selectivity of the main oxygenates (i.e., methanol, acetic acid and ethanol) remained nearly unchanged during the plasma DRM reaction. When using the support (γ-Al2O3), the gas conversions and product selectivities were relatively stable for the first 60 min, but then gradually decreased as the reaction progressed. The high stable reforming performance of these supported catalysts could be attributed to the stable properties of the plasma-modified catalysts in this study. To confirm this hypothesis, we compared the plasma-catalytic reforming performance of the Ni/γ-Al2O3 catalyst prepared by the plasma-modified impregnation method and the traditional thermal method over a longer reaction time (10 h). Higher and more stable gas conversions were observed when using the Ni/γ-Al2O3 catalyst prepared by the plasma-modified impregnation method, as shown in Fig. S3. The influence of different preparation methods, as well as the underlying mechanism, should be investigated further. For instance, the conversion of CO2 and CH4 was both reduced by approximately 10 % (from 17.2 % and 22.1–15.6 % and 20.0 %, respectively) when the reaction kept running for 180 min, compared with that in the initial stage (20 min). This decrease in the reaction performance might be due to the carbon deposition on γ-Al2O3 during the reforming process, as evidenced by TGA results. Fig. 9 shows the performance stability in terms of carbon balance and carbon resistance of catalyst. According to the TGA profiles (Fig. 9(a)), γ-Al2O3 had the highest overall weight loss, followed by Ni/γ-Al2O3, Ag/γ-Al2O3 and Pt/γ-Al2O3. Clearly, the weight loss occurred mainly in two temperature ranges: 25–150 °C and 150 °C–700 °C, which were ascribed to the desorption of water and the elimination of deposited carbon [9]. The catalysts were heated at 110 °C for 2 h to remove humidity before the reaction. The presence of water in the catalysts after the reaction indicated that the reverse water gas shift reaction occurred in the reforming process [61]. In the second stage of weight loss due to the removal of carbon deposition, three types of carbonaceous species can form: active, less active, and inactive carbon species [62]. In this work, the weight loss of the used noble metal supported catalysts (Ag/γ-Al2O3 and Pt/γ-Al2O3) occurred mainly between 200 °C and 550 °C, indicating the formation of active carbonaceous species. These active carbonaceous species could be easily oxidized during the reforming process [63], and thus would not have a severe impact on their activity and stability, as evidenced by the time variations of the gas conversion. Part of the weight loss in other two used samples, especially γ-Al2O3, occurred at temperatures greater than 600 °C, which was attributed to the oxidation of inactive graphite carbon. This type of carbonaceous species was stable and could not be oxidized during the plasma-catalytic DRM at low temperatures, resulting in catalyst deactivation over a long period of reaction time, which could be the main reason for the decrease in gas conversion with reaction time.The carbon balance for all catalytic materials was less than 100 % (Fig. 9(b)), which can be attributed to carbon deposition on both catalysts and the inner electrode surface, as well as the generated liquid oxygenates attached to the inner surface of the quartz tube that were not completely collected [45,47]. Clearly, the carbon balance when using the supported catalysts was higher than 90 %, and the highest value (94.9 %) was achieved when using Pt/γ-Al2O3. However, the carbon balance with γ-Al2O3 was only 75.3 %, which was mainly caused by the relatively severe carbon deposition evidenced by the TGA analysis. Fig. 10 shows the energy efficiencies for both gas conversion and product formation (e.g., syngas, H2, CxHy and CxHyOz) under different packing conditions. Packing γ-Al2O3 and the non-noble catalyst (Ni/γ-Al2O3) exhibited similar energy efficiencies for gas conversion and gas product formation. Higher gas conversions and gas product selectivities were obtained using Ni/γ-Al2O3 compared to γ-Al2O3, (Figs. 6 and 7), but only at higher discharge power (Fig. 5). This was the main reason for their comparable energy efficiency. Packing the noble metal catalysts in the plasma reactor resulted in higher energy efficiencies. For example, the highest energy efficiency for gas conversion (0.22 mmol/kJ), hydrogen (0.079 mmol/kJ) and syngas formation (0.20 mmol/kJ) was attained when using Ag/γ-Al2O3, while Pt/γ-Al2O3 yielded the highest energy efficiency for the production of CxHy (0.020 mmol/kJ) and CxHyOz (0.014 mmol/kJ). The higher energy efficiencies achieved when using these two noble metal catalysts are mainly due to the improved gas conversion and product selectivity even at relatively lower discharge power. This phenomenon suggests that low discharge power may be advantageous in achieving high energy efficiency [36]. Table 1 compares the energy efficiencies for plasma-enabled catalytic reforming of CH4 and CO2 using different DBDs. The highest energy efficiency obtained in this work is comparable to previous work using a similar DBD reactor [33,36]. It should be noted that the highest energy efficiency for gas conversion and product formation is not always achieved simultaneously under the same conditions. In addition, the energy efficiency of DBDs is still lower than that of other plasma technologies such as gliding arc discharge [64,65]. More research is needed to develop a more efficient plasma-catalytic system for promoting the reaction performance in terms of gas conversion, target product selectivity, and energy efficiency. Since high reaction performance was achieved with noble catalysts (e.g., Ag/γ-Al2O3 and Pt/γ-Al2O3), using the Ni catalyst promoted by noble metals could be a useful approach in terms of both catalytic performance and catalyst cost, as demonstrated in the thermal-catalytic reforming processes [1,66,67]. Similarly, bimetallic catalysts using Ni and other transition metals (e.g., Co, Fe, Cu and Mn) could also be a promising option [1,3]. To make plasma catalysis a promising and economically competitive alternative for DRM, more efforts should be put into rational catalyst design, reactor innovation, optimizing plasma-catalysis configurations and operating parameters, and thoroughly understanding the plasma-catalysis mechanism, particularly plasma-assisted surface reactions [2,15,38,40,43,44].The plasma-catalytic DRM over different supported metal catalysts (Ni/γ-Al2O3, Ag/γ-Al2O3 and Pt/γ-Al2O3) was carried out in a DBD reactor at low temperatures. The results showed that the active metals were uniformly distributed across the γ-Al2O3 support, and that loading the active metal slightly decreased the textual properties of the catalytic materials, while significantly enhancing their basic nature. The conversion of CO2 followed the order of the basicity of the catalytic materials, suggesting that the catalyst basicity plays a dominant role in CO2 conversion, whereas the conversion of CH4 was determined by the combined effect of catalyst properties and discharge characteristics. The highest CO2 and CH4 conversions (21.4 % and 27.6 %, respectively) were achieved when using Ag/γ-Al2O3. Coupling the noble metal catalysts (Ag/γ-Al2O3 and Pt/γ-Al2O3) with DBD resulted in the highest selectivity of gas products (34.5 % for H2 and ∼81.0 for CO and CxHy), while using Ni/γ-Al2O3 gave the highest selectivity of liquid products (14.1 %). The supported catalysts prepared by the modified impregnation method exhibited high stability, evidenced by the time variations in gas conversion and catalyst characterization. In addition, high energy efficiencies were achieved when using the noble metal catalysts. Specifically, coupling the Ag/γ-Al2O3 catalyst with DBD yielded the highest energy efficiency for gas conversion (0.22 mmol/kJ), which was comparable to published results. Future research should concentrate on the development of more efficient catalysts, the optimization of plasma-catalysis configurations and operating parameters, and a thorough understanding of the mechanism of plasma-catalytic reforming.Danhua Mei: Conceptualization, Methodology, Investigation, Formal analysis, Writing – original draft, Minjie Sun: Investigation, Formal analysis, Data curation, Validation, Shiyun Liu: Methodology, Formal analysis, Validation, Peng Zhang: Methodology, Formal analysis, Visualization, Zhi Fang: Conceptualization, Supervision, Resources, Writing – reviewing and editing, Project administration, Xin Tu: Conceptualization, Supervision, Writing – reviewing and editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work is financially supported by the National Natural Science Foundation of China (No. 51807087 and No. 52177149), the Natural Science Foundation for Colleges in Jiangsu Province (No. 19KJB470005) and the Project of Six Talent Peak High-Level Talent Team of Jiangsu Province (No. TD-JNHB-006). P. Zhang acknowledges the support from the Postgraduate Research & Practice Innovation Program of Jiangsu Province (No. SJCX22_0424). X. Tu acknowledges funding from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska- Curie grant agreement No. 813393.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jcou.2022.102307. Supplementary material .

Dry reforming of CH4 (DRM) using a plasma-enabled catalytic process is an appealing approach for reducing greenhouse gas emissions while producing fuels and chemicals. However, this is a complex process that is influenced by both catalysts and discharge plasmas, and low energy efficiency remains a challenge for this technology. Here, we developed a water-cooled dielectric barrier discharge (DBD) reactor for plasma DRM reactions over supported catalysts (Ni/γ-Al2O3, Ag/γ-Al2O3 and Pt/γ-Al2O3) prepared via plasma-modified impregnation. Results show that metal loading on γ-Al2O3 enhanced the basic nature of the catalysts and promoted the formation of discharge channels and reactive species. The maximum conversion of CO2 (21.4 %) and CH4 (27.6 %) was obtained when using Ag/γ-Al2O3. The basic nature of the catalytic materials dominated CO2 conversion, whereas the properties of the catalyst and discharge plasma determined CH4 conversion. The highest selectivity of hydrogen (∼34.5 %) and carbon-containing gas products (∼81.0 %) were attained when using the noble metal catalysts (Ag/γ-Al2O3 and Pt/γ-Al2O3), while the highest total selectivity of liquid products (14.1 %) was achieved in the presence of Ni/γ-Al2O3. Compared with γ-Al2O3, the supported catalysts demonstrated higher stability, especially for Ag/γ-Al2O3 and Pt/γ-Al2O3, which also provided higher energy efficiency for gas conversion and product formation.

The sustainable production of hydrogen through electrocatalytic water splitting is a potential pathway for obtaining clean energy due to its environmental friendliness and high energy conversion efficiency [1,2]. Effective electrocatalysts are necessary for high efficient hydrogen evolution reaction (HER). Pt-group catalysts are regarded as high-activity HER electrocatalysts [3]. However, their practical applications are limited by high cost and scarce global resources [4]. In the past few years, cheap, earth-rich, efficient and sustainable catalysts, including transition metal phosphides [5,6], sulfides [7,8], selenides [9,10] and oxides [11,12], are widely studied to alter the noble metal-based electrocatalysts. In addition, these catalysts are usually loaded on carbon or foam supports to form the electrode, which would affect the long-term stability and be difficult for widespread application [13]. Hence, it is very crucial to explore self-supported catalysts with excellent efficiency and good stability for the HER.Many methods, such as template assisted [14], hydrothermal [15], and dealloying [16], were used to synthesize self-supported electrocatalysts. Dealloying is a selective corrosion process involving the dissolution of active atoms and the reorganization of inert atoms [17]. The uniform bicontinuous nanoporous structure produced by dealloying can expose a larger specific surface area and provide more accessible interior active sites [18]. It was reported that nanoporous transition metal compound catalysts exhibited good HER performance in alkaline electrolyte [6]. Nevertheless, the friability of self-supported material is serious problem for its industrialization in flexible electrode devices. Recently, the amorphous alloys with good mechanical properties were suggested to be catalysts for HER. Moreover, amorphous alloys exhibit good catalytic activity due to the unique disorder structure and inherent abundant defects that would generate more active sites [18]. Hu et al. prepared PdNiCuP ribbons with stable amorphous structure as a high activity electrocatalyst for water splitting process [19]. Xu et al. synthesized an amorphous nanoporous Ni-Fe-P material with good properties for both HER and OER in alkaline conditions [20]. Zhang et al. reported that FeCoPC catalyst exhibited a good HER activity [5].In this work, a facile and promising strategy was designed to fabricate self-supported nanoporous Ni-Co-P (np-Ni-Co-P) catalysts via an electrochemical dealloying method. The 3D binder-free amorphous np-Ni-Co-P catalyst exhibits markedly high catalytic activity for HER with overpotential of 114 ​mV ​at a current density of 10 ​mA ​cm−2, low Tafel slope around 57.3 ​mV dec−1 and good stability in alkaline medium.The Ni60Co20P20 master alloy was prepared by melting pure Ni (99.99 ​at.%), pure Co (99.99 ​at.%) and Ni2P (99.7 ​at.%) in a high-frequency induction melting furnace under high purity argon atmosphere. A melt spinning technique was used to prepare amorphous alloy ribbons with the dimensions of ~1 ​mm wide and ~20 ​μm thick. The precursor ribbons were cut to 30 ​mm ​× ​1 ​mm. The part of ribbon with length of 10 ​mm was served as the working area, and the other part was sealed with silica gel as the clamping part. Thus the area of the working electrode was 0.2 ​cm−2. The Ni60Co20P20 ribbons were electrochemically etched under 0.2 ​V vs. saturated calomel electrode (SCE) in 1 ​M HCl solution in a standard three-electrode setup with a SCE as the reference electrode and a graphite sheet as the counter electrode by using an electrochemical workstation. The control samples for nanoporous Ni-P (np-Ni-P) and nanoporous Co-P (np-Co-P) catalyst was prepared by using the same process with Ni60Mn20P20 and Co80P20 as the precursor alloys, respectively. Finally, the dealloyed samples were washed in distilled water and ethyl alcohol for three times and then dried for 12 ​h in a vacuum drying oven.The phases of the as-synthesized samples were analyzed by X-ray diffraction (XRD, BrukerD8) using Cu Kα radiation. The surface morphologies and elemental composition of the as-synthesized samples were obtained by scanning electron microscopy (SEM, Hitachi S-4800) and energy-dispersive X-ray spectroscopy (EDX, JSM-7800F). Transmission electron microscope (TEM, JEOL 2100 ​M) was used to carried out transmission electron microscopy (TEM), high-resolution TEM (HRTEM), and selected area electron diffraction (SAED) tests. The surface valence state were collected by X-ray photoelectron spectroscopy (XPS, PHL1600ESCA).The electrochemical property of all samples were measured on a typical three-electrode setup (Gamry Reference 1000) at room temperature, with 1.0 ​M KOH solution as the electrolyte. The as-synthesized catalysts, a graphite sheet, and a saturated calomel electrode (SCE) electrode served as the working electrode, counter electrode, and reference electrode, respectively. The potential conversion was calibrated (from vs. SCE) to the reversible hydrogen electrode (vs. RHE) based on the Nernst equation: E(RHE) ​= ​E(SCE)+0.241 ​+ ​0.059 ​pH. Linear-sweep voltammetry (LSV) for the HER was tested with a scanning rate of 5 ​mV ​s−1. Electrochemical impedance spectroscopy (EIS) data were obtained at 150 ​mV (vs. RHE) for the HER with the frequency ranging from 0.1 to 105 ​Hz. All polarization curves were corrected for iR compensation by applying the following equation: Ecorr ​= ​Emea−iRs, where Ecorr is the iR-corrected potential, Emea is the measured potential and Rs is the resistance of the system. Cyclic voltammetry (CV) was conducted to estimate the electrochemical active surface area (ECSA) within ±50 ​mV vs open-circuit potential (OCP) under different scan rates of 10, 20, 40, 60, 80 and 100 ​mV ​s−1. The turnover frequency (TOF) value is calculated from the equation: TOF ​= ​J ​× ​A/(2 ​× ​F ​× ​n), where J is the measured current density at the overpotential of 100 ​mV, A is the surface area of the working electrode, F is the Faraday constant (96485 c mol−1), n is the number of moles of active materials loaded on the electrodes [21,22]. The durability of the samples were measured by chronoamperometry (CA) and 3000 cycles of CV scans. Moreover, commercial Pt/C catalyst (5 ​wt%, Alfa Aesar) was used as the control electrodes. 4 ​mg of commercial Pt/C was added to a solution involving 50 ​μL Nafion and 200 ​μL ethanol to form a homogeneous ink. Finally, 20 ​μL catalyst ink was coated on to the surface of a glassy carbon electrode (surface area: 0.2826 ​cm−2) for three times and the dried at room temperature. Fig. 1 a shows the X-ray diffraction (XRD) patterns of all the precursor ribbons. The Ni60Co20P20 and Ni60Mn20P20 precursors are amorphous, while the Co80P20 precursorexists two crystalline phases, including orthorhombic Co2P and hexagonal-close-packed (HCP) cobalt [23]. After dealloying, np-NiCoP and np-NiP still maintain amorphous structure, while the HCP cobalt is significantly reduced for the np-Co-P sample (Fig. 1b). Compared with the crystal phase, the surface chemistry of amorphous catalysts can be optimized to enhance the performance on the level of molecular [8,24]. The retained mechanical flexibility after undergoing eletrochemical dealloying demonstrate a superior self-supported capability of np-Ni-Co-P (as shown in Fig. S1), which would favor in elimination of the interface overpotential and rapid charge transfer during the electrolysis process [25]. Fig. 2 shows the SEM images of the as-synthesized np-Ni-Co-P, np-Co-P and np-Ni-P samples. All samples exhibit the analogous bicontinuous nanostructures with both the ligaments and pores on a large scale. The nano-scale ligaments and pores can efficiently provide abundant active sites and ion-diffusion routes, which are conducive to mass transfer and electron mobility. The corresponding energy dispersive X-ray spectroscopy (EDS) analysis indicates that the atomic ratio of the np-Ni-Co-P, np-Co-P and np-Ni-P are close to 1.34: 0.76: 1, 2.04: 1 and 2.60: 1, respectively (Fig. S2 and Table S1). The phosphorus content in all samples are also similar (about 30 ​at.%). The transmission electron microscopy (TEM) image of the np-Ni-Co-P (Fig. 3 a) further reveals a uniform bicontinuous nanoporous architecture consisting of interlinked metallic ligaments. High-resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED) (inset of Fig. 3b) further certify a homogeneous amorphous nature of the np-Ni-Co-P down to nanoscale (Fig. 3b) because diffraction spots and crystalline lattices cannot be seen. Element mapping result in Fig. 3c manifests that Ni, Co and P elements disperse on the np-Ni-Co-P sample surface uniformly, where no phase separation or elemental segregation is observed.X-ray photoelectron spectroscopy (XPS) test was performed to determine the surface chemistry for np-Ni-Co-P, np-Co-P and np-Ni-P in the Ni 2p, Co 2p and P 2p spectra. In the Ni 2p3/2 spectrum (Fig. 4 a), the binding energy at 852.84 ​eV can be assigned to Niδ+ in metal phosphides, while other peaks with higher binding energies of 853.42, 856.00 and 860.5 ​eV belong to Ni2+, Ni3+ and the appropriate satellite peak, respectively [26,27]. The Co 2p3/2 XPS spectrum (Fig. 4b) could be divided into three peaks. The peak at 778.33 ​eV belongs to Coδ+ in metal phosphides, and the other two peaks located at 780.20 and 782.37 ​eV can be attributed to the Co2+ and Co3+, respectively, which should arise from the surface oxidation after long-time exposure to the air [28,29]. Fig. 4c shows the high-resolution P 2p spectra consisting of a doublet of P 2p3/2 component at 129.25 ​eV and P 2p1/2 component at 130.13 ​eV from metal phosphides. Moreover, the peak at 132.67 ​eV is ascribed to oxidized phosphate species [30,31]. Obviously, Ni 2p3/2 and Co 2p3/2 in the np-Ni-Co-P exhibit as lightly positive energy shift, while P 2p shows a slightly negative shift compared to np-Ni-P and np-Co-P, implying the different electronic structures for all samples. The positive shift of Ni 2p3/2 and Co 2p3/2 binding energies of np-Ni-Co-P catalyst reveals the enhanced electron transfer, while the negative shift of P 2p3/2 indicates the improved electron occupation, which would lead to heightening electron donating ability [16]. In the electrocatalysis process, the positive Ni and Co centers serve as hydride-acceptor, while the negative P centers serve as proton-acceptor sites [32]. The diffusion of electrons from metallic centers Ni and Co to P can effectively accelerate absorption/desorption of H atoms on active site for the HER [33].The electrocatalytic HER properties were evaluated by means of a typical three-electrode setup in 1 ​M KOH solution, where the as-synthesized materials were directly employed as binder-free catalytic electrodes. Fig. 5 a shows the linear sweep voltammetry (LSV) curves of all the as-synthesized samples with a scanning rate of 5 ​mV ​s−1. As seen, the commercial Pt/C (20%) requires a lowest overpotential of 26 ​mV to reach the current density of 10 ​mA ​cm−2. The np-Ni-Co-P catalyst exhibits good catalytic activity with overpotential of 114 ​mV to achieve the current density of 10 ​mA ​cm−2, which is 9 and 206 ​mV lower than those of np-Co-P and np-Ni-P. According to the LSV results, bimetallic phosphides (np-Ni-Co-P) display a lower overpotential compared with the single-metal phosphides (np-Co-P and np-Ni-P). This result illustrates that the synergistic interplay of Ni and Co could efficaciously enhance the HER electrocatalytic performance.The HER reaction kinetics is assessed by the Tafel slope. In Fig. 5b, the np-Ni-Co-P exhibits a Tafel slope of 57.3 ​mV dec−1, outperforming the np-Co-P (69.9 ​mV dec−1) and np-Ni-P (123.1 ​mV dec−1), suggesting the markedly accelerated HER kinetics of the np-Ni-Co-P catalyst. The Tafel slope value of np-Ni-Co-P catalyst within the range of 40∼120 ​mV dec−1 implies that the HER processes via the Volmer-Heyrovsky mechanism [34,35], which can be described as follow: (1) Volmer reaction:H2O ​+ ​M ​+ ​e−→ M-Had ​+ ​OH− (2) Heyrovsky reaction:H2O ​+ ​M-Had ​+ ​e− → H2+OH− where M presents the active site of the catalyst for HER, and Had presents a H atom absorbed at the active site of the catalyst. The low overpotential and Tafel slope values demonstrate that the np-Ni-Co-P catalyst has better electrocatalytic activity than other reported metal-based HER catalysts (Table S2 and Table S3).The electrochemical active area (ECSA) of the as-synthesized catalysts were estimated by CV curves in the non-Faraday region of ±0.05 ​V versus open-circuit potential (OCP) (Fig. S3). Noticeably, the number of active sites of np-Co-P is 84 times higher than that of np-Ni-P, suggesting that Co might form the major active sites toward HER. The number of active sites with np-Ni-Co-P catalyst is reduced by 4.7 times compared to the np-Co-P catalyst. Nevertheless, according to the turnover frequency (TOF) result, the TOF of np-Ni-Co-P catalyst is calculated to be 0.13 ​s−1 ​at η ​= ​100 ​mV, which is 1.3 and 37 times higher than that of np-Co-P (0.10 ​s−1) and np-Ni-P (0.0035 ​s−1), respectively (Fig. 5c). The intrinsic catalytic activity of the np-Ni-Co-P catalyst is significantly enhanced with the addition of Co. The enhanced intrinsic activity might be ascribed to the joint action of Ni and Co due to the change of electron structure of P which further optimizes the free energy of H adsorption [36]. The charge-transfer resistance is also an indispensable parameter affecting the performance of electrocatalysts. EIS was carried out to further evaluate the HER reaction kinetics of as-synthesized catalysts. As displayed in Fig. 4d, the charge-transfer resistance (Rct) value of np-Ni-Co-P is 5.9 ​Ω, which are much smaller than of np-Co-P (9.2 ​Ω) and np-Ni-P (47.7 ​Ω), revealing rapid electron transport ability and charge-transfer kinetic during the HER process. This is favorable for HER performance.The durability of the electrocatalysts was assessed by chronoamperometry and cyclic CV tests. Fig. 6 a shows a slight current degradation of the np-Ni-Co-P catalyst under constant overpotential of 120 ​mV for 20 ​h test, indicating good durability. However, the CV curve of np-Ni-Co-P has small decay (~28 ​mV) after 3000 cycles (Fig. 6b). The structure of the np-Ni-Co-P catalyst after the electrochemical durability test for 3000 CV cycles in alkaline medium was also characterized by using XRD, SEM, TEM, HRTEM and XPS. The XRD patterns (Figure. S4) of the np-Ni-Co-P catalyst after 3000 CV cycles is almost same as that of the initial catalyst. The SEM, TEM and HRTEM images (Figure. S5 and Figure. S6) still display similar nanoporous structure and uniform distribution of all elements with those in Figs. 2a and 3. Fig. S7 shows the XPS spectrum of np-Ni-Co-P after 3000 CV cycles. The extinction of Niδ+/Coδ+ species is connected with the decrease of Ni/Co-P bonds, which indicates that metal phosphide is gradually transformed into metal oxide/hydroxide. Meanwhile, the appearance of a new peak at 529.18 ​eV in the O 1s spectrum further illustrates the formation of Ni-Co oxides/hydroxides. This should be due to the large polarization during the CV test. The Ni-Co oxides/hydroxides are intrinsically adverse electrocatalytic species for the HER, resulting in initial decay of current density [28]. Nevertheless, the intrinsic active of np-Ni-Co-P could transform the partial surface electronic state of the Ni-Co oxides/hydroxides, thus retaining the outstanding HER catalytic activity inalkaline electrolyte.An amorphous nanoporous Ni-Co-P electrocatalyst has been successfully fabricated via a facile electrochemical delloying process. The np-Ni-Co-P catalyst exhibits outstanding HER catalytic activity with the overpotential of 114 ​mV ​at a current density of 10 ​mA ​cm−2 and small Tafel slope of 57.3 ​mV dec−1 in alkaline solution. The outstanding HER performance is ascribed to the bicontinuous nanoporous structure, the alloying effect and disorder atomic arrangement. The synergetic effect of the Ni and Co elements improves the intrinsic activity of active sites. The facile synthesis of np-Ni-Co-P electrocatalysts with high performance and great durability may open up a new strategy for the application.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.We gratefully acknowledge support by the National Natural Science Foundation of China (51771131).The following is the supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.pnsc.2020.12.006.

Hydrogen evolution reaction (HER) through electrocatalysis using cost-efficient and long-term stable bimetallic phosphide as electrocatalyst holds a great promise for sustainable clean energy technologies. In this study, self-supported nanoporous Ni-Co-P (np-Ni-Co-P) catalyst with amorphous structure was synthesized by utilizing a facile electrochemical dealloying strategy. The results showed that due to the nanoporous structure, disorder atomic arrangement and alloying effect, the np-Ni-Co-P exhibited outstanding electrocatalytic performance for HER with low overpotential of 114 ​mV ​at a current density of 10 ​mA ​cm−2, small Tafel slope (57.3 ​mV dec−1) and good long-term durability in 1 ​M KOH. The synergetic effect of Ni and Co elements improves the intrinsic activity of the active sites. This research provides a direction for the exploration of bimetallic phosphides toward HER in the water splitting process.

Hydrogen as a clean and renewable energy carrier has attracted more and more interests owing to the perspective vision of future hydrogen energy society [1]. However, conventional hydrogen production technology based on coal, petroleum or natural gas are still needed to cope with resource exhaustion, environmental pollution and CO2 emission etc. [2,3]. And the technology based on water reserves by electrolysis or photolysis is in its infancy with high energy consumption and low efficiency [4,5]. Therefore in recent years, increasing attentions have been paid on hydrogen production, based on biomass and biomass-derived oxygenates (e.g. ethanol, ethylene glycol and glycerol), due to their sustainable and renewable properties [6–9].Ethylene glycol (EG) can be obtained directly and efficiently by thermal cracking and catalytic reforming of biomass [10,11]. It is the simplest polyol, often been used as representative compound to investigate steam reforming (SR) of bio-derived polyols due to the same C/O ratio [12]. Some base and noble metals, such as Ni, Co, Rh and Pt, have been evidenced to be effective for SR of EG [13–15]. It has been established that Ni-based catalysts favor C—C rupture and are ideal for steam reforming of biomass-derived oxygenates [16–19]. However, they are easily to be deactivated due to the coke formation at low temperature and the sintering of Ni particles at high temperature [20–22]. The approaches such as changing character of support, optimizing the catalyst preparation procedures and adding a second metal have been used to improve the stability of Ni catalyst [23–29]. It is reported that silica has a large specific surface area favoring the dispersion of the active metal [30]. Ceria has been widely used in SR reaction owing to its excellent redox property, it can also enhance the dispersion of the active metal, increasing reforming capacity and suppressing coke formation [31]. A second metal can be added into SR catalysts to promote the activity and stability by synergistic effect. Tupy et al. [18] reported that the catalytic activity and the stability of NiPt catalysts were better than Pt catalysts in steam reforming of EG and the selectivity of by-products depended both on support and metal. Moraes et al. [27] investigated steam reforming of ethanol over Ni and PtNi catalysts and proposed that Pt addition increased the hydrogenation rate of carbon species and minimized the formation of nickel carbide, improving catalyst stability.The steam reforming of EG to produce hydrogen on Ni catalysts were mainly investigated at high temperature (usually > 400 °C). However, a low temperature reaction is more favored due to the energy saving. The aim of this paper is to enhance the stability and activity of EG steam reforming at low temperature over Ni catalysts via modification with Pt. The catalytic properties of bimetallic catalysts were compared to those of monometallic ones, and the CeO2 supported ones was compared with the SiO2 supported. After characterization with nitrogen adsorption–desorption, X-ray diffraction (XRD), H2 temperature-programmed reduction (H2-TPR) and X-ray photoelectron spectroscopy (XPS), the intrinsic reasons of improved stability and activity over bimetallic catalysts were revealed.Commercial SiO2 (Qingdao Ocean Chemical Co., Ltd.; SSA = 420.7 m2 g−1; dp = 4.88 nm) was washed with distilled water and dried at 100 °C for 12 h and then calcined at 500 °C for 4 h before use. CeO2 was prepared by precipitation method. Cerium nitrate aqueous solution (0.08 mol L−1) was precipitated by slowly adding KOH solution (6 mol L−1) drop by drop and later calcining at 500 °C for 4 h.The supported monometallic Pt and Ni catalysts and bimetallic Pt and Ni catalysts were prepared via impregnation methods. The preparing processes of catalysts are noted in Table 1. The monometallic catalysts were prepared through impregnating SiO2 or CeO2 support with an aqueous solution of H2PtCl6·6H2O or Ni(NO3)2·6H2O, for 12 h respectively. Then the samples were dried at 100 °C overnight and calcined at 500 °C for 4 h. The Pt/CeO2 sample was further treated at 400 °C for 12 h under hydrogen (50 mL min−1) to remove residual Cl−, and then cooled with a nitrogen flow to room temperature during 6 h. The catalysts are identified as Ni/SiO2, Ni/CeO2 and Pt/CeO2. And the Ni and Pt loading is 10 wt% and 3 wt%, respectively.The preparation of bimetallic catalysts is similar to that reported in the literature [32]. The total metal contents are 10 wt%, and the subscripts of metals in Table 1 represents the content of corresponding metals. Pt + Ni/SiO2 catalyst was prepared by co-impregnation with a mixed aqueous solution of H2PtCl6·6H2O and Ni(NO3)2·6H2O. Ni–Pt/SiO2 catalyst was prepared by consecutive impregnation with Ni first and then Pt, namely the calcined Ni/SiO2 sample was further impregnated with an aqueous solution of H2PtCl6·6H2O. After impregnation, the Pt + Ni/SiO2 and Ni–Pt/SiO2 samples were dried at 100 °C overnight and treated using the same procedure as Pt/CeO2 to remove residual Cl−. For H2-TPR characterization, some of Pt + Ni/SiO2 catalyst was calcined at 500 °C for 4 h, denoting as Pt + Ni/SiO2 calcined. The Pt–Ni series catalysts were prepared by consecutive impregnation also but with the inverse sequence, namely the Cl−1 removed Pt/SiO2 or Pt/CeO2 sample was impregnated with an aqueous solution of Ni(NO3)2·6H2O and then drying (100 °C) and calcining (500 °C).Temperature-programmed reduction (TPR) measurement was conducted on a TP-5000 apparatus (Tianjin Xianquan Adsorption Instrument Ltd Co., China). About 50 mg sample was loaded in a quartz tube reactor each run. The catalyst was heated from room temperature to 800 °C at a heating rate of 10 °C min−1 under a flow of 20 mL min−1 of a mixture gas of 10 vol% H2/Ar. The H2 consumption was monitored by a thermal conductivity detector.The crystal structure and phases of catalysts were determined on an XRD-6100 powder diffractometer (Shimadzu Analytical Instrument Co., Japan) using a Cu Kα source. It was operated at 50 kV and 30 mA with a scanning angle (2θ) of 10°–80° and scanning speed of 6° min−1.Textural properties of catalysts were measured with NOVA 3000e (Quantachrome Instruments, USA) using N2 as the adsorbate at 77 K. The special surface areas were determined according to BET method.X-ray photoelectron spectroscopy (XPS) analysis was recorded on ESCALAB 250Xi spectrometer (Thermo Fisher Co., USA) using an Al Kα monochromator with a light spot of 500 μm. The binding energy is calibrated with a reference of C1s peak at 284.6 eV.High Resolution Transmission Electron Microscopy (HRTEM) analysis was performed on a JEOL JEM-2010 microscope operated at an accelerating voltage of 200 kV.The analysis of carbon content on the spent catalyst was operated on a Vario EL Cube analyzer (Elementar, Germany).The catalytic performance for SR of EG to hydrogen was carried out in a continuous flow fixed-bed micro-reactor at atmospheric pressure and 300 °C with 0.25 g catalyst loading. The catalysts were reduced at 400 °C for 1 h under a flow of 30 mL min−1 of pure H2 before running the reaction. And then the reactor was purged for 30 min with a flow of pure Ar, and then the system was cooled down to the reaction temperature. An aqueous solution of EG (10%) as the feedstock was pumped continuously via an HPLC pump (Beijing Weixing manufacturer, China) at a flow rate of 0.06 mL min−1 and vaporized by a band heater at 210 °C. The outlet gas products were analyzed online using a GC-950 gas chromatograph (Shanghai Haixin group Co., Ltd) with TCD detector and two columns of 5A molecular and TDX-01.In comparison of activity and selectivity over various catalysts, the data were reported by averaging three times analyses at steady state. In stability experiment, the data were directly recorded.The carbon conversion (X EG), H2 selectivity ( S H 2 ) and H2 yield ( S H 2 ) was calculated according to following definitions as described previously in the literature [14]: (1) X EG = ( C CO + C CO 2 + C CH 4 ) / 2 C EG × 100 % (2) S H 2 = C H 2 / ( C CO + C CO 2 + C CH 4 + C H 2 ) × 100 % (3) Y H 2 = X EG × S H 2 × 100 % where C x is the content of various product.The SR reaction pathway of EG is quite complex and many side reactions can take place. Extensive studies have evidenced that EG is firstly decomposed to produce H2 and CO, which is followed by water–gas shift (WGS) reaction to convert CO to CO2, leading to more H2 production. In addition, the SR process of EG can be accompanied by side reactions such as methanation reaction, carbon deposition reaction and so on [33].To determine the fitful preparation method, we firstly investigated the influence of metal impregnation sequence of bimetallic catalysts on the performance of EG steam reforming at 300 °C during 2 h on stream. The activity of Pt–Ni bimetallic catalysts supported on SiO2 followed an order of Pt1–Ni9/SiO2 > Ni9–Pt1/SiO2 > Pt1 + Ni9/SiO2 (Table 2 ). The Pt1–Ni9/SiO2 catalyst, prepared by first impregnation of Pt and then Ni, gave the highest H2 yield (60.9%) and EG conversion (88.5%). And this catalyst showed the lowest CO and CH4 contents in effluent gas. Ni9–Pt1/SiO2 catalyst with inverse impregnation sequence presented 83.8% EG conversion and 58.2% H2 yield. Co-impregnating Pt1 + Ni9/SiO2 catalyst has the lowest catalytic activity due to considerable amount of CO intermediate and low EG conversion. Based on these results, the optimized impregnation sequence of first Pt and then Ni is used for the preparation of other bimetallic catalysts.EG steam reforming was conducted to compare catalysts of monometallic Pt and Ni and bimetallic Pt–Ni supported on SiO2 and CeO2. As shown in Table 3 , all of the catalysts supported on SiO2 had considerable activity, with more than 85% EG conversion during 2 h on steam. Compared to monometallic Ni/SiO2, addition of Pt improved the catalytic activity. And the H2 yield increased with the increase of Pt/Ni ratio, Pt3–Ni7/SiO2 catalyst exhibited the highest EG conversion (90.0%) and H2 yield (64.9%). By analyzing gas product distribution, a small amount of CH4 was detected over the Pt3–Ni7/SiO2 catalyst, but no CO existed, which indicated that the Pt addition has positive effect on the WGS reaction.The activity and selectivity properties of monometallic catalysts (Pt, Ni) and bimetallic catalysts (Pt–Ni) supported on CeO2 were shown in entries 4–9 of Table 3. Compared to Ni/CeO2, Pt–Ni/CeO2 catalyst series also presented higher activity. And the influence of Pt/Ni ratio was consistent with that when SiO2 was used. The Pt3–Ni7/CeO2 catalyst with the highest Pt content showed the highest EG conversion (87.8%) and H2 yield (63.5%) among CeO2 supported catalysts. As seen in entry 8 of Table 3, the 3 wt% Pt/CeO2 catalyst exhibited modest catalytic activity with low CH4 concentrations but with high CO content. In contrast, 10 wt% Ni/CeO2 catalyst (entry 4 in Table 3) generated considerable amount of CH4, but no CO was detected. Interestingly, Pt–Ni bimetallic catalysts showed very high selectivity towards H2 with little CO and CH4 contents, suggesting the synergy effect between Ni and Pt.It was reported that both Ni and Pt single metal catalyst revealed a good ability in the C—C bond rupture of the EG, but Pt has low activity for methanation and WGS, in contrary Ni exhibits high activity for both reactions [17,18]. Here the bimetallic Pt–Ni catalysts promoted only the WGS reaction but inhibited the methanation reaction, resulting in increase of H2 selectivity. Thus the Pt–Ni catalysts combined the advantages of Pt and Ni metals and avoided their disadvantages through the synergy interaction between Pt and Ni. To verify the existence of this synergy effect between Pt and Ni, 3 wt% Pt/CeO2 + 7 wt% Ni/CeO2 hybrid by physical mixing was tested in EG steam reforming (last entry of Table 3). As expected, the catalytic performance was lower, because the metal amount of 3 wt% Pt/CeO2 + 7 wt% Ni/CeO2 is only half of Pt3–Ni7/CeO2. However, an appreciable amount of CO (7.3%) was detected over the mechanic mixing catalyst, far greater than 1.3% of Pt3–Ni7/CeO2. Thus the increased activity and decreased CO selectivity of Pt–Ni bimetallic catalysts should be attributed to the synergy interaction between Pt and Ni [18].The properties of support can influence the catalytic activity apparently. In the previous article [14], the Ni supported on CeO2 was proved favorable for the steam reforming reaction of EG due to the presence of surface oxygen vacancy on CeO2. However, in this study, the SiO2 supported catalysts always exhibited slightly higher activity than corresponding CeO2 supported ones. The reasons can be ascribed to the interaction of CeO2 with active metals which was discussed in detail in the following text.It was observed that the Ni catalysts is sensitive to deactivation in the steam reforming of biomass, especially in low reaction temperature [21]. For investigating the stability, the catalytic activity of monometallic Ni and bimetallic catalysts were tested with elongated reaction time under operating conditions (0.06 mL min−1 of 10% EG in water at 300 °C and 1 atm). Fig. 1 showed the H2 yield and EG conversion over Ni/SiO2, Pt3–Ni7/SiO2, Ni/CeO2 and Pt3–Ni7/CeO2 catalysts. For the monometallic Ni catalyst, the activity of Ni/SiO2 was significantly higher than Ni/CeO2 at initial stage. But the EG conversion and the H2 yield decreased rapidly during 24 h on steam. It is well known that the coke is easy to form and deposit on the Ni surface at low reaction temperature, the rapid loss of activity can be considered as that the Ni active sites are blocked by the formation of coke [21]. As for Ni/CeO2, although the yield of H2 and the conversion of EG are lower at beginning, they decreased more slowly with prolonged reaction time, indicating that the stability of Ni/CeO2 was better than Ni/SiO2. It was reported that the rich oxygen vacancies of CeO2 surface are capable removing the carbon deposition and the interaction between NiO and CeO2 inhibits the aggregation of NiO particles [31].For the bimetallic Pt–Ni catalyst, both of the two investigated catalysts showed very good stability and very high activity, as shown in Fig. 1b. The H2 yield of Pt3–Ni7/SiO2 and Pt3–Ni7/CeO2 catalyst kept unchanged basically and the EG conversion declined slightly during 100 h-on-stream in EG steam reforming reaction. This evidenced that the addition of Pt into Ni catalyst not only improves the reaction activity but also enhance the stability.Selectivity of C1 gas products including CO, CH4 and CO2 in EG steam reforming during 24 h reaction time on Ni and Pt–Ni catalysts was shown in Fig. 2 . For the supported Ni catalysts (Fig. 2a), an increase of the CO content companied by a decrease of the CO2 content was observed on Ni/SiO2 with the time on stream. This was ascribed to the deactivation of Ni catalyst. However, the concentrations of CO and CO2 hardly changed over Ni/CeO2, evidencing a better stability. Ni supported on CeO2 showed higher selectivity of methanation than on SiO2. And a slight decrease of CH4 content over the two monometallic Ni catalysts was observed during the 24 h reaction period, which may related with the deactivation of catalysts.For Pt3–Ni7/SiO2 catalyst (Fig. 2b), the CO content gradually increased companied by the decrease of the CO2 content during the 24 h reaction time, but the rate of change was much slower than that over Ni/SiO2. Meanwhile, CH4 content only showed little drop. As for Pt3–Ni7/CeO2 (Fig. 2b), the selectivity of C1 gas products remained almost stable throughout the reforming process, further indicating the excellent stability. Furthermore, no CO was detected throughout the 24 h reaction time. Similar to the Ni monometallic catalysts, the bimetallic Pt3–Ni7 supported on CeO2 presents higher selectivity for methane than supported on SiO2. But the content of CH4 was lower over Pt3–Ni7/CeO2 catalyst than over Ni/CeO2 catalyst, suggesting that interaction of Pt with Ni can depress the methanation reaction.Compared Fig. 2a and b, CeO2 as support exhibited higher stability for C1 components than SiO2, which is consistent with the results of EG conversion and H2 yield shown in Fig. 1. But CeO2 lead to higher methane selectivity, a detrimental effect on the H2 production. This can be partly suppressed by the addition of Pt. Both Ni and Pt–Ni supported on CeO2 showed almost zero CO and 25% CO2 selectivity, suggesting the beneficial effect of CeO2 to WGS reaction. The CO augment over Ni/SiO2 and Pt3–Ni7/SiO2 with the extended reaction time can be attributed to the WGS reaction inhibition by the catalyst deactivation. With Pt addition, the deactivation was apparently slowed by comparison of Ni/SiO2 with Pt3–Ni7/SiO2.Further evidence was obtained by measuring the carbon formation rate over the four catalysts after stability test in EG steam reforming, as shown in Table 4 . For the monometallic Ni catalyst, the carbon content of Ni/SiO2 was significantly higher than Ni/CeO2. The high content of coke covered the active centers, leading to the low stability of Ni/SiO2. For the bimetallic Pt–Ni catalyst, however, only a small amount of carbon deposition was observed after long reaction time (100 h), especially for Pt3–Ni7/CeO2. This results implied that Pt can effectively inhibit carbon deposition to improve stability. The rate of carbon deposition followed an order of Pt3–Ni7/CeO2 > Pt3–Ni7/SiO2 > Ni/CeO2 > Ni/SiO2, consistent with the stability performance of the catalysts. The main cause of catalyst deactivation can be attributed to the carbon deposition, which blocks or covers the active sites.The reaction equation of EG steam reforming can be expressed simply as: (4) C 2 H 6 O 2 + 2 H 2 O = 5 H 2 + 2 CO 2 . At full conversion of EG, the carbon conversion (X EG) is determined as 100%, and H2 selectivity ( S H 2 ) is calculated as 71.4% according to Eqs (1) and (2), so the ideal or the highest H2 yield for the EG steaming reform is 71.4% using this valuating method. The stability experiments showed that during 100 h reaction the H2 yield over the Pt modified catalysts can keep at the extent of 60–65%, corresponding to 84–91% of the ideal H2 yield. These results are among the best results of EG steam reform observed in literature [12,15,18]. Fig. 3 shows the H2-TPR profiles of Pt and Ni bimetallic catalysts supported on SiO2 prepared with different impregnation sequence. For Pt1 + Ni9/SiO2 and Ni9–Pt1/SiO2, only a small peak at low temperature (164 °C and 143 °C, respectively) was observed. The reason is that the two catalysts had been previously reduced to remove the residue Cl−1 in catalyst preparation, the small peak was the reduction of surface Ni oxide [24]. This was evidenced by the calcination experiment of Pt1 + Ni9/SiO2, a much larger peak appears at about 290 °C for Pt1 + Ni9/SiO2 calcined (Fig. 1c). The Ni/SiO2 shows a single reduction peak at 375 °C attributed to the reduction of NiO phase. The addition of Pt results in a shift of the reduction peak to the lower temperature.The H2-TPR result shows that the impregnation sequence can apparently affect the reducibility of bimetal catalysts. As we know, the Pt oxide is much easier to be reduced than Ni oxide. During reduction of Pt–Ni oxide catalysts, the Pt oxide is reduced to metal before the Ni oxide. Thus the hydrogen is firstly activated on Pt surface, then the active hydrogen is transferred to Ni by the hydrogen spillover. Meanwhile, the Ni can also be reduced by the electron transfer from Pt metal. With an addition sequence of Ni first and then Pt, the Pt was exposed on surface, which is favorable for the hydrogen activation and hydrogen spillover, thus this Ni9–Pt1/SiO2 catalyst was reduced at lowest temperature (Fig. 3b). With co-impregnation, Ni and Pt is strongly interacted, the hydrogen activation become relatively difficult owing to less Pt on the surface, thus the reduction temperatures of Pt1 + Ni9/SiO2 and Pt1 + Ni9/SiO2 (calcined) were higher than that of Ni9–Pt1/SiO2. As for Pt1–Ni9/SiO2 with Pt impregnating first and then Ni, Ni was sufficiently exposed on surface, contrarily, the Pt was covered by Ni, leading to that the hydrogen activation and reduction of Ni oxide occurred at higher temperature. But this catalyst still exhibited lower reduction temperature than Ni/SiO2. The interaction between Ni and Pt facilitated the electron transfer from Pt to Ni responsible for the enhanced reducibility.From H2-TPR results of catalysts with different impregnation sequence, it is suggested that the reduction of Ni9–Pt1/SiO2 is mainly through hydrogen spillover, electron transfer plays a less important role. On the contrary, the reducibility of Pt1–Ni9/SiO2 is mainly influenced by the electron transfer from Pt to Ni. And both hydrogen spillover and electron transfer contribute to the reduction of Pt1 + Ni9/SiO2 series.Although the Ni9–Pt1/SiO2 and Pt1 + Ni9/SiO2 showed better reducibility, they had less H2 yield than Pt1–Ni9/SiO2. The order of SR activity of EG is Pt1–Ni9/SiO2 > Ni9–Pt1/SiO2 > Pt1 + Ni9/SiO2 > Ni/SiO2. Thus the reducibility of metals was not necessary to be directly related with the catalytic activity. It is deduced that the surface metal dispersion and synergy interaction of Ni and Pt contribute to the activity and selectivity of SR reaction of EG. The intimated contact and strong Ni–Pt interaction in Pt1 + Ni9/SiO2 owing to the co-impregnation resulted in high CO and CH4 content with low EG conversion and H2 yield. Ni9–Pt1/SiO2 with the sequence of Ni first and then Pt impregnation had proper interaction showed improved catalytic activity. And Pt1–Ni9/SiO2 with inverse impregnating sequence had Ni surface enrichment and proper interaction, giving the best catalysis performance.The influence of support and the Pt/Ni ratio on the reducibility was also studied by the H2-TPR (Fig. 4 ). As expected, the single reduction peak of Ni/SiO2 shifted to a lower temperature with the Pt/Ni ratio increases owing to the low temperature reducibility of Pt. The profile of Ni/CeO2 showed three typical reduction peaks at 217, 276 and 371 °C, which reflected the strength of the interaction between NiO particles and CeO2 (no interaction, weak interaction and strong interaction, respectively). The Pt–Ni/CeO2 catalysts show only two reduction peaks, the lower peak temperature is attributed to the reduction of PtO x . It's worth noting that reduction peak of Ni species at higher temperature and Pt species are obviously shifted to low temperature with the Pt/Ni ratio increases. The shift can also be assigned to hydrogen spillover and the electron transfer from Pt to Ni [27,34]. The peak at lowest temperature, which is attributed to the NiO particles without interaction with CeO2 disappeared, and the peak area at lower temperature increase gradually in the expense of the peak area at higher temperature with the increase of Pt/Ni ratio. These suggested that the strong Ni interaction with CeO2 was weakened with addition of Pt. The TPR data are consistent with that of Palma et al. [35].By comparison of SiO2 supported catalysts with CeO2 supported ones, it is observed that Ni and Pt–Ni metals are easier to be reduced on CeO2 support. And the interactions of Ni metal with CeO2 mainly possess two states with different strengths, which is attributed to co-existence of Ce3+ and Ce4+ ions. The electron transfer from the oxygen vacancy can facilitate the reduction of active metal oxide. But the increased reducibility did not improve the catalytic activity. The CeO2 supported catalysts showed lower EG conversions and H2 yields than corresponding SiO2 ones during 2 h reaction. This can be attributed to the interaction of active metals with the CeO2 [31]. However, this interaction of CeO2 certainly resulted in improved stability. Fig. 5 showed XRD patterns of the SiO2 supported catalysts. For all the catalysts, diffraction peaks appear at approximately 36.9°, 43.3°, 62.7°, which are corresponding to the (111), (200), (220) planes of NiO, respectively. By increasing Pt/Ni ratio, the diffraction peaks of NiO move to the low angle region and became wider, indicating that the existence of interaction between Pt and Ni atoms [29,36]. The shift of NiO peaks to lower angle illustrates the increase of distance between crystal surfaces, which may be ascribed to that some Ni ions were substituted by Pt ions with larger radii. The peak broadening is due to smaller NiO particles, suggesting the better dispersion with Pt addition. No Pt diffraction peaks were found, which indicated that Pt metal is highly dispersed on the catalysts.The XRD patterns of the catalysts supported on CeO2 were shown in Fig. 6 . Regarding CeO2, the diffraction peaks are located at 28.5°, 33.3°, 47.5°and 56.3°, corresponding to the cubic fluorite structure of CeO2. Compared with the catalysts supported on SiO2, the peak intensity of NiO was lower, indicating a well-dispersed NiO phase on CeO2 [14]. And no Pt diffraction peaks are detected. This illustrates that INTERACTION of CeO2 enhance the dispersion of Ni and Pt species. And the Pt modification can further increase the dispersion of NiO, which was evidenced by the decreased diffraction peaks of NiO in Fig. 6b–d.The textural properties of Ni/SiO2, Pt3–Ni7/SiO2, Ni/CeO2 and Pt3–Ni7/CeO2 catalyst were summarized in Table 5 . Larger specific surface area and pore volume were observed for the catalysts supported on SiO2, and larger pore was obtained with the CeO2 as support. The addition of Pt does not significantly change the textural properties of the catalysts, only the specific surface area is slight decreased, probably due to the twice calcinations process leading to little sinter of oxide support.XPS analysis was used to investigate the valence states of elements and the nature of the surface of reduced catalysts. The XPS spectra of Ce 3d, O 1s, Pt 4f and Ni 2p were illustrated in Fig. 7 and detailed results of XPS data were summarized in Tables 5 and 6 . The spectra region of Ce 3d core usually lies within 880 and 920 eV, complicated spectra are obtained due to the presence of multiple oxidation state and spin orbit coupling [22]. As shown in Fig. 7A, the main features are represented by two sets of peaks (V, V′, V″, V‴ and U, U′, U″, U‴). The peaks labeled as V, V″, V‴ or U, U″, U‴ are assigned to 3d3/2 and 3d5/2 of Ce4+ respectively and the peaks labeled as V′ or U′ are attributed to 3d3/2 and 3d5/2 of Ce3+ respectively. It notes that the three catalysts all contain a small percentage of Ce3+ oxide. As reported in early paper [37], the existence of Ce3+ implies the defect structure of CeO2-x due to oxygen vacancies. The Ce3+/(Ce3+ + Ce4+) ratios were calculated according to the measured peak areas (Table 6). This ratio showed no apparent difference in three catalysts, suggesting that impregnation of active metals may not change the reduction of CeO2. But the binding energy shifts to higher energy with the impregnation of metals, suggesting interaction of CeO2 support with the active metals.XPS spectra of O 1s region is composed of two obvious peaks, as shown in Fig. 7B. The peak with lower binding energy (labeled as O I) is ascribed to the lattice oxygen in CeO2 and the high binding energy peak (labeled as O II) is assigned to the chemisorbed oxygen species or defect-oxide involving the surface oxygen vacancies [22,25]. The percentage of O II in total oxygen was calculated in the same way as for Ce3+ (Table 6). Unlike the percentage of Ce3+, the O II contents show apparent difference among CeO2 supported catalysts. It is due to that O II species not only relate with the oxygen vacancies but also involve in the oxygen species interacting with active metals. Thus the O II percentage in some extent reflects the interaction of CeO2 support with active metals. It can be seen that Pt3–Ni7/CeO2 possess the highest O II content among the three catalysts. This partly explains its high stability and activity [38].The XPS results of Pt 4f were shown in the Fig. 7C. The two peaks at about 71.5 and 74.8 eV are attributed to Pt 4f7/2 and Pt 4f5/2. These two peaks can be decomposed further according to the valence state of Pt [27,33]. For 3%Pt/CeO2, the Pt 4f7/2 peak at 71.5 eV and the Pt 4f5/2 peak at 74.8 eV are ascribed to Pt0. The Pt 4f7/2 peak at 72.8 eV and the Pt 4f5/2 peak at 76.0 eV are assigned to Pt2+. In the case of Pt3–Ni7/SiO2 and Pt3–Ni7/CeO2, an additional peak appeared at low binding energy region at about 68.5 eV, which is attributed to the Ni 3p species. Table 7 gives the distribution of different valence states of Pt species on the catalysts. The content of Pt0 in Pt3–Ni7/CeO2 is higher than that in 3%Pt/CeO2. This is due to the interaction of Pt with Ni. And the highest Pt0 (69.8%) is detected on Pt3–Ni7/SiO2 catalyst owing to the weakened interaction of SiO2 with Pt.The Ni 2p core level profiles of all the Ni containing catalysts are shown in Fig. 7D. It can be seen that the Ni 2p3/2 spectra region is broad including multiple valence state (Ni0, NiO and Ni2O3) between 851 and 859 eV. In addition, the Ni 2p1/2 region partially overlaps with the low binding energy tail of Ce 3d5/2 region [27]. The binding energy of Ni 2p1/2 and Ni 2p3/2 of all Ni containing catalysts are summarized in Table 7. The addition of Pt did not change the binding energy of Ni 2p for the SiO2 supported catalysts. But Ni 2p peak shifts to the higher binding energy region for Pt3–Ni7/CeO2 when compared with Ni/CeO2. This illustrates that the interaction of Pt and Ni suppressed the electron transfer from the oxygen vacancy. Fig. 8 shows the HRTEM images of fresh Ni/SiO2, Ni/CeO2, Pt3–Ni7/SiO2 and Pt3–Ni7/CeO2 catalyst. With the amorphous SiO2 supported ones, the metals particles are not even dispersed and the morphology is not regular also. The size of metal particles are several to a dozen nanometers. The weak support interaction of SiO2 give rise to high activity but also lead some sinter [35,39]. On the other hand, the metal and CeO2 particles are dispersed more evenly with the size of about 10 nm and no obvious aggregates can be observed. Owing to the interaction of active metal with CeO2, the enlargement or sinter of active metals are inhibited. This may have some contribution to the stability of CeO2 based catalysts.From the characterization and catalytic experiments, it was observed that the activity, H2 selectivity and stability of EG SR over Ni-based catalysts were significantly improved due to Pt addition. Using the Pt first and then Ni impregnation sequence, the Ni was enriched on the surface and was interacted with Pt through the electron transfer from Pt to Ni. These facilitate the C—C rupture and WGS reactions, meanwhile inhibit the methanation reaction. However, the Pt–Ni interaction cannot be too strong, which was the situation of co-impregnation Pt + Ni catalyst, and cannot be too weak, such as the mechanical mixing catalyst. Increasing the Pt content improves the reaction properties. Thus Pt3–Ni7 catalysts with the suitable interaction showed the best catalytic performance in our study.The main reason of deactivation of Ni-based catalysts is the coke formation. It is more serious at low temperature. With addition of Pt, the Ni site is stabilized in the reduced state by the electron transfer and the hydrogen spillover from Pt to Ni [24], thus the formation of coke carbon species on the Ni surface can be efficiently inhibited.The interaction of CeO2 with Ni and Pt stabilize the active metals avoiding sinter at reaction temperature. And surface oxygen vacancy on CeO2 like Pt can provide electron transfer, stabilizing the Ni in reduced status. Thus the usage of CeO2 support further improved the stability of Ni and Pt–Ni catalysts. However, the interaction of CeO2 support with active metals lowered the catalytic activity, resulted in the decreases of EG conversion and H2 yield, especially at initial reaction stage. These comparison investigations between Ni and Pt–Ni, as well as SiO2 and CeO2, provide important clues for the H2 production.As for hydrogen production by EG steam reforming at low temperature (300 °C) over Pt–Ni series. The impregnation sequence of active metals influenced the catalytic activity apparently. The Pt–Ni catalysts with Pt first and then Ni impregnating sequence possessed proper Pt–Ni interaction and the Ni enrichment on surface, showing high activity and H2 selectivity. It is due to that the high methanation activity of Ni was suppressed by Pt modification. And the significant role of Ni in WGS reaction was remained. Increasing Pt/Ni ratio further increased activity and H2 selectivity, suggesting the importance of Pt addition. The stability experiment showed that the deactivation rate of Ni/CeO2 is lower than Ni/SiO2 mainly due to the less coke deposition. The existence of surface oxygen vacancy of CeO2 leads to the electron transfer from support to Ni, which can suppress carbon deposition formation. Interaction of CeO2 with Ni metals increase the dispersion avoiding the sinter, but decrease the catalytic activities. Pt stabilizes the Ni0 and suppress the coke formation by the electron transfer and hydrogen spillover. Thus the Pt3–Ni7 bimetallic catalysts exhibited the excellent activity and stability.There is no conflict of interest.The work was supported by Natural Science Foundation of China (Grant 21273193, 21473231 and 20973148). The authors also gratefully thank Mrs. Gao Ling from Lanzhou Institute of Chemical Physics, Chinese Academy of Sciences, for XPS analysis and Prof. Song Liguo from department of chemistry, University of Tennessee-Knoxville, for improving language level of the manuscript.

Hydrogen production by steam reforming of ethylene glycol (EG) at 300 °C was investigated over SiO2 and CeO2 supported Pt–Ni bimetallic catalysts prepared by incipient wetness impregnation methods. It was observed that impregnation sequence of Pt and Ni can affect the performance of catalysts apparently. Catalyst with Pt first and then Ni addition showed higher EG conversion and H2 yield owing to the Ni enrichment on the surface and the proper interaction between Pt and Ni. It was observed that although SiO2 supported catalysts exhibited better activity and H2 selectivity, CeO2 supported ones had better stability. This is attributed to the less coke formation on CeO2. Increasing Pt/Ni ratio enhanced the reaction activity, and Pt3–Ni7 catalysts with 3 wt% Pt and 7 wt% Ni showed the highest activity and stability. Ni surficial enrichment facilitated the C—C bond rupture and water gas shift reactions; and Pt addition inhibited methanation reaction. Electron transfer and hydrogen spillover from Pt to Ni suppressed carbon deposition. These combined effects lead to the excellent performance of Pt3–Ni7 supported catalysts.

The abrupt climate changes caused by global warming and the non-stoppable ever-increasing energy consumption are complex challenges humankind must urgently deal with. The extreme dependence of up-to-date energy production technologies on non-renewable sources had been of negative effect on environmental and energy security issues [1]. Environmentally, in order to maintain the present standard of living, producing energy from non-renewable sources is unsustainable as it is directly or indirectly correlated to large amounts of greenhouse gases [1,2]. Economically, the establishment of a competitive energetic matrix, minimally dependent on foreign oil-based sources, is a matter of national sovereignty and of external political relationships [3,4]. A viable solution for the problem presented above could be the development of the so-called hydrogen economy, in which hydrogen would be inserted in the energetic matrices, diminishing crude-oil participation.Hydrogen is considered an important energetic vector for the future generations and its production techniques have been exhaustively studied in the literature [5–9]. This important energy source can be produced by various routes as exemplified by coal and biomass gasification [6,7,10], steam reforming and partial oxidation of ethanol and the consolidated catalytic steam reforming of methane [11–15]. An interesting process that has called attention recently is the steam reforming of bio-oil produced by different types of biomass transformation. Bio-oil is a complex mixture of at least 200 different compounds, including acids, aldehydes, ketones, alcohols and lignin oligomers emulsified in aqueous medium [16–20]. Biomass derived liquids composition is fully dependent of the biomass origin and the applied technology used on its conversion. Due to the complexity of bio-oil composition and to the fact that the catalytic phenomena are utterly correlated to the substrate/surface interactions, authors have studied the catalytic activation of its major component, acetic acid, which can have a 12–14 wt% content in bio-oil [18,21,22].The steam reforming of acetic acid is presented by Eq. (1): (1) CH 3 COOH + 2 H 2 O → 2 CO 2 + 4 H 2 Steam reforming of acetic acid has been studied on different catalytic systems, such as supported noble metals [5,23,24]. Supported noble metals catalysts showed themselves quite active and relatively stable against coke formation, but these systems have the disadvantage to be quite expensive when compared to nickel and copper-based materials [16,25]. The advantage in using nickel-containing catalysts is due to its low price and relatively high availability if compared with noble metals. It has been reported that nickel-based catalysts are as active as noble metal systems but deactivation phenomena due to coke formation is still a problem to be deeper understood in those systems. To better understand the effect of the synthesis route on the steam reforming of acetic acid, Xiang and co-workers [26] have synthesized diverse batches of Ni/γ-Al2O3 using nitrate, chlorate, acetate and Sulphur-containing nickel precursors calcined at 600 °C. The work demonstrated the inadequacy of S-containing synthesis routes due to nickel silicate formation and the interaction of alumina with chlorine revealed to be the main reason to the support’s sintering. The acetate and nitrate routes produced Ni-based catalysts with similar activity towards syngas production, being the coke formation still an issue. Lee and co-workers [27] have evaluated the effect of Ni/γ-Al2O3 catalyst promotion with Mg, La, K and Cu dopants on acetic acid conversion and coke suppression. Mg doping was shown to suppress coke formation circa 60% by comparison with unpromoted Ni/γ-Al2O3 and the presence of lanthanum and potassium seemed to have contributed to an increase in the total catalyst’s basicity. Still on the evaluation of the effect of synthesis route on the final performance of Ni-based catalysts, Lago and co-workers [28] have verified that destruction of ABO3 perovskite-type precursors in a reducing atmosphere, prior to the reaction, could generate a B0/A2O3 catalyst with B0 metallic particles finely dispersed on A2O3 oxide. During the last decade Noronha and co-workers have been studying the application of perovskite-type oxides on syngas production [29–33]. Specifically, on steam reforming of acetic acid, a positive effect of La-site substitution by Pr and Sm was observed on the suppression of coke formation, without significative effect on acetic acid conversion. The authors agreed that the tradeoff between rare earth’s elevated cost and their activity respect to acetic acid conversion makes their large-scale utilization not feasible.In the present work, nickel-based catalysts were obtained in situ, by hydrogen reducing treatment of the La1−xCaxNiO3 system and their activity towards syngas and stability due to coking formation were evaluated in the steam reforming of acetic acid during 23 h time on stream (TOS).Perovskites La1−xCaxNiO3 (x = 0, 0.15, 0.30, 0.50) were prepared by citrate method [34]. This method consists in a simple dissolution of stoichiometric amounts of metallic nitrates, together with an excess of citric acid, that is introduced to guarantee first the formation of the gel-like viscous syrup after water evaporation and finally, the formation of the glassy material that will be calcined in further steps [35]. Stoichiometric quantities of lanthanum, calcium and nickel nitrates were dissolved in water, then citric acid was added to this solution, being the ratio between citric acid molar amount and total molar amount of metallic ions equal to 1.5. The solution was heated for 1 h at 90 °C until the formation of a gel that immediately was calcined in air in different steps: 100 °C for 1 h; 300 °C for 2 h; 800 °C for 4 h, always with heating rate equal to 10 °C/min. The non-Ca containing perovskite was named LaNiO3 and the calcium-containing samples were named Ca 15%, Ca 30% and Ca 50% (molar fraction based).X-Ray Powder Diffraction (XRD) analyses of the calcined samples were conducted at room temperature with Cu Kα radiation (λ = 1.5418 Å) using a Shimadzu XRD-6000 diffractometer. Data were collected in the 2θ range of 10° to 80°, with a scan rate of 0.25°/min. Temperature programmed reduction (TPR) analysis was performed by reducing an amount of 20 mg of perovskite diluted with 20 mg of quartz powder in a flow of 30 mL/min (5% H2/He) from room temperature until 800 °C at a ramping rate of 10 °C/min and the hydrogen consumption was monitored by mass spectrometry (Balzer, model QMS 200). In situ diffraction experiments were conducted in a furnace installed into a Huber goniometer operating in Bragg–Brentano geometry at the D10B-XPD beamline at the Laboratório Nacional de Luz Síncrotron at Campinas – São Paulo, using a radiation with λ = 1.5500 Å. The analyses were performed during the reduction of the catalyst precursors under a flow of 30 mL/min of 5% H2/He mixture.Steam reforming (SR) of acetic acid was performed in a fixed-bed reactor using a Microactivity Reference equipment (PID Eng & Tech.) previously described in [29,30]. The samples (10 mg of catalyst diluted in 150 mg of SiC) were previously reduced with 30 mL/min pure H2 stream at 800 °C for 1 h. The reactant mixture containing a water/acetic acid ratio of 3 on molar basis was pumped (0.25 mL/min) into a vaporizer and diluted with He (200 mL/min) and then, it was fed to the reactor. Initial proof experiments and thermodynamic calculations presented in topic 3.3 were performed to determine the range of temperature for the catalytic tests in order to minimize carbon formation. Being so, LaNiO3 catalyst were evaluated from 400 to 700 °C and the Ca-containing catalysts at 600 °C. Reactants and products were analyzed by gas chromatography (Agilent 7890A) equipped with a thermal conductivity detector (TCD) and a Porapak Q column.Eqs. (2), 3 and 4 were used to calculate conversion, compositions, molar flow in the entrance (ninlet) and outlet (noutlet) of the reactor. (2) Conversion o f r e a c t a n t i : X i = 100 ∗ n i i n l e t - n i o u t l e t n i i n l e t (3) Composition o f p r o d u c t i : Y i = 100 ∗ n i o u t l e t n t o t a l o u t l e t (4) n t o t a l o u t l e t = ∑ n i o u t l e t , i = A c e t i c a c i d , H 2 O , H 2 , C O , CO 2 , CH 4 a n d a c e t o n e Fig. 1 shows the XRD diffraction lines of the calcined La1−xCaxNiO3 samples.XRD patterns of the calcined LaNiO3 sample revealed the characteristic lines of LaNiO3 rhombohedral phase (PDF 33-0711), as mentioned in literature [36], indicating that the perovskite structure is the main phase obtained after calcination for the Ca-free sample (x = 0). The substitution of Lanthanum by Calcium resulted in the appearance of segregated La2NiO4 (PDF 34-0314), CaO (PDF 03-1123) and NiO (PDF 44-1159). Lima and co-workers [34] have showed that values of x ≥ 0.1 on La1−xCaxNiO3 system unfavored the formation of the perovskite-type structures, giving path to the segregation of part of the cations into NiO, CaO and La2NiO4 spinel. Fig. 2 shows the LaNiO3 TPR-H2 profile and some of the temperatures retained for in situ XRD studies (λ = 1.5418 Å) presented by the Figs. 3–5 .The reduction profile for LaNiO3 (Fig. 2) shows four events. The first event starts at 171 °C followed by a shoulder at 230 °C that is rapidly superimposed to a peak with its maximum value at 330 °C. Until 280 °C as shown in Fig. 3, there is no detectable difference between the corresponding diffractograms, indicating that the initial hydrogen consumption could be correlated to the gradual oxygen loss until the maximum structural limit, with the formation of LaNiO2,7, as proposed by Jia et al. [37]. The transformation presented by the perovskite between 280 °C and 420 °C can be verified in those diffractograms presented in Fig. 4 and corresponds to the reduction of Ni3+ to Ni2+, forming La2Ni2O5 (PDF 36-1230). This latter transformation is totally in agreement with the literature [38,39]. The difference between LaNiO3 and La2Ni2O5 (2θ = 26; 28,8; 29,86; 39,16) XRD profiles are very subtle as already shown by Valderrama et al [38]. The conversion from Ni2+ to Ni0 is favoured at temperatures higher than 420 °C, as shown in Fig. 5, generating Ni0 (PDF 00-001-1258) supported on lanthanum oxide (PDF 01-073-2141). TPR allied to in situ XRD measurements showed that the destruction of La2Ni2O5 to generate Ni0/La2O3 occurs in the 420–600 °C temperature range. The TPR profiles of La1−xCaxNiO3 precursors were significantly changed and shifted to higher temperatures as can be seen in Fig. 6 .These profiles presented a first peak of H2 consumption at (400–440 °C), followed by a broad peak (450–620 °C) and a third peak at high temperature (650–670 °C). Attempting to obtain a better understanding also on the redox properties of those calcium-containing precursors, studies of in situ XRD were performed with Ca 50%. The results can be seen in Fig. 7 . As a matter of fact, the analysis of Ca 50% can be extended to the other calcium-containing materials as the TPR profiles are quite similar.For the La0.5Ca0.5NiO3 precursor, the diffractogram obtained after reduction at 490 °C showed that the intensity of the diffraction lines characteristic of La2NiO4 phase decreased whereas weak lines due to La2O3 and metallic Ni were observed. In the TPR profile of this catalyst, the first hydrogen consumption (maximum at 440 °C) can be attributed to the reduction of segregated NiO, in agreement with the disappearance above 490 °C of the corresponding diffraction line (2Θ = 63°). In the presence of calcium (50%) La2NiO4 appears more stable at high temperature if compared to the Ca-free system. The TPR reduction peak above 645 °C is therefore attributed to the complete reduction of La2NiO4 [40,41]. Fig. 7 exhibits only the lines of La2O3 and metallic Ni after further increase of the reduction temperature up to 740 °C. Although the average size of La2NiO4 crystallites is apparently not affected by the Ca content, the average crystallite size of reduced nickel crystallites obtained after reduction of the catalysts is affected by the presence of Ca. the size of Ni° crystallites increases strongly when adding 15% of Ca in the LaNiO3 perovskite, and then in a more moderate way when going from 15 to 50% of Ca. This increase in Ni° crystallite size when increasing Ca content is probably linked to the difficulty to fully reduce the catalysts as evidenced by TPR curves in Fig. 2. The maximum temperature of the final reduction process is 510, 655, 670, 680 °C for the catalysts containing respectively 0, 15; 30 and 50% Ca. The higher the final reduction temperature, the more difficult reduction, but the higher the Ni° diffusion rate, generating therefore larger Ni° crystallite sizes. The sintering of Ni crystallites is apparently not directly correlated to the crystallite size of the parent oxide structure, suggesting that the reduction destroy at least in part the spatial organization of the precursors. Then, together with an agglomeration of Ni° species during reduction, LaOx species also reorganize to generate La2O3 based compounds associated in an unknown way to Ca species. As no change in metallic nickel phase diagram is observed after reduction, the amount of Ca inside Ni° structure must be negligible. Table 1 shows the average crystallite size of La2NiO4 found on calcium-containing calcined samples and the metallic Ni formed after reduction in situ at 800 °C, calculated by Debye Scherrer Equation, using the respective most intense diffraction lines (La2NiO4 at 2θ = 32 and Ni0 at 2 θ = 44,2).The average size of La2NiO4 crystallites did not change significantly with calcium content, but the size of metallic nickel crystallites formed after “in situ” reduction, is clearly increased as the amount of Ca is increased.In order to check the feasibility of hydrogen production from acetic acid steam reforming on the conditions previously detailed in the experimental section, thermodynamic data were calculated by Gibb’s free energy minimization using a home-made code with Mathematica® software. It was considered the presence of carbon graphite in the equilibrium and the ideality for all the gaseous species. In the experimental conditions, the steam reforming of acetic acid leads only to hydrogen, acetone, methane, carbon dioxide and monoxide gaseous products and being so, the thermodynamic allowed conversion of acetic acid and water and the products composition (dry basis) are confronted with the homogeneous experimental results. The confrontation is shown in Fig. 8 (A)–(C).The complete conversion of acetic acid in this range of temperature is not limited by thermodynamics but its steam reforming in gas phase is kinetically unfavoured. The maximum conversion is reached only at 700 °C and it is below 10%. The conversion of water is even lower as can be seen in Fig. 8(A). In the equilibrium, as can be seen in Fig. 8(B), only traces of acetone and solid carbon were expected in the 400–700 °C range and the formation of hydrogen and carbon monoxide are favoured as the temperature increases. In the entire range of temperature acetic acid is not found in the equilibrium and at 400 °C it is fully decomposed in methane and carbon dioxide. The presence of hydrogen at low temperatures accompanies water and methane consumption, being the presence of the later directly correlated to the velocity in which the hydrogen is released. Once methane lacks in the system hydrogen tends to find its maximum yield and the concentration of water does not change anymore. The quantity of carbon expected by thermodynamics is irrelevant.Homogeneous run is presented in Fig. 8(C). Carbon balances shows that the conversion of acetic acid is small but leads to a great formation of coke. The carbon containing substances in gas phase does not justify the quantity of acetic acid converted and only at 700 °C traces of gaseous products were detected.In order to understand the effect of temperature on catalytic steam reforming of acetic acid, LaNiO3 sample was tested on 23 h TOS in a 400–700 °C range, with increments of 100 °C. The results of conversion and compositions are shown in Fig. 9 .The increment in temperature had a significant effect in acetic acid conversion and on products distribution. At 400 °C, the low acetic acid conversion leads almost completely to its ketonization product (Eq.8) and coke, and it is completely in agreement with the presence of carbon dioxide and acetone in gas phase. (8) Ketonization : 2 C H 3 C O O H → ( C H 3 ) 2 C O + H 2 O + C O 2 In the first hours of reaction the conversion of acetic acid is low but rapidly increases as hydrogen is being released in the system and this fact could be correlated to the oxido-reducing character of the gas composition. The gas composition at the beginning could partially reoxidize the Ni0 species and as hydrogen is been produced, nickel oxidized species tend to return to their original metallic state. At this temperature water conversion was very low and so are the yields of hydrogen and carbon monoxide.When the temperature increases, the acetic acid conversion increases up to 600 °C. It should be pointed out that up to 600 °C the conversion of acetic acid has its steady state delayed as the temperature increases and it is achieved exactly when acetone concentration remains constant, also reaching its steady state condition. In all temperatures, excepted 400 °C, the conversion of acetic acid is at its maximum value at the beginning of the run, but rapidly decreases, due to catalyst deactivation. The deactivation observed, as Takanabe et al. reported [24], could be explained by coke formation that is directly correlated to the presence of acetone in the system. Acetone could suffer aldol-condensation type reactions forming coke deposits [24]. The molar fraction of carbon was calculated by mass balance and is presented in Fig. 10 .As already presented in Fig. 9, acetic acid conversion has no thermodynamic limitations and due to kinetic effects, it could also generate carbon deposits on the catalyst. The Fig. 11 shows the potentiality of acetic acid ketonization occurrence in the experimental conditions stablished in this work. (9) K = p acetone ∗ p H 2 O ∗ p C O 2 p acetic a c i d 2 The constant Kexp (experimental) and Keq (equilibrium) are calculated using (Eq. (9)). The former was calculated using the experimental partial pressures verified during the tests and the later using thermodynamic data. In all experiments the Kexp/Keq ratio is lower than 1, meaning that the formation of acetone is favoured by thermodynamics during the entire run, no matter the temperature.The presence of methane at 700 °C is higher than in any other reaction temperature during the first hours of run and its presence is attributed to acetic acid decomposition [24]. Takanabe et al. proposed a kinetic scheme in which acetic acid decomposes into methane, COx and hydrogen, showing that the contribution in methane formation due to carbon monoxide hydrogenation is irrelevant. A bifunctional mechanism could be retained for the catalytic process, in which water could be activated by the support, generating hydroxyl groups that could be used in steam reforming or WGS reactions (Eq. (10)). (10) WGS : C O + H 2 O → C O 2 + H 2 The acetic acid could be decomposed into hydrogen, methane, carbon monoxide and dioxide and CHx species that could oligomerize, blocking catalytic sites and also could recombine to give path to acetone synthesis that is also raw material for oligomerization products. The activation of water on the support surface plays an important role not only in hydrogen formation step (recombination of hydrogen adsorbed atoms) but also in cleaning the active phase by hydrogenating those CHx groups. The effect of support composition is studied in the next section.The effect of calcium content on La1−xCaxNiO3 perovskites in the steam reforming of acetic acid and the activity of the corresponding catalysts is presented in Fig. 12 .The conversion of acetic acid presented by the calcium-containing catalysts is lower than the one achieved by the reduced LaNiO3. Table 2 shows the values of the reactant’s conversion achieved after 23 h TOS and the Ni0 average crystalize size calculated by Debye Scherrer’s Equation for all samples.As can be seen in Table 2, a rather good correlation is observed between the conversion of acetic acid and the Ni0 average crystallite size. The average diameter of Ni0 crystallites when LaNiO3 is reduced in situ is less than 50% of those obtained by reduction of calcium-containing catalysts and so, the metallic area accessible when calcium is present is approximately half of that accessible in the reduced LaNiO3. The smaller is the active phase crystallite size, the higher is the presence of steps and kinks on its surfaces, increasing so the turnover rate [42] what could explain the higher acetic acid conversion on Ni/La2O3 obtained from LaNiO3 precursor. The presence of calcium has a positive effect on water conversion, as can be seen in Table 2. It increases with calcium content and when calcium substitutes lanthanum in Ca 50% sample, the conversion of water increases by a factor of 4.As it has already been mentioned, it is believed that steam reforming of acetic acid proceed via bifunctional mechanism, being the water activated on the support generating hydroxyl groups that could be recombined to give hydrogen and to slow down the deactivation by steam reforming of those possible CHx intermediates derived from acetic acid decomposition. In fact, the increase in calcium content showed an increase in water conversion and also a significant increase in hydrogen production. It could be inferred that the increase in calcium content produces a more efficient support for water activation. The presence of calcium seemed to mitigate acetic acid conversion to acetone and the increase in calcium content makes acetone depletion come sooner with time on stream. The system containing reduced LaNiO3 displayed detectable acetone during the entire 23 h TOS but in those systems containing calcium, acetone formation was pulled back and its presence could not be detected after 5 h TOS in presence of Ca50%.The catalysts generated in situ were active when applied to the steam reforming of acetic acid. XRD studies showed that the presence of calcium increased significantly the Ni0 crystallite average size. The presence of calcium seemed to anticipate both production and complete depletion of acetone, an undesired side product that is the precursor of carbon solid structures that could accelerate the catalyst deactivation. It was possible to verify a linear correlation between Ca content and the conversion of water, being the presence of the former beneficial to water activation. Hydrogen and carbon monoxide generation was also directly proportional to Ca content.The authors acknowledge the Brazilian Synchrotron Light Laboratory for the acceptance of missions D10B-XPD 9253 and 10799.

Ni/CaO-La2O3 catalysts generated by in situ reduction of La1−xCaxNiO3 perovskite systems (x = 0; 0.15; 0.30 and 0.50) were prepared and evaluated in steam reforming of acetic acid under steady state conditions. The objective of this work was to study the effect of calcium content towards activity and syngas formation in such catalytic systems. The catalytic materials were characterized by in situ X-ray diffraction and temperature programmed reduction. The catalytic activity was evaluated in a packed bed reactor in a temperature range from 400 to 700 °C for LaNiO3 reduced samples and at 600 °C for the La1−xCaxNiO3 reduced precursors. The tests indicated that the presence of calcium oxide directly promotes hydrogen formation, by permitting a greater amount of water to be converted and limits the occurrence of ketonization.

To tackle the problems of energy crisis and environmental pollution, clean and renewable energy sources are required to replace the fossil fuels widely used to generate electricity. As an alternative energy sources, hydrogen is becoming an important part of the future energy system because of its high energy density and environment-friendliness. In recent years, hydrogen production by hydrolysis has garnered attention worldwide as an economic and feasible method. At present, the development of catalysts with high hydrolysis efficiency is an unresolved issue; various rare-earth-rich non-precious metal catalysts containing transition metal compounds such as sulfides, phosphates, carbides, nitrides, oxides and selenides have been developed (Chang et al., 2016). The addition of non-metallic elements (O, S or N) to the transition metal-based electrocatalyst can also adjust the kinetics of the reaction and improve the catalytic activity (Xu et al., 2017; Hao et al., 2017; Anjum et al., 2018).As yet, nickel-supported catalysts were used for hydrogenation, oxygen reduction and olefin oxidation of nitrobenzene and nitrophenol; these catalysts have attracted interest because of their low cost and excellent catalytic performance. To facilitate catalysts recovery, nickel particles are usually dispersed in a solid matrix. Recently, much efforts were made to prepare heterogeneous nickel catalysts using various materials such as silica (Mitchell et al., 2021), alumina, graphene/amorphous carbon (Sung et al., 2018), zirconia (Wojciech Gac, et al., 2020), titanium dioxide (Jiang, et al., 2017.), magnesia (Yusuf et al., 2021) and carbon. Among them, porous carbon is the most commonly used economic carrier, and carbon-based carriers such as cellulose paper with metallic nickel particles chemically deposited on the surface (Sahasrabudhe et al., 2018), nickel based mesoporous carbons (Yang et al., 2014), in situ prepared of Ru nanoclusters and porous carbon (Ding R et al., 2020b) have many advantages over other carriers because of their chemical inertia and stability. Compared with other carbon materials, the advantages of carbonised fibre obtained from biomass include easier availability, easier regeneration and lower cost (Lai et al., 2019).Meanwhile, the gradual shift of technology towards green synthetic strategy has necessitated the use of nontoxic, renewable and environmentally benign chemicals (Zhou et al., 2018; Kuo et al., 2019). Hence, designing high-value products with long life, reusability, cost-effectiveness and high efficiency has become an urgent need. Unfortunately, the products of agricultural, industrial or forestry wastes are complex and difficult to separate. Therefore, one of the great challenges in the preparation of biological carriers is to transform them into specific products with specific properties and complexity. Other complex factors include excessive accumulation of chemicals during the use of the product, natural ageing, the recycling process itself and the flow of materials and products associated with it (Kümmerer et al., 2020). However, these biomass catalysts are mostly powders with small particle size. To solve the problem of small particle size of carbon-based solids, nickel-supported catalysts were prepared from poplar with the original skeleton structure. Carbonised wood has the potential to be used as catalyst carrier owing to its good microstructure. Poplar wood has the advantages of low weight, long fibres, high content and easy processing, and hence, it is widely used in housing and as pulp and plywood. Poplar is widely grown because of its fast-growing nature and huge carbon emission reduction potential. It is regarded an important industrial raw material in many countries and its use as an energy crop for biomass or biofuel is gaining interest.Steam explosion is an optional and mature pretreatment technology in the field of biomass conversion. The effect of this technology on hardwood is second only to that on gramineous plants. Since most gramineous plants are a part of crop waste, and the sampling time is greatly affected by seasons, researchers focused on poplar among the perennial broad-leaved trees. The particle sizes of different types of biological raw materials differ after steam blasting. In the process of the steam explosion treatment of biomass raw materials, a large amount of water vapour permeates into raw biomass materials, resulting in the formation of hydrogen bonds with some hydroxyl groups on the cellulose molecular chain. In addition, high temperature and high pressure exacerbate the fracture of the hydrogen bond in cellulose and cause the release of new hydroxyl groups; the specific surface area of cellulose increases and the adsorption capacity of blasting products is improved. Because lignocellulose is renewable and rich in hydroxyl groups, it is an ideal carbon source for preparing carbon carriers. During the preparation process, embedding the nickel nanoparticles directly into carbon materials is crucial for preparing ‘embedded’ catalysts.Herein, we report a simple and effective method for preparing a non-metallic ion-doped nickel-supported catalyst using economical and recyclable fibre raw materials as carriers. The nickel-supported catalysts were prepared by adsorption and reduction at room temperature; among the catalysts, non-metallic ions and Ni-Fe metal particles are highly dispersed. The nanoparticles dispersed and anchored on a rational support can efficiently inhibit the aggregation and thus enhance the catalytic activity (Fu et al., 2019a). Non-metallic ion-doped nickel-supported catalysts exhibited catalytic activity and durability, and can be used in various catalytic reactions, such as electrochemical reactions, 4-nitrophenol (4-NP) reduction and so on. In general, the reported preparation method of the nickel-supported catalyst is convenient, economic and environment-friendly, which is in line with many green chemistry and sustainable development principles and employs widely available starting materials.Poplar was steam exploded at 213 °C for 5 min. A compositional analysis of steam exploded poplar (SEP) on a dry basis was performed. Analytical grade hydrogen sodium borohydride (NaBH4) and 4-NP were procured from Sigma-Aldrich (Shanghai, China). Nickel nitrate hexahydrate, iron nitrate nonahydrate and ethanol were analytical grade and procured from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). All chemicals were used without further purification.SEP was prepared at 213 °C for 5 min by steam explosion. First, 2 g SEP and Nickel nitrate hexahydrate (5 mmol) were dispersed in 100 mL of deionised water for 15 min under ultrasonic treatment, and stirred for 40 min to completely dissolve. After Ni2+ was completely adsorbed by the SEP, 5 mL of NaBH4 (0.5 mol/L) solution was added, and the boron-containing metal oxide was grown vertically in situ in the SEP at room temperature. The resulting product was collected by centrifugation and washing with deionised water and ethanol, and then dried overnight in vacuo. The obtained carbon material is hereafter referred to as N-B-Ni/SEP. Thereafter, the obtained product was used to pyrolyse the feedstock as follows: the precursor and 200 mg NaH2PO2 will be prepared at both ends of the alumina crucible. Temperature was increased at a rate of 5 °C/min in a nitrogen atmosphere, and the N-B-Ni/SEP was held for 1.5 h at 350 °C, followed by cooling to room temperature inside the furnace. The carbon material obtained thus is hereinafter referred to as N-B-NiP/SEP. N-B-Ni5Fe5P/SEP was prepared by adding nickel nitrate hexahydrate (5 mmol) and iron nitrate nonahydrate (5 mmol), under identical conditions as described in the previous sentences.A Fourier transform infrared (FT-IR) spectrometer (Karlsruebrook, Germany) using KBr pellet technology was employed to measure FT-IR. A Zeiss Merlin instrument was used for scanning electron microscopy (SEM) under 10 kV voltage. Energy dispersive spectroscopy (EDS) was performed to determine the elemental composition. The crystal structure of the sample was analyzed by using an Ultima IV X-ray diffractometer. The working voltage of the X-ray diffractometer was 40 kV, and the current density was 30 mA. X-ray photoelectron spectroscopy (XPS) analysis was carried out by using an ESCALAB 250 analyzer (Thermo Science) and a monochromatic Al Ka X-ray source. Raman spectroscopy (Horiba evolution) measurements were employed to study the physical properties of the samples. The adsorption–desorption isotherms of nitrogen were determined using a BELSORP-mini II instrument and the Brunauer-Emmett-Teller (BET) method. The ultraviolet (UV)- visible (vis) absorption spectra were recorded via a UV-2900 spectrophotometer (Hitachi, Japan). Inductive coupling plasma emission spectroscopy (ICP-OES) was performed using a PerkinElmer 8300 analyzer.The reduction of 4-NP was performed in a quartz cuvette and monitored by performing UV–vis spectroscopy (Hitachi UV-2900) at room temperature. For comparison, an aqueous 4-NP solution (0.01 M) was prepared and measured prior to monitoring the change in absorption. Then, a total of 25 μl of aqueous 4-NP solution was mixed with 2.5 mL of fresh NaBH4 (0.01 M) solution. Subsequently, a fixed amount of nickel catalyst was added to start the reaction, and UV spectrometry was employed to monitor the reduction in situ by measuring the absorbance of the solution at 400 nm over time.The electrochemical measurement was carried out at room temperature using a three-electrode device using CHI760E electrochemical workstation. A glassy carbon electrode was the working electrode (opposite electrode), and the Ag/AgCl electrode was the reference electrode. The linear sweep voltammogram (LSV) was recorded at a scanning rate of 5 mV/s in 1.0 M KOH electrolyte for OER, and in 0.5 M H2SO4 for HER. The scanning range was 1.0–1.8 vs. reversible hydrogen electrode (RHE). The LSV curve was obtained at a scanning rate of 5 mV/s, and the LSV curve was corrected by 90% IR compensation method. Using the Nernst equation (ERHE = EAg/AgCl + 0.059·pH + 0.197), the measured potential was converted into the corresponding RHE potential. The current density (J) was normalized to the geometric surface area, and the measured potential EAppl (vs. Ag/AgCl) was converted into the RHE. The overpotential (η) of OER when the current density was 10 mA/cm2 was calculated by using the equation η = ERHE-1.23 V. The overpotential (η) of the HER when the current density was 10 mA/cm2 was calculated using the equation (η = ERHE). According to the Tafel equation (η = a + b·log (J) to calculate the Tafel slope (b), the Tafel slope was obtained by fitting the linear part of the Tafel curve (Cao et al., 2020; Lan et al., 2019).The FT-IR analyzer was used to identify the functional groups on the catalyst samples and SEP surface, as shown in Fig. 1 . The FT-IR spectra showed strong absorption at 3421 cm−1, which is attributed to the stretching of the phenolic and aliphatic hydroxyl groups. The peaks at around 2921 cm−1 that were related to the C-H functional group changed after the nickel-supported catalyst samples were prepared by the SEP. The results showed that chemical interactions and ion changes occurred between OH, C–H, C=O and heavy metal ions in the nickel bio-adsorption process (Foroutan et al., 2019b). The FT-IR spectra of N-B-NiP/SEP, N-B-Ni5Fe5P/SEP and N-B-Ni5Fe5P/SEP-1 confirmed the existence of NO3 − and OH− group in the nickel–iron loaded catalyst (Fig. 2 ). The bands at 1596, 1363 and 777 cm−1 were the characteristic vibrations for H2O, –NO3 −and Metal-O (M–O) (Lee et al. 2019; Yang et al. 2019), respectively, thereby showing again that Ni-Fe formed on SEP. Compared with the blank SEP, the change in the absorption peak at 777 cm−1 indicated that metal particles were attached to the surface of SEP. In contrast, the weak peaks at 1112 cm−1 are characteristic of the C-N stretching mode (Coates 2006). The absorption peaks of the repeatedly used catalysts at 777 cm−1 did not diminish, indicating that the catalytic process did not affect the transition metal particles on the carrier surface.The morphology and microstructural information of the N-B-Ni5Fe5P/SEP and N, B-NiP/SEP were systematically studied using electron microscopy techniques (Xiao et al., 2016). The closely packed Ni-Fe coating deposited at room temperature did not change the fibre structure of the SEP (Fig. 3 a-b). The nickel-plated iron or nickel SEM contain a large number of voids in the bracket. SEM images (Fig. 3a-b) show that the growth of the Ni-Fe layer with vertically arranged nano-thin sheets, with interconnected macroporous morphology, will not hinder the underlying macroporous structure. This interesting morphology is beneficial for electrocatalysis because it provides a large number of exposed catalytic active sites and enables electrons to travel rapidly along vertical nanoflakes. Energy dispersive X-ray (EDX) spectroscopy was performed to further characterize the elemental composition and distribution of the N-B-Ni5Fe5P/SEP sample by performing EDS surface scans (Fig. 3d-g). The results show that Ni, Fe, P, B and N are uniformly distributed in the sample, and that the atomic ratio is 1.28 (Ni): 1.21 (Fe). Further, B and N atoms were confirmed to have successfully entered the SEP. The above results further prove that the N-B-Ni5Fe5P/SEP was successfully realized by introducing zero-valent N and B atoms.The detailed structural features of the obtained sample were first investigated by an X-ray diffraction (XRD) study. All the diffraction peaks were ascribed to the hexagonal NiP (JCPDS card No. 03-065-1989) without any peaks for impurities, suggesting that the N-B-NiP/SEP precursor was successfully converted into nickel phosphide/SEP. (Pinilla et al., 2016; Sun et al., 2020). The diffraction pattern for PC has a broad peak at 26°, which is a characteristic of the (002) plane of graphitic carbon (Fig. 4 ).Compared with N-B-NiP/SEP, the four diffraction peaks of NiP in the XRD spectrum of the Fe-doped catalyst (N-B-Ni5Fe5P/SEP) shifted to a larger diffraction angle with Fe doping, indicating that Fe atoms enter the Ni lattice to form an Fe-Ni alloy. The intensity of the diffraction peaks of 111, 201, 210 and 300 of Ni decreases with Fe doping, indicating that Fe doping affects the crystallinity of the alloy particles.The average crystallite size was determined to be about 10.78 nm for N-B-Ni5Fe5P/SEP, and 17.97 nm for N-B-NiP/SEP from the (111) reflection by utilizing Scherrer’s equation that relates the coherently scattering domains with Bragg peak widths: D = kλ/B cos(θ), where k = 0.89 for spherical particles, and B is the full angular width at half-maximum of the peak in radians. Combined with the aforementioned energy spectrum (Fig. 3d-g), it can be seen that the Ni-Fe elements are uniformly distributed on the carrier surface. From these results, we come to the conclusion that the metal particles are well dispersed on the fibre surface, and Fe doping affects the crystallinity of the alloy particles. The above results show that the addition of Ni can effectively promote the miniaturization of Fe grains (Mansouriieh et al., 2016). The XRD pattern of Ni5Fe5/Ni-P electrode (Fig. 4) further confirms the amorphous nature of Ni-Fe catalyst layer as no new peaks are observed besides those corresponding to the catalyst. In fact, it has been proposed that amorphous Ni-Fe electrocatalysts are much more active than their crystalline counterparts because the amorphous electrocatalysts have good structural flexibility and high density of co-ordinatively unsaturated sites that help in the adsorption of oxidized intermediates.The XPS survey scan spectrum (Fig. 5 a) clearly confirmed that Ni, Fe, B, P, N, O, and C elements were present in the samples. According to the XPS analysis, the Ni and Fe contents in N-B-Ni5Fe5P/SEP were 11.14 and 14.84 wt% (Table 1 ), respectively. The molar ratio and actual total loading content of Fe and Ni in the N-B-Ni5Fe5P/SEP and N-B-NiP/SEP catalysts were further determined by ICP-MS, as listed in Table 1. The results are in good agreement with the theoretical molar ratio, indicating that the Ni and Fe metal particles are uniformly dispersed on the SEP carrier. Note that the Ni and Fe loading of N-B-Ni5Fe5P/SEP from ICP-OES (17.19 and 15.58 wt%, respectively) analysis were much higher than the outmost surface Ni and Fe content (11.14 and 14.84 wt%, respectively) as measured by XPS. Hence, we conclude that the tiny Ni and Fe particles are embedded in the carbon fibre instead of being anchored on the surface (Ding et al., 2020). This phenomenon is more obvious in the nickel content of the N-B-NiP/SEP.The high-resolution spectra of the Ni 2p region showed two peaks, 2p3/2 (856.82 eV) and 2p1/2 (874.47 eV) corresponding to the Ni2+ derived from the oxidation of the NiP surface (with the corresponding shakeup satellite peaks at 862.26 and 880.03 eV, respectively) (Ding et al., 2020). The Fe 2p spectrum (Fig. 5h) was fitted into two separate peaks at 711.76 and 724.68 eV corresponding to the spin–orbit states of Fe 2p3/2and Fe 2p1/2, respectively. This finding also confirms that the Fe predominantly exists in the Fe3+state. As shown in Fig. 5g, compared with the N-B-NiP/SEP, the negative shift of NiP indicates a decrease in the number of electrons at Fe site and the accumulation of electrons around the Ni site (Jiao et al., 2019). These changes in electron accumulation cause changes in the distribution of electrons, thus changing the local electronic structure of the metal position.The XPS spectrum (Fig. 5c) for O 1s of samples can be deconvoluted into two peaks at binding energies of 531.08 and 532.08 eV, which were attributed to the surface-adsorbed water (–OH) and C-O species (oxygen vacancies), respectively. The oxygen vacancies indicate a defect site with low oxygen coordination, which decrease the barrier for the adsorption of OH− and promotes OER. In particular, N-B-Ni5Fe5P/SEP and N-B-NiP/SEP show a clear difference in the area of oxygen vacancies because of the presence of Fe metal ions (29.49 %: 51.62 %) (Xu et al., 2018; Kim et al., 2020).As shown in Fig. 5b, the four components of C1s spectrum (284.77, 286.36, 288.49 and 291.54 eV) were attributed to sp2 C-C, sp3 C-C, C-O and carboxylic groups, respectively. In the high-resolution XPS spectra, P 2p exhibits three contributions, P 2p3/2 and P 2p1/2, located at 129.49 and 130.46 eV (Fig. 5e), respectively, which are assigned to NiP, and the peak at 133.72 eV that is attributed to the oxidized P species.The B 1s spectrum (Fig. 5f) clearly evidences the presence of three chemical environments for phosphorus atoms (B-O, B-C, and B-Ni). The existence of B3+ in N-B-Ni5Fe5P/SEP and N-B-NiP/SEP catalyst is evidenced by the peak at 191.60 eV (Fig. 5f), which can be attributed to the surface oxidation of the borate species. Compared with N-B-NiP/SEP, the peak intensity at 190.59 eV of N-B-Ni5Fe5P/SEP that corresponds to B-C bonds was higher, indicating that some C atoms in the carbon fibre were replaced by B atoms. Pleasantly, the peak at 187.33 eV can be attributed to B(0) in the Ni-B bonds, which matches well with the literature. This result suggests that there are abundant zero-valent B atoms in the N-B-Ni5Fe5P/SEP and N-B-NiP/SEP after N2 treatment.In the high-resolution N 1s spectrum, in addition to the characteristics related to pyrrolic-N (402.07 eV) and pyridinic-N (400.00 eV), a characteristic peak with a 397.10 eV binding energy is observed in the N regions. It is attributed to metal-nitrogen bonds, indicating the presence of zero-valent N(0) atoms in the N-B-Ni5Fe5P/SEP and N, B-NiP/ SEP (Fig. 5d). The presence of N dopant in the sample will inherently improve the interaction ability with the reactants and produce a higher positive charge density on its adjacent carbon atoms, which may also contribute to the high activity of the sample (Sun et al., 2020). Therefore, the above results indicate that the zero-valent N and B atoms were successfully doped into N, B-NiP/SEP and N-B-Ni5Fe5P/SEP.The crystallization and graphitization degree of the carbonized SEP support on the Ni-supported catalyst and Ni-Fe bimetallic catalyst were studied by Raman spectroscopy. In general, an ID/IG ratio less than one is ideal. As shown in Fig. 6 , the carbon fibre carriers have high quality and crystallinity, and the peak intensity ratio (ID/IG) is less than one, and the spectra of carbon samples show two distinct bands. The first band is the well-known D band, located at 1363 cm−1, attributed to the disorder in the carbon structure, such as defects in the carbon structure or amorphous carbon (Msda et al., 2002; Awadallah et al., 2013). The vibration of sp2 carbon atoms in the graphitization region forms the G band located at 1589 cm−1 (Ali et al., 2017; Allaedini et al., 2015). Generally, the ratio of the D-band strength to G-band strength ID/IG is used to reflect the degree of graphitization. The ID/IG ratio of N-B-Ni5Fe5P/SEP is 0.89, which indicates that a large number of defects and irregular structures were introduced into the carbon fiber carrier. The ID/IG ratio further increased to 0.95 for N-B-NiP/SEP, indicating the enhanced number of structural defects, increased localized sp3 defects in sp2 framework and high electrical conductivity.The specific surface area and porosity of the obtained materials were investigated by N2 adsorption–desorption experiments. In the curves of N-B-Ni5Fe5P/SEP and N-B-NiP/SEP (Fig. 7 a), the type IV adsorption branches corresponded to the mesoporous structure. According to IUPAC classification, the isotherms (Fig. 7) of the mixed oxides were classified as type IV with an H3 hysteresis loop, suggesting the existence of mesoporous materials with an incision-like pore geometry. The specific surface area of N-B-Ni5Fe5P/SEP and N-B-NiP/SEP were calculated to be 55.44 and 57.18 m2/g, respectively. The pore size distributions are shown in Fig. 7b. The average pore size of N-B-Ni5Fe5P/SEP was about 11.86 nm, while those of N-B-NiP/SEP was around 8.42 nm. It was clear that N-B-Ni5Fe5P/SEP and N-B-NiP/SEP were mainly composed of micropores and mesopores of size around 10 nm. As shown in Fig. 7d, the average pore widths of two samples follow the order of N-B-Ni5Fe5P/SEP > N-B-NiP/SEP and the pore volumes of N-B-Ni5Fe5P/SEP and N-B-NiP/SEP were 0.19 and 0.11 cm3/g, respectively. The pore structure of materials play an important and even decisive role in determining many material properties. When using carbon materials as carriers, their porous properties are conducive to the diffusion of substrates and products and can lead to the exposure of more active sites, thus improving the overall activity of the catalyst.The electrocatalytic OER performance of N-B-Ni5Fe5P/SEP and N-B-NiP/SEP were studied in O2-saturated 1 M KOH. The LSV data (Fig. 8 a-b) were recorded with the scan rate of 5 mV/s. For N-B-NiP/SEP, the Ni2+/Ni3+ was oxidized in the potential range of 1.35–1.5 V (all potentials were versus the RHE) (Fig. 8b). The presence of the oxidation peak indicated that because of insufficient oxidation, a fully protected NiO shell may not be formed outside the Ni nanoparticles, leading to corrosion of metal Ni and the formation of NiOOH during OER in the alkaline solution (Sivanantham et al., 2016). The curves of polarization (Fig. 8a-b) showed that the N-B-Ni5Fe5P/SEP exhibited excellent OER performance with an overpotential of 395 mV at 10 mA/cm2 and 488 mV at 30 mA/cm2 current density, compared to N-B-NiP/SEP (431 and 579 mV, respectively).In addition, to investigate the kinetics of these catalysts, the Tafel slopes obtained from the LSV polarization curves are shown in Fig. 8c. The Tafel slope of N-B-Ni5Fe5P/SEP (101 mV/dec) was considerably smaller than that of N-B-NiP/SEP (151 mV/dec), confirming the faster OER kinetics of the former. Our research results indicate that the synergetic effect of the Ni-Fe bimetal loading and carbon carrier played an important role in facilitating the OER kinetics (Li et al., 2020; Jiang et al., 2018; Yue et al., 2019).To assess the electrocatalytic HER activity of the N-B-Ni5Fe5P/SEP and N-B-NiP/SEP, the related electrochemical measurements were performed using a three-electrode system. Fig. 8d shows the polarization curves of the N-B-Ni5Fe5P/SEP and N-B-NiP/SEP in N2-saturated 0.5 M H2SO4 solution. While the N-B-NiP/SEP, which has a η10 value of 397 mV, the N-B-Ni5Fe5P/SEP requires 392 mV to reach 10 mA/cm2, implying that the Fe trace in N-B-Ni5Fe5P/SEP does not contribute to the electrochemical activities and remains a mere spectator species. Tafel slopes were drawn to evaluate HER kinetics (Fig. 8e). The Tafel slope is 122 mV/dec for N-B-Ni5Fe5P/SEP, which is much smaller than that of the N-B-NiP/SEP (119 mV/dec). In the study of the mechanism of electrocatalytic hydrogen evolution in acidic media, it is generally believed that the reaction process is divided into the following three steps: the first step is the electrochemical reaction process; the second step is the electrochemical desorption process; the third step is the compound desorption process. The general HER mechanism includes at least an electrochemical process and a desorption process, and hence, it can be divided into the Volmer-Heyrovsky mechanism or Volmer-Tafel mechanism according to the different rate steps. As seen from Fig. 8c and e, the Tafel slopes of the N-B-Ni5Fe5P/SEP and N-B-NiP/SEP are 122 and 119 mV/dec, respectively. So the hydrogen evolution process of the catalyst in acidic medium is a slow discharge mechanism, and the Volmer reaction process has a rate-control step, which is the Volmer-Heyrovsky mechanism (Conway and Tilak, 2002; Li et al., 2014; Li et al., 2011).The removal of 4-NP from wastewater is of significant importance from the perspective of environment protection as 4-NP is a prevalent contaminant produced in industry and agriculture (Choi and Oh, 2019; Ding et al., 2020). It is known that 4-aminophenol (4-AP) is very useful and important in many applications, and it is used in analgesic and antipyretic drugs, photographic developer, corrosion inhibitors and anticorrosion lubricants. The reduction of 4-NP to 4-AP has been extensively used as a benchmark system to evaluate the catalytic activity of metal NPs (Chang et al., 2012; Yang et al., 2014).Therefore, the reduction of 4-NP toward 4-AP in the presence of NaBH4 was selected as a model reaction to further confirm the generality of the N-B-Ni5Fe5P/SEP and N-B-NiP/SEP. As shown in Fig. 9 a, the adsorption peak of 4-NP was red-shifted from 317 to 400 nm immediately upon the addition of NaBH4 solution which corresponds to a colour change from light yellow to yellow green because of the formation of the 4-nitrophenolate ion under alkaline conditions. When the catalyst was added, the intensity of the characteristic peak at 400 nm rapidly declined. The reduction of 4-NP was completed within 10 min over 10 mg N-B-NiP/SEP and N-B-Ni5Fe5P/SEP (Fig. 9c). Considering that the reductant concentration is much higher than that of 4-NP (CNaBH4/C4-NP = 100) in the reaction mixture, the pseudo-first-order rate kinetics with respect to 4-NP concentration could be used to evaluate the catalytic rate. The reaction kinetics can be described as − ln(Ct/C0) = kt, where k is the rate constant at a given temperature and t is the reaction time. C0 and Ct are the 4-NP concentration at the beginning and at time t, respectively. As expected, a good linear correlation of ln(Ct/C0) vs. reaction time t was obtained (Fig. 9b), and the kinetic rate constant k was estimated as 0.19 (R2 = 0.99) and 0.344 (R2 = 0.99) min−1 for N-B-NiP/SEP and N-B-Ni5Fe5P/SEP, respectively. To compare different catalysts, we calculated the ratio of rate constant K over the total weight of the nickel catalyst, where K = k/m. Thus the activity factor K was calculated as 19 and 34.4 min−1·g−1 for N-B-NiP/SEP and N-B-Ni5Fe5P/SEP, respectively. It is clear that N-B-Ni5Fe5P/SEP clearly had the largest activity factor, compared with other precious metal catalysts such as Ru/C (0.034 min−1) and Ru/PC-IM (0.198 min−1) (Ding et al., 2020). With an increase in the number of cycles, the conversion of 4-NP decreased slightly, possibly because of the partial loss of active surface area caused by the partial loss of catalyst during recovery.The excellent catalytic performances of N-B-Ni5Fe5P/SEP for 4-NP reduction lead to the following advantages. From the point of view of catalysis, SEP is an ideal substrate for the growth of an active catalyst layer. Because there are abundant coordination hydroxyl groups and epoxy functional groups on the cellulose microfiber, the ultra-fine and clean metal nanoparticles formed in situ are uniformly dispersed on the surface of the carrier rather than being embedded in the carrier. Together, these two functions can lead to stronger binding and faster mass transfer kinetics.In summary, using economical and recyclable fiber raw materials as carriers, nickel-supported catalysts were prepared by adsorption and reduction at room temperature. For the model catalytic hydrogenation of 4-NP by NaBH4, the N-B-NiP/SEP and N-B-Ni5Fe5P/SEP catalysts exhibited much better catalytic performances than the other catalysts recently reported in terms of the catalytic activity (with the proposed catalysts, the reaction was completed within 10 min) and reaction rate constant (0.19 and 0.344 min−1 for N-B-NiP/SEP and N-B-Ni5Fe5P/SEP catalysts, respectively). The catalyst showed activities for electrocatalytic HER and OER under ambient conditions. In general, the reported preparation method of nickel-supported catalyst is convenient, economical and environment-friendly, which is in agreement with many green chemistry and sustainable development principles, and the method employs widely available starting materials.The authors are grateful for the support of the National Nature Science Foundation of China (NSFC, No. 21978074).

A simple and effective method for preparing a non-metallic ion-doped nickel-supported catalyst is reported. Using economical and recyclable fibre raw materials as carriers, nickel-supported catalysts were prepared by adsorption and reduction at room temperature. The nanoparticles dispersed and anchored on a rational support, efficiently inhibiting their aggregation and thus enhancing the catalytic activity. For the model catalytic hydrogenation of 4-nitrophenol by NaBH4, the N-B-NiP/steam-exploded poplar (SEP) and N-B-Ni5Fe5P/SEP catalysts exhibited much better catalytic performances than the other recently reported catalysts in terms of the catalytic activity (the reaction was completed within 10 min for both aforementioned catalysts), reaction rate constant (0.19 and 0.344 min−1, respectively) and the activity factor K (19 and 34.4 min−1·g−1, respectively). The catalysts showed activities for electrocatalytic hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) under ambient conditions. In general, the reported preparation method of nickel-supported catalysts is convenient, economical and environment-friendly, and is agreement with many green chemistry and sustainable development principles; further, it employs widely available starting materials.

The development of sustainable, clean, and environmentally friendly energy is becoming increasingly important to address the growing threat of energy exhaustion and greenhouse gas emissions that contribute to dangerous climate change (Kılınç and Sahin, 2018; Chen et al., 2011). Hydrogen (H2) is a potential alternative to meeting the world's increasing energy demand while also serving as an eco-friendly energy carrier molecule for future applications (Kılınç and Sahin, 2018). From the standpoint of the application, the safe generation, storage, and transportation of H2 are necessary conditions (Cai et al., 2016). Chemical hydrides are considered suitable materials for all of the applications as mentioned above. Chemical hydride hydrolysis is gaining popularity as a potential in-situ H2 supply method for proton exchange membrane fuel cells (PEMFCs) (Crisafulli et al., 2011; Kassem et al., 2019). Amongst chemical hydrides, sodium borohydride (NaBH4) is a preferable material for H2 storage and generation due to its high H2 capacity (∼10.9 wt%) (Lee et al., 2021; Abdelhamid, 2021; Abdelhamid, 2021), high stability in alkaline solution (Ritter, 2003), pure H2 production (Kim, 2004), recycling of the by-products (Calabretta and Davis, 2007; Santos, 2010; Santos and Sequeira, 2011) and non-flammable and less expensive (Chinnappan et al., 2011). Theoretically, one mole of NaBH4 can produce four moles of H2 in water (Eq. (1)) (Yao et al., 2020): (1) NaB H 4 + 2 H 2 O catalyst → 4 H 2 + N a B O 2 Self-hydrolysis of NaBH4 is slow; thus, the addition of an appropriate catalyst can greatly accelerate the hydrolysis reaction (Wechsler et al., 2008). Hitherto, various nano-catalytic materials (e.g., Pt, Ru, Rh, Pd, Ni, Co, Fe, and their alloys) have been used in the hydrolysis reaction of NaBH4 (Dinc et al., 2012; Oh et al., 2015; Shen et al., 2015; Ding et al., 2010; Park et al., 2008; Arzac et al., 2012; Lee et al., 2021; Larichev et al., 2010; Xu et al., 2008; Chen and Kim, 2008). Ni-based catalysts provide objective interest as a catalyst because of their low cost and environmentally friendly construction (Ozay et al., 2011). However, because Ni has a high energy surface and magnetic properties, it must be dispersed and stabilized by appropriate materials to achieve long-term durability without the formation of aggregates (Wang et al., 2021). Furthermore, the formation of NaBO2 as a by-product during NaBH4 hydrolysis leads to deactivation of the catalyst surface (Chinnappan et al., 2011). Because metal-based catalysts are typically used in powder form, they are inconvenient for start-and-stop applications. Separation of the catalyst powder from the reaction media and the possibility of catalyst particle aggregation significant practical issues (Chinnappan and Kim HJIjohe, 2012). The supporting materials significantly impact catalyst activity and durability (Li et al., 2012). Polymer substrates, as we know, have flexible design structures and are easily separated from reactants. As a result, various Ni–polymer hybrids with varying morphologies have been synthesized using various preparation methods (Kılınç and Sahin, 2018; Chen et al., 2011; Cai et al., 2016; Sagbas, 2012; Seven and Sahiner, 2014; Liu et al., 2013; Yan et al., 2009; Chen et al., 2015; Cai et al., 2016; Özhava et al., 2015; Chen et al., 2009). They demonstrated a high value in H2 generation from NaBH4 while overcoming the mentioned above. As we know, the method of preparation and the morphology of the catalysts directly impact their catalytic activity. Polymer nanofiber membrane (PNFM) have been proposed as supporting materials for various NPs in various chemical reactions. When compared to other supporting materials, PNFM are easily recyclable and reused with high efficiency. Nanofibers are able to form a highly porous mesh therefore their usage is almost endless (Chinnappan et al., 2011). Li et al (Li et al., 2014), They prepared composite nanofibers by immobilizing Cobalt (II) chloride on polyacrylonitrile NFs, which demonstrated excellent catalytic performance and stability in H2 generation from NABH4 solutions. Kim and his group developed hybrid NFs based on polyvinylidene fluoride (PVDF) as a support substrate in the production of H2 from NaBH4 (Chinnappan et al., 2011); Y-zeolite/CoCl2-PVDF (Li et al., 2012), dicationic tetrachloronickelate (II) anion (dicationic ionic salt [C6(mpy)2][NiCl4]2)-PVDF (Chinnappan and Kim HJIjohe, 2012); Ni NPs-PVDF (Sheikh et al., 2011). They found that the prepared hybrid membranes have good catalytic activity and are reusable. Electrospun poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) has recently been introduced as a polymer electrolyte and membrane in various applications (e.g., fuel cells, dye-sensitized solar cells, lithium-ion batteries, and water separation) (Raghavan et al., 2008; Mališ et al., 2013; Vijayakumar et al., 2015; Zhang et al., 2021; Tian and Jiang, 2008). In comparison to other host polymers (e.g. poly(ethylene oxide) (PEO), poly(ethylene glycol), poly(urethane acrylate), (PVdF), poly(methyl methacrylate), and PEO-modified poly(methacrylate)), PVDF-HFP is regarded as the most suitable host polymer for preparing hybrid composites (Raghavan et al., 2008). It has a high affinity to absorb electrolyte solution with good chemical and electrochemical stability (Zhang et al., 2014). In this study, we exploited the hydrophobic properties of PVDF-HFP to deposit NPs on its surface because the salt contains tetrahydrate, which makes deposition of NPs on the surface preferable to embedding NPs supporting on PVDF-HFP NFs. This hypothesis has two consequences: (1) reduce polymer crystallinity and (2) increase solution uptake, which may improve contact between the NaBH4 and catalyst surface (Raghavan et al., 2008). Accordingly, PVDF-HFP is considered an efficient supporting material for NPs which is a chemically stable and easily recyclable polymer. To our knowledge, no research has been conducted on the preparation of Ni NPs@PVDF-HFP membrane NFs for H2 production from NaBH4 hydrolysis as an efficient and easily reusable catalyst material. Metallic Ni nanoparticles embedded in PVDF-HFP NFs were investigated as non-precious catalysts for H2 generation from NaBH4 in this study. The NFs presented here were prepared using an electrospinning technique and a chemical reduction process. Typically, electrospun nanofibers composed of NiAc and PVDF-HFP were dried and then reduced in-situ with NaBH4 to form Ni NPs supporting on PVDF-HFP NFs. The in-situ reduction process has been done in methanol solution. The utilized physicochemical characterizations indicated that the reduction process leads to form Ni supporting on the PVDF-HFP NFs.Nickel (II) acetate tetrahydrate (NiAc, 98 % assay), poly(vinylidene fluoride-co-hexafluoropropylene) ((PVDF-HFP), 98 % assay) with a molecular weight of 65,000 g/mol, and sodium borohydride (NaBH4, 98 % assay) were purchased from Aldrich Co., USA. N, N-dimethylformamide (DMF, reagent grade, 99 % assay), and acetone were brought from Fluka. All these chemicals were used without further purification.First, 15 wt% PVDF-HFP solutions were prepared by dissolved 1.5 gm PVDF-HFP in a mixture of DMF and acetone (4:1 wt ratio). Four different percentages of NiAc-based PVDF-HFP (10 % (HFB-10), 20 % (HFB-20), 30 % (HFB-30), and 40 % wt. (HFB-40)), NiAc was dissolved in a determined amount of DMF before added to the PVDF-HFP solution, have been added to PVDF-HFP solution in the separated laboratory glass bottles. The solutions were kept in the magnetic stirrer overnight. The prepared sol-gels were subjected to the lab-scale electro-spinner machine. The sol–gel was placed in a plastic capillary syringe. A copper wire was inserted inside the syringe from one side, and the other side was connected to a high-voltage power supply (positive electrode). The negative electrode was connected with a ground iron drum covered by aluminum foil to collect electrospun NF mats. 20 kV voltage was applied between syringe and drum. The collected electrospun NF mats were dried at 30 °C under a vacuum overnight. The pristine PVDF-HFP-free NiAc membrane has been prepared with the same procedure.Typically, determining amounts from prepared electrospun NF mats were immersed in a 500 mL beaker containing methanol solution. A determined amount of NaBH4 was added to the previous solution. The molar ratios between metals precursor and NaBH4 were adjusted at 1:5 to obtain a full reduction reaction. As soon as the mat was attached to the NaBH4 solution, the color was changed from green to black, and the membranes looked like a black dying piece of cloth (Fig. 1 ). The mat was left in solution until the gas bubble stopped. The produced mat was washed three times with deionized water and ethanol to remove any residues. Finally, the reduced mat was dried under vacuum at 30 °C overnight.The morphology of the prepared catalytic NFs was observed by scanning electron microscope (SEM, Hitachi S-7400, Japan), equipped with energy dispersive X-ray (EDX) before inspection samples were coated with gold. For the investigation of the morphology of PVDF-HFP membrane NFs and their interactions with the Ni NPs, Fourier transform infrared (FTIR), using the smart ATR-FTIR model “Nicolet iS 10” (Thermo Fisher Scientific, MA USA) equipped with the specular reflectance. The scanning range was 400–3500 cm−1, and the samples were placed on the top of the spectrophotometer. The catalysts' crystalline structure and crystal size were determined by X-ray diffraction (Rigaku Co., Japan) with Cu Kα (λ = 1.54056 Å). An X-ray photoelectron spectroscopy analysis (XPS, AXIS-NOVA, Kratos Analytical, UK) was conducted with the following conditions: base pressure of 6.5 × 10−9 Torr, resolution (pass energy) of 20 eV, and scan step of 0.05 eV/step. A thermogravimetric analyzer (TGA) was used for the thermal analysis and stability of NFs samples with a heating rate of 10 °C/min and nitrogen flow rate of 20 mL/min. The temperature range for the analysis was set between room temperature and 900 °C.The H2 gas produced during the reaction was passed through a tube and collected in an inverted burette using the water displacement method. The volume of hydrogen generated was calculated by measuring the change in the height of the water level in the burettes at different time intervals. The catalytic reaction was carried out in a reactor made up of Pyrex round bottom flask reactors. The volume of H2 produced was calculated using the water displacement method. The reaction vessel was immersed in a temperature-controlled water bath to control reaction temperature. The determined concentration of aqueous NaBH4 solution and amount of catalyst was added to the reaction vessel. The hydrolysis reaction's kinetics investigated by varying the amount of catalyst, NaBH4 amount, and temperature. It also investigated the long-term durability of the introduced membrane NFs via the recycling process.Proposing the electrospinning technique for preparing polymeric nanofibrous membranes could display many distinct features, including increased interconnectivity, flexibility, excellent porosity, and extraordinary surface-to-volume ratios (Gibson et al., 2001; Yousef et al., 2012). Among the commonly used polymeric chemicals for fabricating these films, PVdF-HFP is the most preferred due to its semi-crystalline nature, good thermal stability, increased dielectric constant, and hydrophobicity besides its piezo and pyroelectric characteristics (Shin et al., 2010; Kumar GGJJoMC. , 2011). Fig. 2 a and b show the low and high magnifications SEM image of electrospun PVdF-HFP NF mats after drying; as shown in the figure, a good nanofibrous structure without any beads are formed. Furthermore, nano-cracks appeared on the surface of the NFs. This could be due to the rapid evaporation rate of the acetone solvent during the electrospinning process before NFs reach the drum's surface that helps produce this nano-cracks structure as a suitable site for the nucleation of Ni crystals. The water content of the used nickel precursor salt appreciably improved the hydrophilicity of fabricated polymeric membranes with enhanced demixing rate inside the liquid–liquid phases to favor the formation of numerous pores onto their structure (Chen et al., 2015). Afterward, the analyte molecules could be easily trapped in these pores with the lowest diffusion resistance and facilitate the H2 evolution.Moreover, the presence of metals salts has a beneficial role in increasing the electrical conductivity and the gelation of the polymer solution with the generation of maximum elongation of a jet along its axis to produce extremely small-sized polymeric nanofibers (Kang et al., 2014) finally. During the in-situ reduction of Ni(II) ions used NaBH4 as a powerful and effective reducing agent in methanol media, metallic Ni NPs are produced and deposited onto the PVdF-HFP membrane surface. SEM images of electrospun Ni@PVdF-HFP NF membranes NFs were shown in Fig. [2c (HFB-10), d (HF-B20), e (HF-B30), and f (HF-B40)], the inset show the high magnifications SEM images. As shown in the figure, the rough and beads-free NFs are formed. Furthermore, reduced Ni ions are covered the surface of PVdF-HFP NF mats as they built up based on the nano-cracks present on PVdF-HFP NFs surface.During the methanolysis process (Aydın et al., 2020), sodium tetra methoxy borate (NaB(OCH3)4) is formed as a by-product (Equation (2) ). During washing, sodium tetra methoxy borate reacts with water (Equation (2) ) to produce sodium borate and methanol that wash out with excess water. (2) NaB H 4 + 4 C H 3 O H → N a B OC H 3 4 + 4 H 2 (3) NaB OC H 3 4 + 2 H 2 O → N a B O 2 + 4 C H 3 O H Using methanol for dehydrogenating NaBH4 has advantages over the use of water. One of them is related to the nature of the by-product. In methanolysis, NaB(OCH3)4 forms. Unlike NaB(OH)4, NaB(OCH3)4 does not have the propensity to polymerize into polyborates. Avoiding the NaB(OH)4 precipitation inhibits catalyst poisoning (Aydın et al., 2020; Lo et al., 2007). Hongming et al. (Zhang, 2020), prepared ultrafine Co NPs @ carbon nanospheres using hydrothermal and reduction processes as an efficient catalyst for H2 production from NaBH4 hydrolysis. They indicated that the reduction of Co ions in the ethanol solution media is better than in water media, in which partial NaBO2 was precipitated out with the Co NPs due to the insolubility of NaBO2 in ethanol could separate and further prevent the agglomeration of the Co nanoparticles. At last, the NaBO2 was washed off with DI water. The electrospun Ni2+/PVdF-HFP membranes had white-colored surfaces. The chemical reduction of these metallic ions tends to change the color of their related polymeric membranes into a black-look-like black dying piece of cloth (Fig. 1), which suggests the growth of tiny Ni NPs on the surface of PVdF-HFP membranes. In other words, the complete coverage of PVdF-HFP membranes surface with skin layers of black nickel dots might resemble the shell “Ni nanoparticles”-core “polymeric film” arrangements. An elemental mapping image of the HFP-40 membrane was presented in Fig. (3 a, b, and c) . It is clear that the high distribution of Ni NPs around the membrane NFs, is confirmed by the SEM images. EDX chart of HFP-40 membrane was presented in Fig. (4 a and b) . The related peaks of carbon, nickel, and fluorine were detected to ensure the successful fabrication of these composite membranes. The XRD study in Figs explored the crystal structure of PVdF-HFP and HFP-40 membranes. (5 a and b), respectively. Three main diffraction planes were observed at 2θ values of 18.04°, 20.24°, and 36.19° in the XRD chart of PVdF-HFP membrane corresponding to (100), (020), and (021) crystal indices, respectively (Stephan et al., 2006)[see Fig. 5 a]. Besides these defined PVdF-HFP membrane peaks, nickel species were identified through three characteristic planes at 42.77° (111), 49.75° (200), and 73.60° (220) to ascertain the formation of the face-centered cubic crystalline structure of nickel [JCPDS card No. 04–0850] (Barakat and Kim, 2009) [see Fig. 5 b]. The particle size of Ni nanoparticles is calculate used Scherrer equation (Yao et al., 2016), it was found to be 19.5 nm. It is worth mentioning that the reduced membrane in water media showed the same color as pristine PVDF-HFP spectra. As the membrane kept its green color. The stability of PVdF-HFP and Ni@PVdF-HFP membranes when subjected to elevated temperatures was examined through TGA charts in Fig. 6 . One weight loss section was observed for bare HFP-10 film at 420 °C due to the random scission of its units during the degradation process (Kim et al., 2011; Babu et al., 2015). This step was shown at a lower temperature value [346 °C] after nickel species were incorporated into HFP-10 film, besides a small change at 95 °C when physisorbed water molecules were evaporated. The presence of Ni NPs in the structure of this polymeric membrane was responsible for weakening the van der Waals′ interacting forces between its chains. This,facilitated the degradation of the metal-supporting polymeric membrane at decreased temperatures when related to the case at bare film (VijayaáKumar, 2001). FTIR spectra of PVdF-HFP and Ni@PVdF-HFP membranes were also described in Fig. 7 . Common vibrational bands were noticed in both charts that depicted the polymeric membrane. Its α and β phases were confirmed by their respective peaks at frequency values of 749 and 837 cm−1 (Kumar and Nahm, 2008). Another two specific vibrational bands were centred at 672 and 872 cm−1 that were assigned for C − F and CH2 wagging of vinylidene units in the amorphous phase of PVdF-HFP film. Furthermore, the symmetric C − F stretching, CF2 stretching, and deformed vibrations in this membrane were also shown through their corresponding bands at 1071, 1175, and 1400 cm−1 (Mandal et al., 2014). Additional two peaks appeared when nickel precursor salt was introduced during the polymeric film fabrication. The formation of Ni − O species into this nanomaterial was supporting by its stretching vibration peaks at 1561 cm−1 (Nath, 2014).Their shape influenced the catalytic activity of the catalysts. Shape-anisotropic nanostructures possessing more active sites for catalysis led to improved catalytic performance. NFs have a large surface area compared to other nanostructures, leading to better performance as catalyst support. Nonetheless, it was demonstrated that the nanofibrous morphologies corresponding to the long axial ratio provide a significant performance when compared to other nanostructures. In this regard, NFM outperforms typical powder-like catalysts in terms of catalyst separation and capacity to be reused. Ni@PVDF-HFP MNFs were prepared by electrospinning technique, tested as a catalyst in the hydrolysis of NABH4 and found to be the highly active catalyst to generate H2. The hydrolysis of 1.34 mmol of alkaline NaBH4 occurred without any catalyst; however, it obtained 28 mL after 60 min. The addition of pristine PVDF-HFP MNFs to 1.34 mmol of alkaline NaBH4 did not significantly affect the NaBH4 hydrolysis. However, the H2 generation is increased with a reduction in the time using electrospun Ni@PVdF-HFP MNFs (Fig. 8 ), as shown by the variation in the catalytic activity using different Ni contents. Experiments were performed by adding 100 mg of MNFs from all formulations, in the separated glass reactor, into 1.34 mmol of alkaline NaBH4 at 25 °C to determine the best activity of the synthesized membrane NFs. As manifested in the figure the HFP-40 (103 mL in 60 min) membrane, NFs have shown the highest H2 generation rate than other ratios [HFB-10 (68 mL in 60 min), HFB-20 (81 mL in 60 min), and HFB-30 (93 mL in 60 min), so it has chosen to be used in further experiments. The fabricated MNFs have shown better catalytic activity, and it produced 103 mL in 60 min using 1.34 mmol NaBH4 and 100 mg catalyst at 25 °C, than PVDF-[C6(mpy)2][NiCl4]2 NFs composite catalyst (Sheikh et al., 2011), ]; it was produced about 140 mL H2 in 60 min used 158.72 mM NaBH4 and 40 mg from the catalyst contain 2.5 % Ni, and 40 % Ni@TiO2 (Dönmez and Ayas NJIJoHE. , 2021), it was produced 37.89 mL H2.g−1 cat min−1 used 100 mg NaBH4, 100 mg catalyst, 5 mL 0.25 M NaOH at 20 °C. Jaeyeong et al. (Lee et al., 2019), demonstrated that the Ni powder needed the longest time (280 min) to generate the same amount of hydrogen (500 mL) compared with Ni thin film (240 min), and etched Ni foil (60 min) used 1.5 g NaBH4 and 0.01 g from catalysts at 25 °C although Ni powder high surface area. This could be due to the easily contaminated or oxidized. Accordingly, the rougher surface, which was made by in situ difficult reduction reaction of NiAc to metallic Ni on the surface of PVDF-HFP, made the catalyst more efficient. Catalytic hydrolysis of NaBH4 depends on the type of catalyst and other parameters such as NaBH4 concentration, reaction temperature, catalyst amount, and reusability. For this reason, the effect of these parameters on the hydrolysis of NaBH4 was investigated in the presence of the HFB-40. The effect of NaBH4 concentration on the reaction rate has been tested (Fig. 9 a) using 100 mg of HFP-40 MNFs at 25 °C. As shown in the figure, the initial H2 generation rate is linearly increased with an increase in NaBH4 concentration. When the amount of NaBH4is increases from 50 mg to 125 mg, the hydrogen production rate increases from 255.2 mL.g−1 cat min−1 to 689.9 mL.g−1 cat min−1 , respectively. The estimated slope of the best-fit line was 0.74 (Fig. 9 b), which clarifies that the H2 production rate follows pseudo-first-order kinetics concerning NaBH4. This is due to the use of low NaBH4 concentration as the higher concentration follows the pseudo-zero order reaction, in which at higher concentration, the viscosity increases, the reactant diffusion resistance, reaction rate decreases, and the by-product (NaBO2) formed in hydrolysis can adsorb on the catalyst surface and block the active sites (Ozay et al., 2011; Sagbas, 2012; Walter et al., 2008; Saka and Eygi, 2020). Our study was investigated at low concentration compared to the zero-order reaction. To examine the influence of the reaction temperature effect (Fig. 10 ), experiments were executed using 2.67 mmol of alkaline NaBH4with 100 mg of HFP-40 MNFs at temperatures ranging from 298 K to 328 K to obtain the activation energy (Ea) of NaBH4hydrolysis catalyzed HFB-40 using the Arrhenius equation (Eq. (4) ). (4) ln k = ln A - E a RT Where k is a rate constant, A is a pre-exponential factor, R is a gas constant, and T is the reaction temperature. As expected, the H2 generation increased as the reaction temperature and H2 volume vs reaction time changed linearly (Fig. 10a). As shown in the figure, the reaction time for H2 generation has been reduced at elevated temperature. Furthermore, the H2 generation is increased. As seen in Fig. 10, when the reaction temperature rises from 25 °C to 55 °C, the hydrogen yield increases from 50 % to 100 % in 28 min. This obtained is compatible with the studies in the literature (Ekinci et al., 2020; Wei et al., 2017; Li et al., 2014). The rate (k) value is obtained from the linear portion of temperature graphs. From Arrhenius plot ln (k) versus 1/T in Fig. 10 b and Arrhenius equation (Eq. (2) ), the Ea was estimated to be 23.52 kJ mol−1. Activation energies of non-noble metals were reported in the literature between 16.28 and 42.45 kJ mol-1 (Dönmez and Ayas NJIJoHE. , 2021; Tamboli et al., 2015; Hua et al., 2003; Soltani, 2020; Kılınç and Şahin, 2019). Activation energy values of prepared NFs and Ni-based catalysts are compared (Table1 ), indicating superior catalytic performance of the introduced Ni@PVDF-HFP. The H2 generation gathered over time by employing different catalyst amounts (100 to 250 mg) of HFP-40 MNFs is given in Fig. 11 . As expected, the H2 generation increased when the amount of HFP-40 was increased as more catalyst provides more active sites for NaBH4 dehydrogenation. One can note that the H2 generation from the hydrolysis reaction of NaBH4 proceeds very slowly and then stops without catalyst (Fig. 8). The H2 generation increased as the catalyst amount increased (Fig. 11 a), attributed to the fact that hydrolysis reaction of NaBH4 is a catalyst-controlled reaction. This phenomenon can be attributed to the fact that the reaction rate increased due to the active sites of the catalyst increasing in direct proportion to catalyst amount, called structure sensitivity. Thus, it is clear that H2 generation can be determined by controlling catalyst amount. Fig. 11 b shows the ln H2 generation rate vs ln HFB-40. The determined slope of the best-fit line is 1.29, suggesting that the produced H2 agrees with pseudo-first-order kinetics regarding the amount of catalyst. According to the results obtained from the effect of catalyst concentration, NaBH4 concentration, and reaction temperature, the NaBH4 dehydrogenation kinetic equation catalyzed by the introduced NFs membrane can be written according to Eqs. (5), (6), and (7). (5) r = - d SBH d t = k HFB - 40 1.29 SBH 0.74 (6) k = A e - E a RT → ln k = ln A - E a RT (7) r = - d SBH d t = 22337.01 e 2828.7 T HBF - 40 1.29 SBH 0.74 Gibbs free energy of activation (ΔG, (kJ mol−1)) can be determined using thermodynamic data (activation enthalpy (ΔH, (kJ mol−1)) and activation entropy (ΔS, (J mol − 1 K−1)) according to Eqs. (8) and (9): (8) ln k D = ln k B h + Δ S R - Δ H RT (9) Δ G = Δ H - T Δ S Where kD = (k/T), KB and h are the Boltzmann constant (1.381 × 10−23 J K−1) and the Planck constant (6.626 × 10−34 J s−1), respectively. According to Eq. (9) and Fig. 10 , ΔH and ΔS is estimated to be 20.92 kJ mol−1 and 0.0272 kJ mol−1, respectively. ΔG equation can be summarized as follows: (10) Δ G = 20.92 - 0.0272 T ΔG values are estimated to be 12.81 and 11.9 kJ mol−1 at 298 and 323 K, respectively. This main that the reaction's spontaneity is directly increased with the temperature.Reusability and stability are significant factors in determining whether the catalyst is suitable for a practical application. The reusability of the HFB-40 membrane NFs was tested repeatedly ten times to confirm its stability in the presence of 2.67 mmol of alkaline NaBH4, at each cycle, with 100 mg of HFP-40 MNFs at 25 °C (Fig. 12 ). The membrane NFs was used ten times without makeup, reactivation, or regeneration. As seen in the figure, the catalytic activity of the membrane NFs catalyst is maintained even if the use is repeated for up to 8 cycles as the H2 generation rate remains unchanged. The slight decreases have been demonstrated as the cycle number increases. The same amount of H2 has been generated extended reaction time. This could be due to the precipitate of reaction products on the membrane that inhibited the active metal sites as the membrane is used without cleaning during all cycles, which this accumulation harms the hydrogen generation rate. The slight decrease in catalytic activity after eight cycles may be due to the increase in the number of boron products on the membrane surface as an increase in the solution viscosity (Yang et al., 2017), which decreases the accessibility of active sites or blockage of pores in the membrane NFs. XPS of the HFB-40 catalyst confirms this after reuse for ten cycles (Fig. 13 ). The Ni 2p XPS spectrum of Ni 2p spectra shows two main peaks positioned at 856.1 and 873.6 eV, which are assigned to Ni 2P3/2 and Ni 2P1/2, together with two corresponding satellite peaks located at 861.7 and 880.6 eV (Zhu et al., 2017). As depicted in Figure, the Ni 2p3 peak located at 856.1 eV demonstrated the formation of Ni(OH)2 after reuse ten times, which is similar to the value obtained in literature (Yao et al., 2016). Furthermore, Na 1 s peak located at 1072.2 eV and B 1 s peak located at 180 eV appeared after reuse, ten cycles because the Na ions are generated from NaBH4 hydrolysis, which mainly constitutes contaminants, are dissolved in the solution during the hydrolysis. The SEM image of HFB-40 catalyst after reuse for ten cycles (Fig. 14 ) indicates that the NFS kept their nanofibrous structure.Metallic Ni NPs @ PVDF-HFP membrane NFs catalysts have been successfully prepared via electrospinning technique followed by in-situ reduction of metal ions to metallic Ni. The prepared metal catalysts have shown an excellent catalytic performance in the H2 generation from NaBH4. The sample was composed of 40 %wt. NiAc showed the highest catalytic activity compared to the other formulations. Whereas 103 mL of H2, from the hydrolysis of 1.34 mmol NaBH4, was produced using 40 % wt. NiAc tis compared to 68 mL, 81 mL, and 93 mL for 10 % wt., 20 % wt., and 30 % wt. NiAc, respectively, in 60 min at 25 °C. The increase in the temperature lead to an increase in the hydrogen production rate and obtained low activation energy. (23.52 kJ mol−1). The kinetics study revealed that the reaction was pseudo-first-order in sodium borohydride concentration and catalyst amount.Furthermore, the catalyst exhibits satisfactory stability in the hydrolysis process for ten cycles. Because of its easy recyclability, the introduced catalyst has a wide range of potential applications in the generation of H2 from sodium borohydride hydrolysis. Considering that the eco-friendly and inexpensive Ni@PVDF-HFP membrane NFs are catalytically effective with superior reusability, they should have potential application in the H2 generation from the sodium borohydride hydrolysis. Abdullah M. Al-Enizi: Conceptualization, Investigation, Methodology, Writing – original draft. Ayman Yousef: Conceptualization, Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing. Shoyebmohamad F. Shaikh: Conceptualization, Investigation, Methodology, Writing – review & editing. Bidhan Pandit: Methodology, Writing – review & editing. M.M. El-Halwany: Investigation, Methodology, Validation.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors extend their sincere appreciation to the Researchers Supporting Project number (RSP-2021/370), King Saud University, Riyadh, Saudi Arabia, for the financial support.Supplementary data to this article can be found online at https://doi.org/10.1016/j.arabjc.2022.104207.The following are the Supplementary data to this article: Supplementary video 1

Nickel nanoparticles (Ni NPs) supported on Poly(vinylidene fluoride-co-hexafluoropropylene) nanofibers (PVDF-HFP NFs) were successfully synthesized through electrospinning and in-situ reduction of Ni2+ salts into the surface of PVDF-HFP NFs to form metallic Ni NPs@PVDF-HFP NFs. Different percentages of nickel acetate tetrahydrate (NiAc) (10 %, 20 %, 30 %, 40 % wt.) based PVDF-HFP. The formation of tiny metallic Ni NPs @PVDF-HFP membrane NFs was demonstrated using standard physiochemical techniques. Nanofibers membranes have demonstrated good catalytic activity in H2 production from sodium borohydride (NaBH4). The sample composed of 40 %wt Ni showed the highest catalytic activity compared to the other formulations. Whereas 103 mL of H2, from the hydrolysis of 1.34 mmol NaBH4, was produced using 40 wt% NiAc compared to 68 mL, 81 mL, and 93 mL for 10 wt%, 20 wt%, and 30 wt% NiAc, respectively, in 60 min at 25 °C. The hydrogen generation has been enhanced with an increase in the Nanofibers membrane amount and reaction temperature. The latter results in a low activation energy (23.52 kJ mol−1). The kinetics study revealed that the reaction was pseudo-first-order in sodium borohydride concentration and catalyst amount. Furthermore, the catalyst exhibits satisfactory stability in the hydrolysis process for ten cycles. Because of its easy recyclability, the introduced catalyst has a wide range of potential applications in the generation of H2 from sodium borohydride hydrolysis.

In recent years, interest in lignocellulosic biomass valorization into value-added chemicals and fuels has gained importance due to decreasing fossil raw materials [1–3]. Along with hemicellulose and cellulose, lignin is one of the three most important biopolymer components of lignocellulosic biomass and comprises an aromatic structure with ether linkages, methoxy-, and hydroxyl groups. Depending on the species, lignin makes up about one-third of the solids in wood. Lignin, which consists of p-hydroxyphenyl (H), guaiacyl (G), and syringyl (S) units, is an amorphous, three-dimensional polymer. Cross-linking occurs via C-O-C (β-Ο-4, α-Ο-4, 4-Ο-5) and CC (β-1, β-β, 5–5) bonds. The largest proportion of these are the stable C-O-C bonds, with the 4-O-5 bond having the strongest bond dissociation energy at 330 kJ mol−1. This makes effective cleavage of such C-O-C bonds the key factor in the depolymerization of lignin [2,4]. A general structure of lignin is shown in Fig. 1 . A variety of different strategies for the depolymerization of lignin have been investigated, such as hydrocracking [5], hydrogenolysis [6], pyrolysis [7], and oxidation [8,9]. The catalytic pathway is considered to be the key technology for lignin utilization technologies. In current industrial processes, more than 80% use various catalysts for the synthesis of a wide range of chemical, petrochemical, and biochemical products as well as polymers [10–12]. Homogeneous catalysis has limited application in the industry compared to heterogeneous catalysis for depolymerization of lignin, due to their inherent disadvantages like the difficult separation from the reaction mixture, and corrosiveness for example in the case of soluble acid catalysts [13]. The reusability and recyclability of heterogeneous catalysts are attracting the interest of the scientific community [14]. Precious metal-based catalysts, such as Pt, Pd, and Ru have high activity in the depolymerization of lignin due to their high hydrodeoxygenation activity [15,16]. To date, many different metal-loaded heterogeneous catalysts have been investigated for lignin depolymerization. The substrates range from model compounds to biomass-derived lignin. Most of these processes are based on reductive depolymerization strategies, which additionally require high pressures of externally supplied hydrogen [14]. Sturgeon et al. used a nickel-loaded HTC catalyst which acts similarly to the base-catalyzed mechanism due to the basic nature of the support material [14]. Metallic nickel has shown particular promise for the formation of phenols from lignin, as it can cleave aryl-aryl C-O-C bonds and C-OH bonds into side chains with CH3 or CH2 functions, usually yielding C1-C3 alkane-substituted guaiacol as the main products [17]. Besse et al. investigated the catalytic conversion of lignin model compounds with a Pt/C catalyst in ethanol/water mixtures, using ethanol as both H-donor and solvent. The study highlights the high selectivity of water/ethanol mixtures at hydrothermal conditions concerning hydrogenolysis products and hydrogenation products. CO bond cleavage of the model compounds could be achieved by H-transfer hydrogenolysis without externally added hydrogen [18]. Huang et al. describe an increase in the rates of Guerbet, esterification, and alkylation reactions when HTC is loaded with up to 20% Cu. A combination of Cu and basic sites also facilitates the dehydrogenation of ethanol, generating the hydrogen needed for the hydrogenolysis and hydrodeoxygenation reactions, resulting in a higher yield of monomers [19]. Chaudharya and Dhepe describe a range of transition metal-free catalysts suitable for base-catalyzed depolymerization of lignin, including zeolites, metal oxides, hydrotalcites, and hydroxyapatite. Although the reaction conditions described allow the preparation of relevant monomer and oligomer building blocks, they do so only in low yield and with high catalyst input [20]. Support materials equipped with transition metals are another class of catalysts suitable for the degradation of lignin. For example, US9631146B2 describes a process using nickel on a double-layered hydroxide as a catalyst. However, studies of such a system showed that the yield of desired products was relatively low due to a high proportion of the coke fraction in the reaction product [21]. The objective of the present study was to provide a process for the catalyzed degradation of lignin with a high yield and high selectivity for phenolic building blocks with minimal formation of the coke fraction and to find a process for degradation under mild reaction conditions, i.e. low temperature, low pressure and without inert gas or hydrogen atmosphere. In addition, part of the present study was to realize a significant reduction in the amount of catalyst required [22]. Due to the great potential of precious metal-containing catalysts, the advantages of HTC as an alkaline support material, and the described advantage of ethanol/water mixtures as solvent and H-donor for the depolymerization of lignin, a catalyst screening followed by optimization of the most promising catalyst system (5% Pt-1% Ni/HTC) via DoE was performed in this study--Organosolv lignin was purchased from Chemical Point. All catalysts used in this work were supplied from Heraeus Deutschland GmbH & Co. KG. Ethanol (EtOH) was obtained from AustrAlco, ethyl acetate (EtOAc), tetrahydrofuran (THF) and sodium sulfate from Roth, and hydrochloric acid (HCl) from VWR. All chemicals in this work were used without further purification.All catalysts were prepared by impregnation. The preparation of the 5%Pt-1%Ni/HTC catalyst is described as an example. 5 g Pt as platinum(II) nitrate (Pt(NO3)2 solution with 15.2% Pt, Heraeus) and 2 g Ni as nickel(II) nitrate hexahydrate (Ni(NO3)2*6 H2O with 20% Ni, Merck) were diluted to 30 mL and homogenized. The solution was added to 93 g HTC (Sasol, BET surface area 19 m2 g−1) and the mixture was homogenized. The mixture was dried overnight at 110 °C under a nitrogen atmosphere in a vacuum. This was followed by thermal treatment for 14 h in an oxygen-containing atmosphere, during which the temperature was gradually increased to 250 °C. Finally, the material was treated for 16 h with forming gas (95 vol% nitrogen, 5 vol% hydrogen) at up to 250 °C.The characterizations of the catalysts were carried out by Heraeus Precious Metals Company in Hanau, Germany.The determination of the BET surface area of the catalysts was carried out at −195.8 °C using the Q-Surf SA-9601 measuring device from Horiba. Each catalyst was analyzed using a 3-point measurement with a 70% helium/30% nitrogen gas mixture at three different points of the adsorption isotherm (i.e. at different relative pressures). To remove gases and vapors adsorbed on the catalyst, especially water, the samples were degassed at 250 °C for 60 min before measurement.The precious metal surface area of the catalysts was determined via Carbon monoxide (CO) adsorption. Each catalyst was first reduced for 20 min at 400 °C under forming gas consisting of 95% argon and 5% hydrogen in a closed container. CO (with helium as carrier gas) was then dosed in pulses into the container. This was done until constant CO peaks were detected behind the catalyst. By determining the peak area of the dosed CO and the peak area of the converted CO, the amount of CO absorbed by the catalyst was determined. From the amount of CO absorbed obtained, it was calculated how much CO was stored per amount of catalytically active composition used. From the measured amount of CO stored at the active centers, the surface area of the active precious metal centers (often referred to as precious metal surface area) could be determined by conversions.The pH of the catalysts was determined with a WTW inoLab pH 7310 instrument. This was done by adding 10 mL of water to 800 mg of dry catalyst in a test tube and mixing well. For the ionic equilibrium to be established, the measurement is carried out after one hour.Powder XRD patterns of the catalysts were recorded with an STOE Stadi P Diffractometer in the 2θ range of 5–81° using a Cu-Kα radiation source (k = 1.54056 Å). The operating voltage and current were 40 kV and 30 mA, respectively, with a scanning speed of 4 deg. min−1 for data acquisition.Focused ion beam scanning electron microscopy (FIB-SEM) images were acquired by using a Field emission SEM of ZEISS, 1540XB. All the samples were sputtered with 5–10 nm of gold.Transmission electron microscopy (TEM) investigations were conducted using an FEI Talos 20–200 transmission microscope at 200kV, The measurements were performed in TEM mode. Energy-dispersive x-ray spectroscopy was used to detect differences in local chemical composition. Carbonized Cu-grid, Plano Mesh 300, was used. In scanning transmission electron microscope (STEM) collected images using Bright Field (BF) and high-angle annular dark-field (HAADF) detector. The pictures from HAADF and BF are in the same folder. BF is almost the same that you see in TEM mode, HAADF shows Z contrast (atomic number), which means that Pt or Pd particle will be very visible on C or Al matrix.The heterogeneously catalyzed depolymerization experiments were carried out in a 450 mL stainless steel autoclave (PARR, 4871 Process Controller, Software: SpecView3). The scheme of lignin depolymerization and the workup of the fractions was similar to the previous study [2] and is schematically shown in Scheme 1 . In a typical experiment, 20 g of lignin and 4 g of heterogeneous catalyst were used. The catalyst:lignin ratios described in the literature for the use of supported HTC catalysts for the depolymerization of lignin (2:1 wt:wt Strungeon et al.; 1:2 wt:wt Huang et al) are difficult to implement for large-scale use of catalysts for the depolymerization of lignin. Therefore, in this study, we chose to screen for a much lower ratio of 1:5 wt:wt. The reaction parameters for all screening experiments were set to 200 °C, 30 min residence time, and an ethanol/deion. water mixture with an Ethanol concentration of 45.9% (v/v). Lignin was suspended in 200 mL of the reaction mixture and treated in the Parr reactor at a stirring speed of 5 s−1. The gas produced during the reaction was removed and determined volumetrically. Subsequently, the reaction mixture was separated into the product fractions lignin oil (monomer yield), lignin tar (oligomer yield), lignin coke (coke formation), and water solubles. The reaction mixture was transferred in a beaker with a spatula and 45.9% EtOH (v/v) and acidified with concentrated HCl (37 wt%) to a pH value of 2 to precipitate the lignin tar, which were separated by vacuum filtration through a suction filter with a porosity of 3. The solid was washed three times with diluted HCl. The aqueous phase was extracted three times with EtOAc. The organic phases were combined, dried with sodium sulfate, and filtered. The EtOAc was evaporated using a rotary evaporator to obtain the EtOAc soluble products (lignin oil). The solids (lignin tar, coke, and catalyst) were suspended in THF to dissolve the THF soluble products. The solid remaining after filtering in the suction filter consisted of lignin coke and heterogeneous catalyst. The liquid phase was evaporated using the rotary evaporator to obtain the THF soluble products (lignin tar).The yields of all fractions were calculated thus: (1) yield of lignin oil wt % = weight of lignin oil weight of lignin ∙ 100 (2) yield lignin tar wt % = weight of lignin tar weight of lignin ∙ 100 (3) coke formation wt % = weight of lignin cok e − weight of catalyst weight of lignin ∙ 100 (4) water solubles wt % = weight of water solubles weight of lignin ∙ 100 The most promising catalyst system was optimized using Statistical DoE (Design Expert ®) to maximize the value-added fractions of lignin oil and lignin tar products and minimize the coke fraction. For this purpose, a reduced response surface central composite design (CCD) was planned. In this experimental plan the three numeric variables temperature (T), residence time (t), and the catalyst amount were studied. The stirrer speed was kept constant at 300 rpm in all experiments. For this design, 13 randomized experiments, including 3 replicates at the center point were performed. Randomization was done to minimize unobserved effects which could influence the results. The influence of each numeric variable and also the interaction between the variables could be studied with this experiment. The effect of the three factors temperature, residence time, and catalyst amount on the depolymerization of organosolv lignin was evaluated using the gravimetric yields of the product fractions lignin oil and lignin tar and the unwanted coke formation as response variables. After analyzing the experiments by ANOVA with a 95% confidence level, an optimization of the reaction parameters was calculated with the software “Design Expert” and a verification experiment was performed.Organosolv lignin was isolated from bluebells and corn by ChemicalPoint in Germany. The characterization was done by NREL analysis: total solids (d.m.) [23], ash [24], carbohydrates and lignin [25]. The relative ratios of the main linkages were determined by 2D-HSQC NMR techniques following reported procedures [26].For determination of the phenolic OH groups the Folin Ciocalteu method [2] and the 31P NMR method [27] with cholesterol as an internal standard was applied. Experimental details for sample pre-treatment and derivatization reaction can be found in the Supporting information.The analysis of the product fractions was done similarly to Süss et al. [2]. Molecular weight determination of the lignin oil, lignin tar, and the origin OL were done by GPC (Thermo Scientific, Dionex ICS 5000+) using a PSS MCX analytical 100A + 1000A + 100000A column (8 mm × 300 mm, Thermo Fischer). It was performed at 30 °C with 0.1 mol L−1 NaOH as eluent at a flow rate of 0.5 mL min−1. For detection, a wavelength detector set at 280 nm was used. The system was calibrated using standards from PSS (Polymer Standard Service) with a molecular range of 891–976,000 g mol−1 and Vanillin. For determination of the phenolic OH groups of both fractions, lignin oil and lignin tar, and the used OL the Folin Ciocalteu method [2] and the 31P NMR method [27] with cholesterol as internal standard was applied. The relative ratios of the main linkages were determined by 2D-HSQC NMR techniques following reported procedures [26]. Additionally, the lignin oil products were analyzed quantitively and qualitatively by a Shimadzu GC MS QP 2020 instrument with an HP SM5 capillary column (60 m × 0.25 mm × 0.25 μm) and electron impact ionization (EI). As carrier gas, He was used with a split ration of 10. The oven temperature was increased from 50 °C to 300 °C with a 10 °C min−1 heating rate. For oven temperatures of 120 °C and 280 °C, holding times of 5 and 8 min were set. The injection temperature was set to 250 °C. For quantification, the internal standard toluene and an external calibration with 41 possible monomer substances that can be formed during depolymerization were used.In our study, we investigated the heterogeneously catalyzed depolymerization of OL in an ethanol-water mixture (45.9% v/v). The HTC-supported catalysts were prepared in cooperation with Heraeus GmbH, Germany. Because of the high purity and total lignin content of 98.24 wt%, no further purification before the depolymerization experiments of the purchased OL was necessary. Analysis results for the determination of the dry matter, ash content, lignin content (acid-soluble, acid-insoluble), and sugar contents, are shown in Table 1 .All prepared catalysts were characterized by Heraeus GmbH using precious metal surface (PM surface) (CO / m2 g−1), BET surface area (m2 g−1) analysis, and determination of the pH value as listed in Table 2 . Fig. 2 shows the XRD pattern of the 5%Pt-1%Ni/HTC catalyst, which was found to be the most suitable catalyst for depolymerization in ethanol/water after catalyst screening. Typical characteristic bands for HTC were found at 11.4°, 34.2°, and 60.0°. The peaks arising from mixed nickel oxides are found at 22.5°and 37.8°. The fact that nickel is mainly present as Ni(OH)2 and other mixed nickel oxides indicate an interaction of nickel with the HTC support. These results are in agreement with results from the literature [14]. Fig. 3 shows three selected catalysts, 0.5%Pt/HTC, 5%Pt/HTC and 5%Pt-1%Ni/HTC. The amount of noble metal or additional loading of Ni does not influence the optical appearance of the support material, as shown in figures d, h and l. The shell shape, which is typical for this substrate, is retained regardless of the loading. In the TEM images, a significant difference between the individual catalysts is observed. With increasing loading, the particles with an average particle size of 2.51 nm for the 5%Pt/HTC catalyst (d) and 1.73 nm for 5%Pt-1%Ni/HTC (g) are significantly more uniform in size and much more homogeneously distributed on the catalyst surface than the 0.5%Pt/HTC (a) catalyst with an average particle size of 7.31 nm which is consistent with the literature [4]. It can be seen that the additional supporting of nickel leads to a different distribution of the metallic particles than when the substrate is only supported with Pt. This effect needs to be further investigated with additional analytic techniques.On the SI-EDS-HAADF images, the actual active metal particles are visible (Pt red, Ni yellow).The influence of the platinum loading and also the loading with a second metal (Cu or Ni) of the HTC supported catalysts on the product yields were examined in detail (Fig. 4 ).20 g of OL and 4 g of catalyst were used for each experiment to compare the different catalysts in terms of yields of the value-added fractions (lignin oil and lignin tar) and reduction of coke formation. The reaction parameters were set at 200 °C, 30 min, and 300 rpm. After depolymerization, the fractions were separated as shown in Scheme 1.Gravimetrical yields of all screening experiments are shown in Fig. 4. Without a catalyst, only small amounts of the value-added product fractions (3 wt% lignin oil and 15 wt% ligni tar) were achieved. The undesired coke formation with 58 wt% is by far the largest fraction formed without a catalyst under these reaction conditions. The influence of the Pt loading showed an increase of the lignin oil fraction and lignin tar fraction from 13 to 20 wt% (lignin oil) and 43 to 53 wt% (lignin tar) and a reduction of the coke formation from 30 to 20 wt% by increasing the Pt loading from 0.5% to 5%. To evaluate the influence of Cu and Ni, depolymerization experiments with 2% Cu on HTC and 2% Ni on HTC were conducted. Both metals showed a significantly higher coke formation (Cu: 27 wt%, Ni: 24 wt%) with approximately 60 wt% higher lignin tar yield compared to the 5% Pt catalyst. Comparing the pure 5%Pt loading with the combination of 5%Pt on HTC with a second metal at 1% or 2% (Cu or Ni) each, the yield of lignin oil decreased from 20 to 16 and 17 wt%, respectively. The yield of lignin tar was increased from 53 to 59 wt%. The positive effect of Cu and Ni on the reduction of coke formation described in the literature could be verified in our study under the tested conditions, compared to the reaction without catalyst [17,19,28]. Comparing the coke formation of the depolymerization reactions catalyzed with 2%Cu/HTC or 2%Ni/HTC with the Pt/HTC catalyzed ones, the coke formation of Cu or Ni is more effectively suppressed, but the yield of lignin oil decreases significantly, whereby loading with Ni achieves both less coke formation and higher yields of lignin oil than that with Cu. To gain the benefits of Pt and the additional loading of Cu or Ni, a combined loading was tested. The combination increased the yields of the value-added fractions lignin oil and lignin tar and further reduced the coke formation when 5%Pt-1%Ni/HTC was used to 15%. Again, Ni showed better performance over Cu in terms of coke avoidance.To compare the molecular mass distribution of the monomeric and oligomeric product fractions and of the purchased lignin GPC was used. GPC data are shown in Table 3 . A representative GPC chromatogram for the untreated OL and depolymerized lignin fractions after reaction with 5% Pt on HTC support at the standard reaction parameters of 200 °C, 30 min, and 300 rpm is shown in the supporting information (Fig. S5).It can be seen that higher loading with Pt or additional loading with Ni leads to higher molar masses in the lignin tar fraction. This can be explained by the fact that both Pt and Ni suppress coke formation and thus more, but larger lignin tar fragments are formed. The molecular mass distribution is narrower when only 5%Pt is loaded on the support than when a second metal is additionally loaded on the support. Since the molecular mass distribution of the catalyst 5%Pt-1%Ni/HTC is nevertheless very narrow compared to the OL and the gravimetric yields of the value-added fractions could be increased again with the loading of a second metal and above all the coke formation was reduced, this catalyst was considered worthwhile for optimizing the process parameters.Based on the catalyst system found to be suitable (5%Pt-1%Ni/HTC), a DoE was prepared to further minimize coke formation and maximize the value-added fractions (lignin oil and lignin tar). For this purpose, a randomized response surface CCD was performed. Using the program Design Expert©, the optimal reaction parameters (catalyst amount, residence time, and temperature) were determined, and a verification experiment was performed. The layout of the design as well as its limitations, surface plots and evaluation can be seen in Tables S1-S5 and Fig. S1.The analysis of the results gave a reduced quadratic model for the lignin oil and lignin tar fractions as described in Eqs. (5) & (6). (5) lignin oil fraction = + 14.8947 + 0.050682 ∙ T − 10.69925 ∙ m cat + 2.60 ∙ m cat 2 (6) lignin tar fraction = + 93.91967 − 0.20497 ∙ T + 13.11667 ∙ m cat − 3.29673 ∙ m cat 2 In the case of coke formation, the analysis led to a reduced linear model as can be described by Eq. (7). (7) coke formation = − 0.18865 + 4.61728 ∙ m cat These models were used to optimize the process parameters for depolymerization of OL in ethanol-water mixture with the heterogeneous catalyst 5%Pt-1%Ni/HTC. The resulting process parameters, the predicted yields of the response fractions yield of lignin oil, yield of lignin tar, and coke formation as well as the yields of the verification experiment are shown in Table 4 .As can be seen in the verification experiment, the yield of the lignin oil fraction and coke formation is slightly lower than predicted and the yield of the lignin tar is much higher. These results indicate, that the model is still not perfect to describe all effects of the depolymerization conditions. These yields are comparable to the homogeneously catalyzed depolymerization of organosolv lignin using NaOH, where 9.4% lignin oil, 86.2% lignin tar, and 0.5% coke formation were achieved [2]. Since the overall yields of the valuable products were higher and the coke formation lower than predicted and thus the share of value-added products was increased overall, the model is considered sufficient for us. Queneau and Han describe the challenges of biomass recovery in their study. The use of catalysts capable of accommodating the multifunctional nature of biomolecular substrates as well as preventing undesirable over-converted products such as coke formation is a challenge for the scientific community [29]. In the process conditions determined by DoE, the catalyst used in this study can increase the product fractions suitable for further processing while minimizing the undesirable coke fraction and the required catalyst amount. To verify the influence of the support material or the optimized process parameters on the gravimetric yields, additional tests were carried out with the pure support material (0.237 g) and tests without catalyst, in each case at 233 °C and a residence time of 88 min. The yields are shown in Table 5 . The use of 0.237 g of pure support material instead of 5%Pt-1%Ni/HTC shows a small positive influence on the yields of the lignin oil and lignin tar fractions and a slightly positive influence on the prevention of coke formation compared to the experiment where neither support material nor catalyst was used at the process parameters determined by DoE. These results show that the loading of the support material has a decisive influence on the gravimetric yields of the value-adding fractions (ligin oil and lignin tar) and a significant influence on the prevention of coke formation. Wang et al. report that the presence of Ni on the surface of the support material has a positive effect on the hydrogenation and hydrogenolysis of large lignin fragments [30]. Using NaOH catalysis in association with Ru/C, Long et al. were able to reduce coke formation to 14.03% [31]. A possible reaction mechanism is shown in Fig. 5 .A possible explanation for the effective prevention of coke under optimized conditions is the hydrogen transfer of ethanol in the solvent, which is supported by Ni in the catalyst. However, it is necessary to investigate the mechanistic background in more detail considering different coke qualities of different catalysts in a follow-up study.2D-HSQC NMR was used to investigate the S/G/H ratios and the number of β-O-4 bonds of the purchased organosolv lignin and the verification of bond cleavage in the degradation products. The analysis showed an S/G/H ratio of 12/83/5 and 28 β-O-4 bonds for the initial lignin and no remaining β-O-4 bonds in the product fractions. The HSQC spectra are shown in Figs. S2-S4.GPC was employed to determine the molecular weight distribution of the used lignin and the lignin oil and lignin tar product fractions. Results are listed in Table 6 and chromatograms are shown in Fig. S5&S6.These data indicate that the molar mass distributions are very similar for the standard and optimized conditions in both cases, the two fractions have significantly smaller molecular weight distributions than the lignin used. The high yield and especially the significantly reduced coke formation make the optimal conditions more preferable compared to the standard conditions.The determination of the phenolic OH-groups is done by the Folin-Ciocalteu method. Results of the measured OH-groups are listed in Table 7 .The content of phenolic OH groups already increased significantly at the standard conditions in the case of the lignin oil fraction, compared to the initial lignin. Under the optimized conditions, the OH group content slightly decreased again. Looking at the lignin tar fraction, the content of phenolic OH groups was reduced in both cases compared to the initial lignin.Additionally for the purchased lignin and the depolymerization products of the reaction with the optimized process parameters, the determination of the phenolic OH-groups was done by a 31P NMR method. With this method, aliphatic OH groups, phenolic OH groups, and acid groups can be determined. The phenolic OH groups can be further divided into C5 substituted, guaiacyl, and p-hydroxyphenyl groups. Results are listed in Table 8 . The spectra are shown in Fig. S7.Comparing the results of the Folin-Ciocalteu method with those of the 31P NMR method, it can be seen that the 31P NMR method can not only split the phenolic OH groups into different subgroups but can also detect the aliphatic OH groups as well as the carboxylic acid groups. The comparison of both methods indicates that the Folin-Ciocalteu method detects the aliphatic and phenolic OH groups as sum parameters. Whether the OH groups increase or decrease in total due to the degradation can be determined with both methods. The Folin-Ciocalteu method is a simple and cost-effective method to determine the trend of phenolic OH groups. Both methods showed an increase in phenolic OH groups in the lignin oil fraction and a decrease in the lignin tar fraction. Using the 31P NMR method, it was shown that the aliphatic OH groups decreased significantly in the case of the lignin oil fraction. The acid groups decrease significantly in both fractions compared to the starting lignin.Chen et al. report that for Pt and Ni catalysts, respectively, direct deoxygenation and hydrogenation are parallel primary reactions, but the selectivity depends not only on the metal used but also on the solvent used and the process conditions. This was confirmed in our experiments, as the selectivities of the obtained products are significantly different between the experiments at standard conditions (200 °C, 30 min, 3 g catalyst input) and the optimized conditions (233 °C, 88 min, 0.237 g catalyst input) [33]. It can be seen that the product distribution is different for the two reaction conditions. Under the standard conditions, the main product was 2-methoxy-4-vinylphenol, whereas, under the optimized conditions, the main products were 2-Methoxy-4-vinylphenol, 3,5-Dimehtoxy-4-hydroxy acetophenone as well as 2,6-Dimethoxyphenol. Significantly more Phenol was formed too. The fact that carboxylic acids are found in the GC–MS analyses can be explained via the workup process by separating the fractions using aqueous HCl. Table 9 shows the qualitatively and quantitatively determined products of both fractions. Chromatogramm can be seen in Fig. S8. To keep the results of the GC–MS arranged, only monoaromatic compounds that account for more than 1 wt% of the identified compounds were evaluated.Organosolv lignin was depolymerized to lignin oil and lignin tar fractions with the developed catalyst system 5%Pt-1%Ni/HTC and process optimization using DoE, avoiding coke formation. By optimizing the process parameters to 233 °C, 88 min residence time, and a catalyst input of only 0.24 g per 20 g lignin, 18 wt% lignin oil fraction, with significantly more yield of Phenol and 2-Methoxyphenol, and 72 wt% lignin tar fraction were obtained. The coke formation could be reduced to 0.4 wt%. The monomer building blocks could be characterized by GC–MS. Both product fractions, lignin oil and lignin tar showed a significant reduction in molar mass distribution. The obtained product fractions have great potential as bio-based substitutes for Phenol in phenol-formaldehyde resins and as antioxidant additives in the plastics industry. Raphaela Süss: Methodology, Validation, Investigation, Writing – original draft, Writing – review & editing, Visualization. Gottfried Aufischer: Methodology, Investigation. Lukas Zeilerbauer: Investigation. Birgit Kamm: Conceptualization, Project administration, Funding acquisition. Gisa Meissner: Resources. Hendrik Spod: Resources. Christian Paulik: Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was funded by Heraeus Deutschland GmbH and Co.KG and the Austrian Research Funding Association (FFG), (Wood K plus Comet Funding Period 2019-2022). Depolymerisation of organosolv lignin by supported Pt metal catalysts Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.catcom.2022.106503.

The conversion of lignocellulosic biomass into value-added chemicals and biofuels has been attracting the attention of researchers in recent years. Lignin is an abundant, natural polymer and a major component of lignocellulose comprising an aromatic structure with ether linkages, methoxy-, and hydroxyl groups. Therefore, it has great potential as a sustainable source to produce basic chemical products. In this study, precious metal-loaded hydrotalcite (HTC) catalysts for the depolymerization of organosolv lignin (OL) were investigated concerning minimizing coke formation and maximizing the value-added lignin oil fraction and lignin tar fraction. The influences of the catalyst support, the platinum loading as well as the loading with a second metal (Cu or Ni) were examined. The resulting depolymerization fractions (lignin oil, lignin tar, aqueous fraction, and coke) were determined gravimetrically. To compare the molecular mass distribution of the lignin oil and lignin tar fractions as well as the purchased OL, gel permeation chromatography (GPC) was used. The lignin oil fractions were analyzed quantitatively and qualitatively by gas chromatography-mass spectroscopy (GC–MS). Regarding the most suitable catalyst system (5%Pt-1%Ni/HTC), a design of experiment (DoE) was prepared to further minimize coke formation and maximize the value-added fractions (lignin oil and lignin tar). This optimization led to 18 wt% lignin oil fraction, 72 wt% lignin tar fraction, and 0.4 wt% coke formation.

The composite Ni/Al2O3-SiC catalyst was designed and synthesized for hydrogen production from propane partial oxidation. Through the composite of high thermal conductivity SiC and porous solid acid Al2O3, it improved the shortcoming of sintering and inactivation of high-temperature reaction for Ni-based catalysts. The samples were characterized by X-ray diffraction, N2 physical adsorption, scanning electron microscopy, transmission electron microscopy, and thermogravimetric analysis etc. From the results, the SiC doped catalyst exhibited superior catalytic properties for hydrogen production by propane partial oxidation. By doping SiC with high thermal conductivity to Al2O3 support, the catalyst was improved for sintering and carbon deposition resistance and reducing the aggregation of Ni active sites reflected by improving the stability of hydrogen production.

Production of waste is rapidly increasing and World Energy Council expected a waste daily production of more than 6 million tons by the year 2025 [1]. In order to convert the organic fraction of the waste into energy, several techniques can be used like direct combustion, pyrolysis, gasification, transesterification, fermentation, and anaerobic digestion [2]. However, anaerobic digestion is one of the most widely used techniques that result in the production of biogas and biofertilizers [2]. The resulting biogas can be used for the production of heat and electricity. Biogas is mainly composed of carbon dioxide (CO2) and methane (CH4) among other components present in lower amounts (H2, O2, volatile organic compounds, nitrogen, etc.) [3].Biogas has a very low heating value (LHV: 15–30 MJ Kg−1) when compared to natural gas (LHV: 50 MJ Kg−1) and hydrogen (LHV: 120 MJ Kg−1). Therefore, thermocatalytic dry reforming of methane is often employed for the up-gradation of biogas into more valuable syngas. However, this method has not yet been implemented commercially [4]. In this process, a mixture of biogas (i.e., methane and carbon dioxide) is converted in the presence of a suitable catalyst into syngas (a mixture of carbon monoxide and hydrogen). This well-studied reaction in the literature [5–6] continues to pique the interest and attention of scientists due to its importance in valorizing carbon dioxide, a major greenhouse gas. The reforming of biogas was studied on various nickel-based catalytic materials [7–11] and several promoters such as Gd, Sc [7], Ce [8,10,11], Co [9], and Mg [11] were used aiming to improve the stability of these catalysts. It is reported that the addition of ceria and lanthanum in zirconia-supported nickel catalysts favored carbon formation at low reaction temperatures [8]. Huang et al. [11] studied the effect of pressure and revealed that high pressure favored filamentous carbon formation on Ni/MoCeZr/MgAl2O4-MgO catalysts.However, the simultaneous conversion of a biogas mixture into syngas and carbon-derived materials has been rarely reported in the literature [12–15]. Generally, this process is known as catalytic decomposition of biogas, a process that yields a syngas mixture and high-quality bio-nanostructured carbon materials through a series of simultaneous reforming, decomposition, and Boudouard reactions between CH4, CO2, and CO [13]. Notably, the carbon materials produced can be used in several applications, including catalyst support [16] and energy storage materials [17]. The production of carbon materials from biogas has the advantage of producing such high added value materials from a renewable source in comparison to other fossil fuel-based methods.In previous studies, the authors used a fusion method to synthesize: i) Ni/Al2O3 [14], ii) bimetallic Ni-Co/Al2O3 [15], and Ni, Co, and Fe supported on alumina catalysts [13]. The same team also worked on a more sophisticated fluidized bed reactor configuration to continuously produce carbon nanofibers and syngas [12]. Based on these studies and among monometallic compositions, Ni/Al2O3 catalyst seems to be the best choice for this reaction showing maximum carbon production at a moderate temperature [13]. These findings are consistent with our recent work on iron- and nickel-loaded mesoporous alumina catalysts. We demonstrated the superiority of Ni over Fe in catalysing biogas reforming while producing greater amounts of carbonaceous deposits, classified as carbon nanotubes [18].It has been demonstrated that one-pot nickel-based mesoporous alumina catalysts (5 wt% Ni) exhibit higher activity and stability in methane reforming reactions than similar catalysts prepared by conventional impregnation [19–21]. To the best of our knowledge, while conventional non-porous alumina was used for carbon nanofibers and syngas production [12–15], mesoporous alumina materials were rarely used as catalysts for combined biogas reforming and decomposition reactions.This study aims to develop and test ordered mesoporous alumina-supported Ni-based (20, 50 wt%) catalysts for biogas (CH4 and CO2 mixture) reforming and decomposition reaction. The catalysts were prepared using a facile one-pot evaporation-induced self-assembly method. Unlike previous studies [18–21], this work exploits catalysts with high active metal loading to achieve a high yield of the targeted carbon nanomaterials (e.g., CNF) and syngas with the desired H2/CO ratio (∼ 1). Furthermore, detailed thermodynamic simulations for biogas reforming and decomposition under similar reaction conditions were performed to elucidate the experimental results and gain a better understanding of the reaction pathways and mechanisms.The mesoporous Ni incorporated Al2O3 samples were synthesized following a well-established “one-pot” evaporation-induced self-assembly (EISA) method [22–23] with some minor modifications [18–21,24]. In a typical synthesis first 1 g of P123 Pluronic triblock copolymer ((EO)20(PO)70(EO)20, Mn = 5800, Sigma Aldrich) is dissolved in 20 ml of absolute ethanol (CH3CH2OH, Sigma Aldrich) under vigorous stirring at room temperature. After complete dissolution of the structuring agent, 1.6 ml of nitric acid (65.0 wt%, Johnson Matthey S.A) is added under stirring, together with A mmol of aluminum isopropoxide (Al (OPri)3, C9H21AlO3, Sigma Aldrich) and B mmol of nickel nitrate hexahydrate (Ni(NO3)2·6H2O, Sigma Aldrich). All chemicals were used as received, without additional purification. For each synthesis, the total molar composition was always kept constant, equal to 10 mmol (i.e. [A + B] = 10 mmol). For the Ni20%Al2O3 sample, 0.58 g of Ni (NO3)2·6H2O and 1.63 g of C9H21AlO3 were used for its preparation. Regarding the high Ni-loaded sample (Ni50%Al2O3), higher amounts of Ni nitrates and lower amounts of aluminum precursor were used respectively. Typically, 1.45 g of Ni (NO3)2·6H2O and 1.02 g of C9H21AlO3 were co-introduced in the same reactor. Ni-free alumina sample was synthesized using the same experimental protocol yet without the addition of Ni precursor. For each preparation, the final mixture was covered with a polyethylene film, continuously stirred at RT for at least 7 h until complete dissolution then, transferred into a double-layer jacketed beaker supplied by a flow of distilled water regulated at 60 °C to undergo slow evaporation (ethanol, acid) for 48 h straight. The obtained greenish xerogels (the intensity of color augments with Ni content) were slowly calcined (air, thin-bed) at 600 °C for 5 h (heating rate 0.5 °C min−1) to give calcined “one-pot” alumina-based materials.Nitrogen adsorption–desorption isotherms were performed on a Micromeritics Tristar (II) 3020, porosity and surface area analyzer. Prior to measurements, calcined samples were degassed under vacuum for 3 h at 200 °C then cooled down to room temperature before being placed at −196 °C (liquid nitrogen temperature). The Brunauer-Emmett-Teller (BET) surface areas were evaluated from the BET equation for a relative pressure (P/P0) range of 0.05–0.25, while pore size distribution (PSD) was calculated using the Barrett-Joyner-Halenda (BJH) method for the desorption branch of the isotherm.The reducibility of nickel species in calcined materials was estimated by temperature-programmed reduction (H2-TPR) upon carrying out the experiments on an Autochem 2920 (Micromeritics) apparatus. The unit was equipped with a thermal conductivity detector (TCD) for a continuous recording of the overall H2 consumption. Each sample (70 mg) was deposited on quartz wool in a U-shaped quartz reactor and degassed using argon at 150 °C for 60 min, followed by cooling to 25 °C. Then, the furnace was heated to 1000 °C at the rate of 10 °C min−1 under the flow (40 ml min−1) of H2/Ar mixture (10 vol% H2). A cold trap made of ice and salt (NaCl) was used to condense water generated during reduction from the effluent gas before it reached the TCD detector. This assures that the obtained signal (difference in thermal conductivity between reference and analysis gases) is essentially linked with H2 consumption.Structural characteristics of calcined materials were carried out at room temperature by performing powder X-ray diffraction (XRD) at wide angles. Diffractograms were recorded on Rigaku (Mini-flex) diffractometer operating at 40 kV and 40 mA and using a Cu Kα irradiation source (λ = 1.5418 Å). The acquisitions were logged for 2θ values between 10.0 and 90.0° and a step size of 0.02°. A comparison with standard powder XRD files published by the International Center for Diffraction Data (ICDD) helped in the identification of present crystalline phases. The HighScore Plus software was used to examine the data file of the instrument. Crystallite sizes were calculated using the Scherrer equation: D (hkl) = (Kλ/βcosθ), where K = 0.9 is the shape factor for spherical particles, λ is the X-ray wavelength (λ = 1.5405 nm for Cu Kα), β is the full width at half maximum (FWHM) of the diffraction peak and θ is the peak position.The morphological aspects of spent catalysts were examined employing scanning electron microscopy (SEM). Micrographs were registered on the JEOL JSM-6360A microscope accompanied by a Li/Si lens for energy dispersive spectroscopy (EDS) analyses.Thermal gravimetric analysis (TGA) was performed to quantify carbon deposition amounts on spent catalysts. Experiments were recorded on a TGA-1515 SHIMADZU thermal analyzer where; 0.010–0.015 g of the spent catalyst was put into a platinum pan deposited on a thermo-balance. During analysis, the temperature was raised from RT to 900 °C (heating rate: 20 °C min−1) in flowing air (50 ml min−1) and the change in mass was constantly monitored and computed.The biogas reforming and decomposition reaction was carried out at atmospheric pressure in a tubular stainless-steel reactor (internal diameter: 9 mm and length of 300 mm). For each experiment, a 100 mg sample (non-diluted powder), sandwiched between two quartz wool beds, was used. The respective catalysts were first in situ reduced at 600 °C for 1 h in pure H2 (30 ml min−1) to ensure complete reduction of NiO to catalytically active Ni0 nanoparticles. Then, the flow was switched to the reactant mixture (CH4:CO2:N2 = 1:1:0.33), and the total gas hourly space velocity (GHSV) was 42 Lgcat -1hr-1. Stability measurements were carried out at 700 °C for 5 h on time-on-stream where the temperature was controlled using a K-type thermocouple placed in the middle of the catalyst bed. The effluent gas was quantified by online gas chromatography using a Micro-GC equipped with a thermal conductivity detector and coupled to two columns placed in parallel for the detection of CH4, H2, and CO (type: Molecular Sieve 5A column) and CO2 (Porapak Q column). Conversions of CO2 (XCO2), CH4 (XCH4), and H2:CO molar ratio were calculated based on Eqs. (1)–(3): (1) X CO 2 % = CO 2 in - CO 2 out CO 2 in × 100 (2) X CH 4 % = CH 4 in - CH 4 out CH 4 in × 100 (3) H 2 : C O ( m o l a r ) = mol of H 2 p r o d u c e d mol of CO produced The HSC (versions 7.1 and 10) Chemistry software (where H, S, and C stand for the enthalpy, entropy, and heat capacity, respectively) was used to generate theoretical models imitating experimental conditions. Maximum allowed thermodynamic values, for different inlet feed compositions and fixed pressure of 1 atm, are plotted as a function of temperature (range: 200–1000 °C). It is worth noting that simulations allowed for both gaseous and solid species. CH4, CO2, H2, CO, and N2 were included as gaseous components, while C(s) (whenever added) was introduced separately in the solid phase. Two of the simulations were performed with an inlet CH4:CO2:N2 molar ratio of 1:1:0.33 with and without the consideration of C(s) as a potential product of the reaction. The third simulation was conducted for pure methane decomposition (CH4:N2 ratio of 1:0.33, no CO2 was included in the feed) and served for comparison purposes on the effect of CO2 on the amount of C(s) and product distribution. The general principle of each simulation is based on a procedure relying on the minimization of the total Gibbs free energy of the reacting system [25–26]. In this case, the system becomes thermodynamically favorable when its total Gibbs free energy value is at a minimum and its differential is essentially zero under specified conditions of temperature, pressure, and feed composition. The equilibrium state is determined without any specification of main and side reactions occurring in the system. Additional details on the overall principle and involved mathematical models can be found in our recent thermodynamic studies on methane reforming [27] and iron oxide reduction in the hydrogen atmosphere [28].All prepared materials (Ni-free and Ni-containing) exhibit type IV N2-sorption isotherms (Fig. 1 ) with H1-shaped hysteresis loops and rises in adsorbed nitrogen amounts within P/P0 values between 0.4 and 0.6. Such observations are classified as typical features of ordered mesoporous materials [29]. Corresponding textural properties are listed in Table 1 and surface area values are in the range of 177–206 m2g−1, as high as in previous reports on mesoporous alumina synthesized under comparable conditions [19–24]. The calcined, Ni-free, Al2O3 sample (curve a, Fig. 1) displays a very steep hysteresis loop indicative of the presence of uniformly arranged cylindrical mesopores [22–23,29], with a mean average diameter of 7.1 nm (Table 1). The incorporation of Ni in the course of preparation induces textural changes manifested by a gradual shift of the hysteresis loop to higher relative pressure values accompanied by a decrease in its steepness. The phenomenon is more accentuated with increasing Ni content (curve c, Fig. 1). Such a shift is the result of pore widening, revealing larger yet slightly ordered mesopores [22–23,29–30]. This can be also illustrated by referring to the larger range of pore sizes, deduced after deriving BJH curves from the desorption branch of N2-sorption isotherms (inset in Fig. 1, Table 1). Such textural changes could not have been derived from thermal treatments since all materials were subjected to the same conditions of evaporation and calcination. In fact, this increase in pore diameter could be attributed to the presence of nitrates-bearing water molecules, specifically, hexahydrates initially present within the composition of the salt precursor. During stirring, nitrates hydrolyze and produce H+ cations (via dissociation of water molecules) acidifying the ethanol solution and creating, therefore, a different atmosphere than that expected in Ni-free solutions. As a result, the hydrogen potential (pH) of the resulting solution changes. Importantly, such pH changes have been reported to be major contributors to influencing the balance at the organic/inorganic interface, causing a disruption in the assembly process. The assembly initiates during stirring at RT and extends to evaporation at 60 °C [31]. Such alterations of acidity levels together with possible access restriction to some pores by occluded species could also contribute to the slight loss of surface area seen in Ni20%Al2O3 compared to Ni-free Al2O3 (Table 1). A main noticeable observation is the surface area of Ni50%Al2O3 being close to that of Ni-free Al2O3 (Table 1), in spite of its highest disordered structural arrangement. The Ni50%Al2O3 contains an excess Ni in comparison to the Ni/Al molar ratio characteristic for nickel aluminate (NiAl2O4) [22]. Therefore, some Ni species are mainly present in the form of free Ni, as confirmed by TPR experiments (Section 3.2), outside of the alumina matrix. The external deposition of (some) Ni species creates structural surface defects accompanied by an increase in the amount of adsorbed N2 during monolayer coverage. Besides, a notable evolution in these Ni-containing samples is the increase of pore volume with Ni enrichment, as previously reported on analogous “one-pot” synthesized Ni-alumina materials in our study [20] as well in another for materials calcined up to 400 °C [22], but not yet clarified. Regardless of these variances between samples, their isotherms classify them as mesoporous materials with high thermal resistance, as they have been treated up to 600 °C, making them suitable for the high-temperature dry reforming of methane reaction.The reducibility of calcined samples was studied by following the H2-consumption behavior of nickel species, under hydrogen flow, from RT up to 800 °C. The H2-TPR profiles of freshly calcined samples are displayed in Fig. 2 . Table 1 shows the overall experimental H2-uptakes (200–800 °C) estimated from respective reduction profiles, as well as the relative contribution (in terms of consumed H2) of each peak. The absence of any H2-consumption peak on the profile of the Ni-free mesoporous Al2O3 (pattern a, Fig. 2) is in complete accordance with the fact that this sample contains no reducible species. A major hydrogen reduction peak is seen for Ni20%Al2O3 (pattern b, Fig. 2), with maxima localized at around 570 °C. Such type of profile already noticed in similarly prepared “one-pot” Ni-alumina materials [19–21,24], is typical for demonstrating the presence of oxidized Ni species strongly interacting with the support matrix, possibly in the form of mixed spinel phases. The reduction of such phases yields small, well-dispersed, Ni0 nanoparticles that are active and stable on time-on-stream [19–21,24]. As for Ni50%Al2O3, the larger area of reduction pattern is attributed to the bigger amount of Ni deposited in its course of preparation. As expected, integration of the area under the curve revealed an overall H2 amount of 4432 µmol g−1 for Ni20%Al2O3 and 13459 µmol g−1 for Ni50%Al2O3 (Table 1). The overall experimental H2 uptake at 800 °C is consistent with the complete reduction of the targeted amount of Ni species (theoretical H2 amount is 4400 μmol g−1 for 20 wt% Ni and 11000 μmol g−1 for 50 wt% Ni) revealing a good accuracy between expected amounts and those recovered after evaporation and calcination steps. Regarding the TPR profile of Ni50%Al2O3 (pattern c, Fig. 2), different types of Ni species are co-present within this sample. Those undergoing reduction at high-temperatures (peak centered at circa 640 °C) are similar in nature to nickel species present in Ni20%Al2O3 and are characterized by the establishment of strong metal-support interactions (MSIs). The other type of species reduces at lower temperatures (reduction peak centered at circa 340 °C) and is best described as free and easily reducible NiO species with weaker MSI [19,32–33]. Based on quantification results (Table 1), around 20% of species are classified as free NiO whereas the majority is constituted of Ni strongly interacting with the support, requiring a higher reduction temperature. Such trends in the TPR profile of the highly loaded Ni sample (Ni50%Al2O3) could result from Ni sintering, in the course of calcination, leading to the deposition of nanoparticles outside the mesoporous framework. Another possible explanation is that these nickel atoms did not chemically interact with aluminum atoms during the early stages of preparation due to the lower availability of aluminum atoms (high Ni/Al ratio) in Ni50%Al2O3 compared to Ni20%Al2O3. It should be recalled that the total molar composition of aluminum and nickel remains fixed across all syntheses. Given the slight temperature differences observed for the main reduction peaks between the two samples (patterns b and c, Fig. 2), it is worth mentioning that TPR experiments are conducted under dynamic conditions, far from equilibrium, which potentially affects the overall reduction signature according to the size and the localization of reducible nickel species.Catalytic tests were performed over in situ reduced Ni20%Al2O3 and Ni50%Al2O3 catalysts at a gas hourly space velocity of 42 L gcat -1 hr-1 and in the presence of excess reactants (i.e. CH4 and CO2) as compared to inert gaseous diluent. Such conditions can be addressed as drastic compared to previously adopted experimental protocols applied on similarly prepared materials for running combined steam and dry reforming, biogas reforming, or dry reforming of methane reactions [18–21]. Corresponding experimental details on reaction medium and conditions previously selected to run reactions as compared to those chosen for this study are listed in Table 2 . As shown based on tabulated data, the molar CH4 (or CO2): diluent ratio used in this study is much higher than in any other study (i.e. 3.03 compared to 0.128 and even lower). In fact, diluents are sometimes blended with the reactants mixture for safety reasons in the event of a leak. However, adding a diluent to the feed affects the conversion of reactants by shifting the equilibrium to lower temperatures [27]. Operating “one-pot” Ni-alumina catalysts under slightly diluted conditions adds a realistic outlook to their catalytic performances. Indeed, raw feed streams, for instance, biogas from various biomass resources, to the reformer unit are usually slightly diluted where; the amount of the inert component does not surpass 25 vol% of the mixture [34]. Regarding the GHSV value of 42 L gcat -1 hr-1, it falls between previously used values (Table 2) and depends on the amount of catalyst loaded in the reactor. Lowering that content decreases the residence time and therefore affects the extent of conversion [20–21]. In this study, an intermediate contact time is allowed inside the reformer, and the effect of catalyzing an equimolar CH4:CO2 inlet feed stream, much richer in chemically active species is evaluated. As for the choice of the temperature, its value is in line with those adopted for running dry reforming of methane, DRM (Table 2) since it allows the attainment of high methane and carbon dioxide conversion levels, as will be shown based on thermodynamic discussions (Section 3.3.3).Catalytic performances in terms of CH4 and CO2 conversion and H2:CO molar ratios as a function of time-on-stream, at a fixed temperature of 700 °C, are displayed in Fig. 3 . Under our reaction conditions and based on thermodynamic calculations accounting for C(s) deposition (dashed lines, Fig. 3), a higher CH4 conversion (by almost 24%) should be anticipated priori compared to that of CO2. Regarding selectivity outcome, expressed in terms of H2:CO ratio, a value of 1.45 is expected signifying excessive H2 production compared to CO for an equimolar inlet feed of CH4 to CO2. Catalytic data of Ni20%Al2O3 (curve a Fig. 3) show stable performances for 300 min on stream with CO2 conversion being constantly higher (by about 10%) than that of methane. A behavior that is opposite to thermodynamic data estimated for C(s)-assisted methane reforming (dashed line, Fig. 3A), being instead in line with those projected for a C(s)-free DRM operation (straight line, Fig. 3A). Consequently, the H2:CO ratio is lower than 1.45 (curve a Fig. 3C), being very close to 0.88, the expected thermodynamic molar value from a C(s)-free DRM reaction (straight line, Fig. 3C). Surprisingly, the high-loaded Ni-sample displayed an un-classical behavior in the course of DRM (curves b, Fig. 3). Initially, conversion values were at the same level then, underwent opposite trends where; methane conversion continued increasing, that of carbon dioxide constantly decreasing and the variation of the molar H2:CO ratio was most affected by methane conversion. It is worth noting that Ni50%Al2O3 operated for 120 min owing to excessive rise in pressure resulting from a blockage due to heavy C(s) deposition. In-depth analysis of such carbonaceous deposits are discussed in upcoming sections. In fact, the pressure reached a value of 8.5 bar after 120 min of reaction. The pressure was initially controlled at atmospheric value then rises to exceed this value as the reaction proceeded because of severe accumulation of solid carbon deposits on the catalyst surface as well as in the reactor tubes. Within the reaction time of 120 min, methane conversion was always higher than that of carbon dioxide attesting to a mechanism involving C(s) development (dashed lines, Fig. 3A). The continuous increment in CH4 conversion accompanied by the incessant reduction in CO2 conversion indicate a preferential reactivity of Ni0 sites toward methane as compared to carbon dioxide. Thus, methane is being preferentially activated into its elementary H2 and C(s) products justifying the rising tendency noted on the molar H2:CO profile (curve b, Fig. 3C).For a better understanding of reactivity data, thermodynamic plots reflecting non-C(s)-developing and C(s)-developing biogas reforming and decomposition operations along with those attributed to pure methane decomposition are presented in Fig. 4 . Based on these plots, and for a temperature value of 700 °C, gaseous CO and H2 products are to be expected for both cases involving CO2 as a reactant (Fig. 4A, B). Their content differs depending on whether C(s) are allowed to be produced in the medium. For example, in simulations involving C(s) generation, the expected amount of CO is around 1 Kmol whereas it is close to 1.5 Kmol in absence of C(s) as a product. This infers that the CO is consumed in C(s)-formation reactions involving CO hydrogenation (CO + H2 ↔ H2O + C(s)) and CO disproportionation (known as Boudouard reaction, 2CO ↔ CO2 + C(s)), both being thermodynamically favored within low to intermediate temperature ranges, accounting in its lower amounts at 700 °C (Fig. 4B). As for H2, close amounts (range: 1.37–1.53 Kmol per 1 mol of CH4) are expected in both cases (Fig. 4A, B). Despite similar contents, it is worth noting that methane conversion is expected to be much higher in the scenario involving C(s) formation than that predicted without C(s) generation (Table 2). This implies that methane will participate in a series of reactions accounting for its consumption through its decomposition, dry reforming as well as reforming with steam. For the latter reaction, it is indeed favored owing to the richness in H2O as compared to a deprived environment for the simulation designed without C(s) (Fig. 4B). Methane is then believed to undergo consumption, solely, via DRM under C(s)-free conditions (Fig. 4A). The similarity in H2 amounts, at 700 °C, resulting from different methane conversion values (CH4 being a main H2 source) attests to the fact that the majority of methane is consumed at lower temperatures before reaching 700 °C, under C(s)-developing conditions, through steam reforming (CH4 + H2O ↔ 3H2 + CO) being characterized by a lower endothermic nature than dry reforming [27].Coming back to experimental data, it can be seen that Ni20%Al2O3 is catalyzing dry reforming of methane in absence of severe C(s) deposition/accumulation (Section 3.3.2), as will be verified based on XRD, SEM, and TGA data (upcoming sections). The performance of this catalyst could have been anticipated as previous “one-pot” synthesized samples displayed very stable reactivity levels in CSDRM, BR, and DRM yet, under more diluted inlet feed compositions (Table 2). Thus, this sample reflects high adaptability to drastic conditions making it efficient for catalyzing realistic biogas compositions. Regarding Ni50%Al2O3, its performance does not really imitate dry reforming in either scenario since for both C(s)-developing and non-C(s)-developing DRM, methane and carbon dioxide should remain stable, under isothermal conditions (Fig. 4A,B). This sample is rather catalyzing, preferentially, methane over carbon dioxide as if it is promoting methane decomposition (Fig. 4C). In fact, it has been recently reported, based on transient CH4 and CO decomposition experiments, that methane decomposition appears to be the main pathway for solid carbon formation (and its subsequent deposition) under typical DRM conditions (atmospheric pressure and temperature close to 750 °C) particularly in presence of supported Ni-based catalysts [35]. Indeed, as time proceeds, more methane is being consumed yielding H2 and C(s), as evidenced by the characterization results of spent materials. A much smaller fraction of CO2 still undergoes consumption yet, its amount decreases on stream.The performance of Ni20%Al2O3 appears to be promising when compared to literature data obtained by other research teams over Al2O3-based catalysts, with comparable properties to ours, even for feeds more enriched in diluent gas and reformed at lower temperatures (Table 2). For a 16 wt%, “one-pot” synthesized mesoporous Ni-Al2O3 sample, lower reactivity levels along with deactivating performances were noted when the catalyst was operated at 600 °C for a CH4:CO2:Ar inlet molar ratio of 1:1:1 and for a smaller GHSV value (ref. 36, Table 2). Catalytic data similar to those noted over Ni20%Al2O3 were found over a Ni5%Al2O3 catalyst operated at the same temperature and under a richer CH4-CO2 reactional feed stream (ref. 37, Table 2). The similarity in outcomes possibly originates from the preparation method being based on the EISA approach. The different Ni content seemed to slightly affect reactivity levels, probably due to differences in situ reduction treatments for the samples prior to catalytic testing. In our case, the reduction was conducted at lower temperatures and under pure H2 stream whereas, in the reported work, reduction took place at 750 °C under a diluted H2 stream (5 vol% H2 in N2) [37]. Depending on the richness of hydrogen in the medium, its atomic recombination is believed to be increased in presence of a diluent improving, therefore, H2-spillover and affecting positively the reactivity of nickel [40]. When “one-pot” synthesized Ni-Al2O3 samples were allowed to catalyze undiluted feed streams (refs. 38 and 39, Table 2), a common trend is witnessed. Catalytic deactivation is noted owing to sintering and subsequently to heavy carbon accumulation on active sites since the early stages of reaction. A possible perspective for stable performances could involve an enrichment in Ni content (for wider availability of active sites), upon adopting the EISA “one-pot” methodology, since catalytically active Ni0 nanoparticles were shown to remain encapsulated within the mesoporous matrix, protected from sintering and from the deposition of graphitic, known as deactivating, carbonaceous species [18–21,24].In view of Ni50%Al2O3, its performance is quite promising when compared to previously reported data on catalysts designed for methane decomposition (MD). High intrinsic activity values are typically expected over Ni-based catalysts for MD. For instance, a 50–60 wt% Ni-loaded Al2O3 supported catalyst converts methane at a rate of roughly 64% [41]. However, when such samples are tested isothermally at 675 °C, they underwent drastic deactivation within just 30 min on stream. When Ni-doped (12.5 wt%) hydrotalcite-like Mg-Al (Mg2+/Al3+:0.24) solid-solution materials were tested for MD, a methane conversion of 47% was obtained at 650 °C for 30 min on time-on-stream. However, other hydrotalcite-based catalysts, synthesized using various Mg2+/Al3+ ratios, deactivated dramatically after a few minutes of exposure to reactional medium [42]. Similar behavior has been also reported over Ni-based catalysts supported on mesoporous alumina [43]. Improvement tactics have been already developed, especially for Fe-based materials, and these include (i) the addition of secondary, transition metal, elements (i.e. Ni, Mo, and Co) to tune the characteristics of the material [43–45] and, (ii) the incorporation of trace amounts of noble metals for reactivity boosting [46]. Although these solutions improve catalytic reactivity and stability to some extent, they are time-consuming and involve multi-step complicated synthesis procedures, increasing the overall complexity of the process. Moreover, some of these steps are also expensive especially when noble metals are involved, which accounts for additional expenses. The “one-pot” methodology reported in this study offered the direct development of a monometallic Ni-based sample that is inherently porous and capable of catalyzing (preferentially) methane reforming and decomposition while undergoing in situ activation, yielding higher and stable conversion on time-on-stream. Most importantly, developed carbonaceous deposits belong to the family of non-deactivating species, mainly carbon nanotubes, having a wide range of industrial applications and high economic interests (Section 3.4).The spent catalysts were characterized by several techniques to quantify and identify the nature of carbonaceous deposits. XRD was performed over fresh and spent materials for a direct comparison of the evolution of Ni species and the crystallographic state of carbon deposits. These patterns are presented in Fig. 5 . Depending on the amount of added metal, XRD diffractograms of calcined catalysts presented different signatures. For Ni20%Al2O3 (pattern a, Fig. 5), the XRD signal was mostly amorphous indicative of high dispersion of species in the mesoporous alumina skeleton. Even if the presence of NiO was not detectable in the calcined material, it was confirmed by TPR (discussed previously in Section 3.2). For the Ni-enriched sample, its XRD signature (pattern b, Fig. 5) displayed apparent diffraction peaks characteristic of cubic NiO. Applying Scherrer’s equation at 2θ of 43.17° (attributable to the 200-diffraction plane) revealed an average NiO particle size of 31.6 nm. For both calcined materials, no diffraction peaks were visible for crystalline alumina attesting to its high thermal stability against dihydroxylation and/or dehydration after calcination at 600 °C. As a result of in situ reductions followed successively by catalysis, XRD patterns of Ni-containing catalysts showed some structural modifications. Diffraction peaks corresponding to the phase transition of the alumina skeleton from amorphous (patterns a and b, Fig. 5) to (partially) crystalline γ-Al2O3 became visible (patterns a’ and b’, Fig. 5). Despite their appearance on diffraction patterns, their crystallographic domains are very small (at least compared to other diffraction peaks on the same patterns) attesting the high structural resistance of the ordered alumina framework. Moreover, the peaks attributable to (cubically arranged) metallic nickel were identifiable over both samples. The determination of the exact average Ni0 particle size, based on XRD data, cannot be possible because of the overlapping of Ni0 diffraction peaks with those corresponding to crystalline γ-Al2O3 (patterns a’ and b’, Fig. 5). Nevertheless, the stable catalytic performance of Ni20%Al2O3 toward dry reforming of methane and the un-classical catalytic behavior (described by a progressive rise in methane conversion) of Ni50%Al2O3 toward methane decomposition attest to the continuous presence of accessible and catalytically active Ni0 centers that remained dispersed and resistant to sintering despite the high temperature and the amount of loaded metal. The sole presence of metallic nickel diffraction peaks over spent catalysts indicates that no reoxidation into (catalytically inactive) NiO took place during the reaction. Additional diffraction peaks typical of crystalline C(s) were also present on the diffractograms with varying degrees of peak intensity (patterns a’ and b’, Fig. 5). With correlation to catalytic data (Section 3.3), the amount of deposited carbon is in accordance with reactivity trends. In fact, higher amounts of carbonaceous deposits were found over Ni50%Al2O3 in line with its preferential reactivity for catalyzing methane decomposition yielding thus 1 mol of C(s) per 1 mol of methane.Representative SEM images of fresh and spent Ni20%Al2O3 and Ni50%Al2O3 catalysts are shown in Fig. 6 . Based on Fig. 6A there is no evidence of externally deposited nickel oxide nanoparticles, over calcined Ni20%Al2O3, because of the absence of brilliant colors which normally result from the diffusion of metals having higher electronic densities than aluminum oxide. However, some light spots are visible on the external surface of alumina grains for calcined Ni50%Al2O3 (Fig. 6B). Their presence indicates that (some) nickel particles have been deposited outside of the mesoporous framework. It results in the formation of surface defects (as deduced from N2-sorption data, Section 3.1) and the presence of an easily reducible population of Ni particles (as shown based on TPR profiles, Section 3.2). Such nanoparticles are reported to be more prone to coking, during catalysis, than those stabilized inside the alumina matrix since carbon formation is hindered over internally incorporated species owing to steric constraints imposed by the support structure [18–21,24,47]. After catalysis, crystalline carbonaceous deposits detected by XRD were also visualized by SEM. Over Ni20%Al2O3 spent catalyst, some (small) domains of sponge-like and short-carbon filaments were shown on the external surface of alumina grains (Fig. 6A’). Compared to a typical SEM image of spent Ni50%Al2O3 (Fig. 6B’), alumina grains in this sample were almost undetectable owing to complete coverage by solid carbon deposits. Higher-resolution SEM images, taken for a clearer visualization of such carbonaceous materials overspent Ni50%Al2O3 (Fig. 7 ), revealed that the catalyst consisted mostly of hollow and randomly oriented nanotubes of several branches with apparently (i) an uneven surface morphology (Fig. 7A) and (ii) recognizable Y-junctions between nanotubes (Fig. 7B). Additionally, these images showed that developed nanotubes had an open-ended geometry where; Ni0 nanoparticles were found hanging on the tips of nanotubes (as indicated by the dashed circles, Fig. 7). The presence of such Ni0 particles at the end of nanotubes is indicative of their formation through the well-established tip-growth mechanism [48]. Therefore, a part of the Ni surface remained accessible for catalysis, where nanoparticles catalyzed preferentially methane decomposition.The amount of C(s) deposits and their reactivity in the oxidative atmosphere (flowing air) have been studied by TGA and corresponding profiles are displayed in Fig. 8 . Based on their oxidation temperature(s), carbon species can be classified as weakly stable amorphous Cα (sp2 C-atoms, graphene-like species, oxidation peak temperature between 300 and 450 °C), moderately stable Cβ (C-nanotubes, oxidation peak temperature between 450 and 600 °C), and highly stable Cγ (sp3 C-atoms, oxidation peak temperature exceeding 600 °C) [49–50]. The first significant information provided by TGA is the total C(s) content over spent catalysts. Notably, the amount of carbon produced over Ni20%Al2O3 spent catalyst is much lower (i.e., ∼5 wt%) than that formed over Ni50%Al2O3 spent catalyst (i.e., around 98 wt%). These results are in accordance with the above-discussed XRD and SEM data, highlighting that each catalyst followed a different mechanistic pathway under a similar reactional feed stream. Furthermore, this finding is consistent with recent data on Ni-supported catalysts tested for DRM, which show a proportional relationship between the size of Ni nanoparticles and amounts of accumulated carbon [35]. Authors showcased that depending on tailored catalyst composition preferential formation of C(s) nanotubes could be favored. It is worth mentioning that for spent Ni20%Al2O3, a small rise in its TGA profile is noted (pattern a, Fig. 8) and this could be attributed to a re-oxidation of (small) Ni0 into NiO under flowing air. The second information deduced from TGA profiles is the type of carbonaceous deposits. As evidenced by the carbon oxidation temperature (ca. 550 °C) and corresponding weight loss, over Ni50%Al2O3 catalyst a significant amount of carbon nanotubes is produced (pattern b, Fig. 8). This confirms the continuous reactivity of this sample on time-on-stream as the produced nanotubes did not encapsulate active centers rather, they detached the Ni particles that are weakly bonded to the support and grew whilst these particles catalyzed methane into hydrogen and nanotubes. Such a preferential activation of methane could be described based on a recent study by Jiang et al. [51] over bimetallic Ni-Co/CeO2 catalysts in DRM. The authors demonstrated, using activation energy calculations, that the support, irrespective of the transition metal site, is crucial in activating CO2 during DRM. As for methane, it can be activated on metallic sites and such activation remains persistent as long as interfacial metallic sites remain available and accessible for reaction. Likewise, Liang et al. [52] proposed a mechanistic scheme showing the importance of support in CO2 activation and the role it plays in carbon gasification. Authors highlighted the fact that methane undergoes activation on metallic while CO2 activates on the support (CO2 → CO + O), owing to the richness in oxygen vacancies, yielding CO and O. Then, the as-produced atomic oxygen interferes in the in-situ oxidation of carbon, resulting from CH4 decomposition, through the reaction C(s) + O → CO. In view of the results reported in our study, we can tentatively attribute the behavior of Ni50%Al2O3 to the coverage of the alumina surface by carbon nanotubes, preventing the direct contact of CO2 with the slightly basic sites of Al2O3, resulting in decay in CO2 conversion while that of methane remained intact and rising. The progressive increase in methane conversion potentially results from the enhanced electronic properties of Ni0 as they are present on the tips of nanotubes, characterized by high electronic conductivities [53]. In fact, the high thermal and electronic conductivities of nanotubes along with their high mechanical strength are making them of great interest for application in multiple fields including electronics, catalysis, and biosensors [53–55].One-pot mesoporous alumina containing 20 and 50 wt% nickel-based catalysts were successfully prepared using the EISA method. The physicochemical characterizations showed a partial disorder in the porosity and the deposition of some nickel outside the mesoporous alumina grains when the Ni loading was increased to 50 wt%. However, a high Ni metal loading (50 wt%) in the catalyst was required for the simultaneous production of syngas and the desired carbon nanofibers. These latter materials, known as bio-nanostructured carbon, were successfully produced without significant deactivation of the catalysts in a continuous and highly concentrated reactants stream (CH4, CO2, and < 15% Ar diluent). This work demonstrated the efficacy of the one-pot EISA method for developing highly active and selective catalysts for the simultaneous production of carbon nanofibers and syngas from biogas. These findings establish a foundation for a more rational synthesis of active and selective catalysts for biogas valorisation applications. In the future, we aim to investigate other transition metals and operate the catalytic reaction under high pressure and industrially relevant conditions. Nissrine El Hassan: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft. Karam Jabbour: Methodology, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization. Anis H. Fakeeha: Validation, Resources, Writing – review & editing. Yara Nasr: Formal analysis, Investigation. Muhammad A. Naeem: Data curation, Writing – review & editing. Salwa Bader Alreshaidan: Formal analysis, Investigation. Ahmad Al Fatesh: Formal analysis, Investigation, Resources.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to extend their sincere appreciation to Researchers Supporting Project RSP-2021/368, King Saud University, Riyadh, Saudi Arabia.

Biogas, a renewable energy source, is primarily composed of CH4 and CO2. It is a promising alternative to fossil fuels and can be used directly for electricity production as well as heat generation via combustion. Concerns about climate change and a greater emphasis on renewable energy sources have recently increased interest in biogas utilization. In this context, biogas reforming and decomposition (BRD) into synthesis gas and carbon nanofibers (CNFs) is viewed as a new and attractive way of efficiently valorising biogas. In this study, Ni-loaded (i.e., 20, 50 wt%) mesoporous alumina materials were prepared using one-pot evaporation-induced self-assembly method for BRD. Synthesized materials were characterized by various techniques: N2-physisorption, X-ray diffraction, temperature-programmed reduction, scanning electron microscopy, and thermal gravimetric analysis. Results showed that textural and structural properties of synthesised materials differed with Ni loading. High Ni-loaded catalyst displayed higher surface area, pore volume, pore size distribution, and average particle size which is the result of deposition of Ni species outside alumina grains creating thus, surface defects. BRD results were greatly influenced by Ni content with Ni50%Al2O3 reflecting catalytic behaviour similar to those expected for pure methane decomposition. Most importantly, this catalyst was also capable of generating, selectively, interesting carbon nanofibers.

Due to the advantageous property and long lifetime, plastics are important materials in our society. The yearly plastics consumption is around 300 million tonnes in worldwide and about 60 million tons in the EU [1]. Most of the waste polymers are landfilled (27%) and valorised with energy recovery (41%) in Europe [1–3]. However, there is a great difference among the landfill rate of different countries; some of them has less than 5%, while others have more than 70%. It is a positive tendency, that the amount of the collected, recycled and energy recovered waste plastic are significantly increased; besides the amount of the landfilled plastic waste is decreased in the last ten years [1]. Currently, on average, 30% of the plastic wastes are recycled, inside and outside of the EU by mechanically or chemically. Pyrolysis (so called chemical recycling) may be an efficient option to resume the energy from waste polymers and to obtain valuable hydrocarbons [3,4]. Gas, pyrolysis oil and coke should be the products of pyrolysis process. The arising oil can be used as fuel oil in furnaces or after further upgrading as fuel in vehicles. Whereas different types of the plastics have variance in their chemical structure and physical nature, the yield and composition of the pyrolysis products are significantly different [5–7].It is also well known, that high quality, waste derived hydrocarbons could be obtained by the pyrolysis of polyolefin (high density polyethylene (HDPE), low density polyethylene (LDPE) and polypropylene (PP)) and polystyrene (PS). On the other hand, contaminations and the presence of other polymers, such as polyvinyl chloride (PVC), polyamide (PA), paper, biomass or even additives for plastics, containing different elements except for carbon and hydrogen, can significantly deteriorate the advance property of the pyrolysis products. The problem can be eliminating by the efficient type selective collection of waste materials, but it needs relatively high installing and operating costs. Regarding the undesired constituents, chlorine is one of the key components, because significant corrosion can be caused by chlorinated components during storage, transportation, combustion, etc. Another problem is that the chlorinated components can poison the catalysts not only during the pyrolysis, but also in further application of pyrolysis oil [8–13]. PVC has a key role in the plastics industry and the occurrence of PVC in streams of municipal plastic wastes is very common. Approximately 10% of the plastic is PVC, used for building and construction, packaging, households, automotive, electric and electronic, etc. sector [1].During the thermal decomposition of PVC toxic and corrosive compounds are generated, especially vast amount of hydrogen chloride in gaseous product and chlorinated organic compounds in pyrolysis oil [8,10,11]. On the other hand, the reduction of chlorinated compounds in pyrolysis product obtained from PVC containing raw materials is an unsolved problem. For this intention the pyrolysis kinetic of both virgin and waste PVC, each alone or mixed with other plastics, have been studied by many authors [10,11,14]. The presence of the other plastics and the PVC concentration can significantly affect the dechlorination reactions [14].The in-situ reduction in the concentration of chlorinated compounds can be divided into different groups: stepwise pyrolysis, catalytic pyrolysis, pyrolysis with adsorbents added to the sample, or even hydrothermal carbonisation are used [3,4,15–23]. In stepwise pyrolysis, a prior low temperature step (up to 350 °C) is used for the sake of eliminating the chlorine from the original sample in the form of HCl, which leaves the reactor system as gas; then the temperature is elevated and the sample is pyrolyzed in a conventional run between 400 and 800 °C [4,12,17–19]. In general, the co-pyrolysis of PVC with other plastics and biomass are investigated by stepwise process. E.g. H. Kuramochi et al. concluded that the HCl emission could significantly reduce due to the presence of wood, because of the hemicellulose as a strong Cl absorbent [18]. Stepwise pyrolysis often applying different or separated reactor vessels with different temperatures, because more than 95% of the dehydrochlorination reactions take place in the first unit at low temperature (300–350 °C), while the chlorine content of the products from the following reactors were less (<0.025%) [19]. By biomass-PVC co-pyrolysis also the reaction kinetic parameter should be modified [20]. Regarding hydrothermal carbonisation, the high temperature favoured to the aromatic hydrocarbons [21]. Nowadays, more and more researcher investigates the co-pyrolysis of plastic waste and different types of biomass. In this case, the initial characterization of feedstock has a great significance to know the suitability of raw material to carry out further experimental sets, because all the biomass could not be suitable, economic, qualitative and high yield bio-oil producer. Thermogravimetric analysis is appropriate to characterize the feedstock. B.B. Uzun and E. Yaman studied the pyrolysis kinetics of walnut shell and waste polyolefins using thermogravimetric analysis. The results provide valuable information on pyrolysis mechanism and based on them, walnut shell have considerable potential for pyrolysis. However, the heating rate has a significant effect on the pyrolysis process [22]. Akancha et al. investigated the co-pyrolysis of waste polypropylene and rice bran wax to produce biofuel. The physical properties of the pyrolysis oil were fairly comparable to Diesel and gasoline [23].In catalytic pyrolysis, some catalysts with metal content have been studied in order to demonstrate their positive properties. In general, the used catalysts are metals on inorganic supporter, such as synthetic zeolites, therefore, they have double role, as pyrolysis catalysts and inhibitors for HCl formation [3,18–21,24–26]. Not only single catalysts, but also metal loaded catalyst and catalyst mixtures/composites are used. E.g. fluid catalytic cracking, hydrocracking catalysts or ZSM-5 and iron oxide composite catalyst can efficiency used for reducing the chlorine content of products of vacuum gas oil and PVC pyrolysis [16].Regarding the adsorbents, especially the HCl emission can be decreased, however, the reduction in chlorinated compounds in pyrolysis oils would be also advantageous. Numerous materials as biomass constituents, petrochemical residues and alkaline adsorbents (NaHCO3, CaO, CaCO3, Na2CO3, Ca(OH)2) have been used as HCl adsorbents [4,16,25,27].Another option for quality improving of the chlorinated pyrolysis products is the upgrading. A. Lopez-Urionabarrenechea et al. reported, that chlorinated oils from waste plastic pyrolysis could be upgraded by red mud, because it can promote the cracking and dechlorination reactions [27]. Andrei Veksha et al. can upgrade the non-condensable pyrolysis gases using CaCO3 supported catalytic sorbent containing 5% NiO during high temperature (700 °C) process [28]. S. Kumagai et al. investigated the effect of calcium oxide and calcium hydroxide catalysts on the product distribution of decomposition at 600 °C of individual and mixed plastics under a steam atmosphere. CaO increased gas and liquid production from the plastic mixture, which was further enhanced in the presence of Ca(OH)2 [29].It is worth to say, that owing to the reaction scheme of decomposition reactions, the high olefin content is the characteristic of the pyrolysis products, however it is changing wide range. Generally pyrolysis oils have 30–70% unsaturated hydrocarbons, depending on the reaction parameters [30–36]. Not only the before mentioned chlorinated compounds in pyrolysis oil, but also the unsaturated hydrocarbons can occur problems during the long term utilization. As it is known, the reactivity of the unsaturated compounds is higher than saturated hydrocarbons due to the distorted electron cloud. The consequence of this fact is that CC bonds can easily react with another CC bond, which lead to oligomerization and polymerization e.g. during long term storage. The high temperature or sunlight radiation redounds to those reactions [30–36]. More information is available about the aging of biomass sourced pyrolysis oils (hardwood, softwood, sewage sludge, wheat straw, chicken manure, etc.), but only few result is published regarding the aging of pyrolysis oil obtained by contaminated or PVC containing plastic waste [37–39]. Regarding the aging test, an accelerated test at 80 °C for 24 h is the mostly used.In this work, the thermo-catalytic pyrolysis of real waste plastics containing PVC was performed for in-situ improving of pyrolysis oil and the long term utilization of pyrolysis oil was investigated via accelerated aging and corrosion tests.Mixtures of polyolefin rich real municipal waste plastics contained PVC was used as raw material in our current work: 35% LDPE, 32% HDPE, 24% PP, 4% PVC, 3% ethylene-propylene dimer and 2% polystyrene. Raw material was supplied by household collection and contained 1.5% chlorine, while the ash content was 4.7%. Waste plastics had been shredded and then crashed by laboratory device (Dipre).Mixtures of Ni/ZSM-5, Ni/SAPO-11 catalysts, Ca(OH)2 (supplied by VWR) and red mud from Bayer process for alumina production were added to the raw materials for the in-situ product improving. Red mud contains Fe2O3 (74.8%), Al2O3 (9.8%), CaO (4.6%), TiO2 (3.2%), SiO2 (2.2%), MnO (1.2%), Na2O (0.9%), V2O5 (0.4%), SrO2 (0.2%) and 2.6% others. The BET surface of the Ca(OH)2 and red mud was 28.4 and 35.1 m2/g, respectively. Table 1 summarizes the composition of catalyst mixtures.The ZSM-5 ([Nan(H2O)16][AlnSi96-nO192]) catalyst supporter is MFI-type 10 membered ring channel synthetic zeolite with 5.3 × 5.6 Å channels, while AEL-type, crystalline silicoaluminophosphate (SAPO-11) catalyst ([Nan][Al20SinP20-nO80]) supporter is a one-dimensional also 10-membered, but elliptical pore channel synthetic zeolite with 6.3 × 3.9 Å pore size.Both the ZSM-5 and SAPO-11 catalysts were loaded by nickel using the following procedure. Catalysts were continuously stirred in 1 M Ni(NO3)2·6H2O (supplied by VWR) at 80 °C for 5 h, then they were washed by deionized water, filtered and dried for 10 h at 110 °C. In the final step, each catalyst was conditioned at 500 °C for 5 h in air.Ni/ZSM-5 had 15.1 Si/Al ratio, 324 m2/g BET surface and 1.55 nm average pore diameter. Contrary, Ni/SAPO-11 catalyst had 0.25 Si/Al ratio, 211 m2/g BET surface and 2.11 nm average pore diameter. The micro, meso and total pore volumes were 0.081, 0.018 and 0.099 cm3/g regarding Ni/ZSM-5 and 0.052, 0.063 and 0.15 cm3/g in case of Ni/SAPO-11, respectively. The nickel content of catalysts was 9%, since the catalyst supporters were able to bind this amount of nickel during the impregnation of catalysts. This amount of nickel content is appropriate according to literature references [40–42].Waste plastics were pyrolyzed in a one stage stainless steel batch reactor at 510–520 °C in nitrogen atmosphere (Fig. 1 ). The heating rate was 15 °C/min. The nitrogen flow was 5 dm3/h [43]. The reactor external wall was fitted with electric heating system and the temperature was controlled by PID controller. Firstly, 50 g of the raw material together with 2.5 g of catalyst was placed in the reactor, then, after the assembling of the rig, the temperature was elevated to the set value.Volatiles from the decomposed hydrocarbons were driven though a heat exchanger, where condensable from hydrocarbon gases had been transformed into liquid product, then the fractions were separated in phase separator into pyrolysis oil and hydrocarbon gases. Hydrocarbon gases were bubbled through a scrubber filled with sodium hydroxide solution. The yields of products were calculated based on the mass balance. The weight of pyrolysis oil and residue were measured after the experiment. Gas yield was calculated in the knowledge of pyrolysis oil and solid residue (Y(gas) = 100-Y(pyrolysis oil)-Y(residue)).A DANI type GC instrument was used fitted with programmed injector, flame ionized detector for analysis of hydrocarbon composition of gases: Rtx PONA column (100 m × 0.25 mm, surface thickness of 0.5 μm) and Rtx-5 PONA (100 m × 0.25 mm, surface thickness of 1 μm). Sample analysis was taken using isotherm conditions (T = 30 °C). The temperature of injector and detector were 240 °C.Hydrogen content of the gas products was measured by a Shimadzu GC-2010 gas chromatograph (TCD detector, Carboxen™ 1006 PLOT column (30 m × 0.53 mm)). The temperature was increased from 35 °C (hold time 2 min) to 250 °C at 40 °C/min heating rate, then the final temperature was hold till 5 min.Pyrolysis oil was analysed by DANI type GC (Rtx 1 dimetil-polysiloxan capillary (30 m × 0.53 mm, thickens of 0.25 μm)) using the following temperature program: 40 °C for 5 min, then the temperature was elevated by 10 °C/min till 350 °C and it was kept at 350 °C till 20 min. Both the injector and detector temperature was 350 °C. Components were identified based on their retention time, while the pyrolysis oil composition was evaluated via peak areas belong to the components.The chlorine content of products was measured by x-ray instruments (PHILIPS MiniPal PW 4025/02 non-polarized EDXRFS). The spectrometer was powered by PW 4051 MiniPal/MiniMate Software V 2.0A, equipped with a rhodium-side window tube anode and Si-PIN detector. The fluorescent x-ray was detected by a Si-PIN detector with a beryllium window. The analysis was taken in helium medium.The long term application of the pyrolysis oils was followed by accelerated aging and corrosion test. Samples were stored in a sample holder at 80 °C till 7 days. The density (EN ISO 12185) and the viscosity (EN 16896) were measured at 25 °C, moreover the Total Acid Number (TAN) (ISO 6618) and Jodine Number (MSZ 19971-1983) were followed at 0, 1, 3, 5 and 7 days of the test. To investigate the corrosion property, a copper plate was stored in pyrolysis oil at 25 °C till 60 days and the weight loss of the copper plate was calculated by the following equation: (1) w e i g h t l o s s = m i − m f m i · 100 A where mi and mf are the initial and final weights of the metal plate, A is the surface area of the plate.The pyrolysis oils were also investigated by Fourier Transformed Infrared spectroscopy (TENSPR 27 FTIR instrument with ATR unit). The resolution was 3 cm−1, the illumination was SiC Globar light, a RT-DLaTGS type detector was used. The change of the olefin content was followed by the integrated peak areas at 890, 910, 950 and 990 cm−1 using the following equation: (2) Δ ( I A ) = I A ( x ) 0 − I A ( x ) i I A ( x ) 0 · 100 where IA(x)0 is the integrated peak area at x = 890, 910, 950 and 990 cm−1 without aging, and IA(x)i is the integrated peak area at x = 890, 910, 950 and 990 cm−1 after i = 1,3,5,7 days.The oligomers and polymers formed by the aging test was separated from the by pyrolysis oil by filtration using VWR (514-0066) filer, Nylon, 0.2 μm, D 25 mm type device. Then the amount of the separated fraction had been weighted and the polymerized ratio was calculated as following: (3) P R = m i − m 0 m 0 · 100 where mi is the weight of the separated oligomers-polymers i = 1,3,5,7 days and m0 is the weight of the tested sample.The yields of pyrolysis products are summarized in Fig. 2 . Pyrolysis was taken till no volatiles could evaporate from the reactor. In general, higher yield of gaseous products (18.8–26.2%) was observed over ZSM-5 based catalyst mixtures; while more pyrolysis oil was obtain using Ni/SAPO-11:Ca(OH)2:red mud catalyst mixtures (64.2–71.9%). As it was before discussed, Ni/ZSM-5 catalysts had larger BET surface area, than Ni/SAPO-11 catalysts, furthermore the micropore volume of Ni/ZSM-5 catalysts were also higher than Ni/SAPO-11. For higher yields of gases especially the larger micropore volume could be blamed. It is also well established that the CC bond cracking properties of Ni/ZSM-5 with a high Si/Al ratio were higher than that of the Ni/SAPO-11 catalyst. This was particularly significant for 1:1:2 red mud:Ca(OH)2:Ni/ZSM-5 catalysts. It is important to mention, that the ratio of pyrolysis oil and gases was higher using SAPO-11 based catalysts, than that of ZSM-5 based. During the pyrolysis, the hydrocarbon macromolecules enter the pores of the catalyst, where they further cracked. Due to the pore size of the catalysts, the decomposed polymer macromolecules could escape from the smaller pore size ZSM-5 supported catalysts only after a larger degradation. The SAPO-11 has a bit larger average pore diameter than ZSM-5; for this reason, the larger molecules can also come out from the pores of the catalyst without further cracking. When these longer carbon-chain molecules reach the active centres of the catalyst they undergo cyclization and polycyclization [44–46]. Due to these reactions, coke formation is initiated which leads to deactivation of catalyst. The more coke deposition on the catalyst, the larger coke agglomerates are formed on its surface.Regarding the residue, a bit more solid phase decomposed products was obtained over SAPO-11 based catalysts, than ZSM-5. Residue was investigated by SEM instrument. It was concluded, that catalysts, inorganic content of the plastic and the coke agglomerated into larger particles in case of Ni/SAPO-11 based catalyst mixtures (Fig. 3 ). Whereas, the catalyst grain were well separated from each other and did not covered by thick and hard coke layer.Gases contained hydrogen and C1 C5 hydrocarbons, dominantly C2, C3 and C4; n-olefin, n-paraffin, branched olefin and branched paraffin. The compositions of hydrocarbon gases are summarized in Table 2 .Without catalyst, the n-olefin (38.5%) and the n-paraffin (41.2%) hydrocarbons were the main products. However, owing to the isomerization effect of the catalysts, the concentrations of branched hydrocarbons can significantly increase during thermo-catalytic pyrolysis, such as isobutane, isobutene, trans-but-2-ene, cisz-but-2-ene, isopentane, 2,2 dimethyl-butane, methyl-pentane. Both the Ni/ZSM-5 and Ni/SAPO-11 containing catalyst mixtures occurred notable increasing in branched hydrocarbons. Without catalyst the concentrations of non-branched and branched hydrocarbons was 80.5 and 19.5%, respectively. Contrary, the summarized concentration of branched hydrocarbons was 51.2–56.1% in pyrolysis oil by Ni/ZSM-5 catalyst mixtures and 47.7–58.6% by mixtures of Ni/SAPO-11:Ca(OH)2:red mud catalysts. It is worth to note, that the isomerization effect could only slightly decreased by the adding of Ca(OH)2 and red mud to the catalyst mixture.The n-olefin/n-paraffin ratio was between 0.57 and 1.15 by catalysts. The thermal and catalytic degradation of hydrocarbons occurs primarily by β-scission, resulting approximately 1:1 ratio of n-olefin and n-paraffin. As it is shown by the results, the amount of olefins can be increased by the using of catalysts. For the isomerization of hydrocarbons obtained by the degradation of the longer molecules, the catalyst must have sufficient acidity for the rearrangement of the hydrocarbon skeletal and dehydrogenation/hydrogenation function. Results well shown, that significantly more isomers of hydrocarbons were obtained using high concentration of both Ni/ZSM-5 (112-Z) and Ni/SAPO-11 (112-S) catalysts. Results show, that the ratio of non-branched/branched hydrocarbons decreased from 3.93 to 0.71–1.09 by the use of catalysts; but there were no trend-like, significant differences among the catalysts. However, the using of catalyst mixtures with 50% Ni/ZSM-5 and Ni/SAPO-11 catalysts the highest branched and lowest non-branched hydrocarbons were resulted.It is worth to investigate the ratio of saturated and unsaturated hydrocarbons within the branched molecules. Less branched olefin and more branched paraffin were obtained by ZSM-5 catalysts (111-Z, 112-Z, 121-Z, 211-Z) than SAPO-11 catalysts (111-S, 112-S, 121-S, 211-S). Comparing the effects of the two catalysts to the formation of branched unsaturated hydrocarbons, it can be concluded, that the higher surface areas and Si/Al ratios were the cause for elevated olefin content regarding ZSM-5 catalysts. For example, the ratio of branched olefin/branched paraffin ranged between 1.74 and 2.43 for ZSM-5 catalysts or 3.58–5.26 for SAPO-11 catalysts. In both cases, the highest value was achieved by 121-Z and 121-S catalyst, while the lowest by 2:1:1 ratios. The SAPO-11 catalyst is particularly beneficial for the formation of isobutene, trans-but-2-en, cis-but-2-en and isopentene. It is also observed, that more multi-branched isomers was obtained over SAPO-11 catalysts.It is right, that the used pyrolysis temperature was low for significant hydrogen production over transition metal loaded catalysts [47]. The hydrogen content of the gas products was 0.5% without catalysts, which could be increased to 1.1–4.5% and 1.5–5.9% using Ni/ZSM-5 and Ni/SAPO-11 containing catalyst mixtures, respectively. That result was also supports the higher isoolefin content. The reason for this may be the following. Firstly, the n-paraffin adsorbed and dehydrogenated on the metallic active centres of the catalyst resulting n-olefins, and then n-olefin transformed to carbenium ion at the acidic sites of the zeolite. That non-branched carbenium ion rearrange to branched olefin beside proton loss. It is also known, that the branched olefin is hydrogenated to paraffin and leaves the catalyst using suitable zeolites. The metallic sites are required to maintain the proper olefin concentration and to maintain the propagation reactions. On the SAPO-11 catalysts, in the first step, hydrogen and carbonium ions are formed, which is also isomerised, but the final hydrogenation of the branched olefin into branched paraffin was slightly occurred, which led to higher hydrogen content in gases. In both cases, the high Ni ion-exchanged catalyst (112-Z and 112-S) could mostly increase the hydrogen concentration, while red mud mixed into the catalysts had synergic effect. The relatively low hydrogen content was also supported by previous results. E.g. A. López et al. demonstrated that less than 1.2% hydrogen was obtained from stepwise pyrolysis of the mixtures of PE, PP, PS, PET and PVC, while the gas fraction contained mainly C2, C3 and C4 hydrocarbons [4].The pyrolysis oils were analysed by GC-FID. Results are shown in Fig. 4 . Based on result, pyrolysis oils were the mixtures of C5 C30 hydrocarbons: n-olefin, n-paraffin, branched hydrocarbons and aromatics.Firstly, the catalysts can also affect the concentrations of different compounds in pyrolysis oils. The concentrations of n-paraffin, n-olefin, branched hydrocarbons (both saturated and unsaturated) and aromatic were 30.8%, 27.0%, 40.3% and 1.9% without catalyst. It is well shown, that mostly the concentrations of aromatics and branched hydrocarbons can increase with the addition of catalysts.Each of the catalysts showed high activity in isomerization reactions which led to the increasing in branched hydrocarbons, however, aromatization reactions was occurred rather on Ni/ZSM-5 containing catalysts. Regarding aromatic compounds in pyrolysis oil from catalyst free experiment, the polystyrene in raw material was their source. Contrary, the aromatic content was 11.4–17.6% over Ni/ZSM-5 containing catalysts, while 8.1–11.0% using Ni/SAPO-11 containing catalysts; mostly benzene, toluene, xylenes and other short-chain substituted aromatic hydrocarbons. It is important to remark, that owing to the difference in catalyst channels, the concentration of short-chain substituted aromatic hydrocarbons was higher when Ni/SAPO-11 containing catalysts were used, than that of Ni/ZSM-5 containing. One dimension of Ni/SAPO-11 catalyst was larger, than Ni/ZSM-5 catalyst; therefore the larger hydrocarbons (e.g. short alkyl chain substituted aromatics) can leave the catalyst surface without decomposition. It is also well shown, that Ni/SAPO-11 catalysts had a bit favourable effect for production of n-paraffin, however the concentration of branched hydrocarbons were tendency-like less, than that of using Ni/ZSM-5 catalysts.The n-paraffin/n-olefin ratio were 1.09–1.18 using Ni/ZSM-5 containing catalysts, while 1.31–1.43 over Ni/SAPO-11 catalysts. The ratio of branched and non-branched aliphatic hydrocarbons in the pyrolysis oil was 0.70 for cracking without catalysts, which increased to 1.31–2.17 by Ni/ZSM-5 catalysts and 0.97–1.75 using Ni/SAPO-11 catalysts.The ratio of branched and non-branched aliphatic hydrocarbons increased in the order of 211, 121, 111 and 112 signed catalysts using both ZSM-5 and SAPO-11 supporters. This was due to the catalyst skeleton rearranging property. However, more n-paraffin and less n-olefin hydrocarbons were produced over SAPO-11 catalyst. This is due to local hydrogenation reactions. During the decomposition of polymer waste, the source of hydrogen is primarily the cyclization and aromatization reactions of aliphatic hydrocarbons and the forms of isomerization reactions when non-branched hydrocarbons transformed to branched unsaturated hydrocarbons. It is also important to remark, that the hydrogen content of the gas products correlates with the ratio of n-parafin/n-olefin in pyrolysis oils in case of SAPO-11 catalysts. Regarding the amount of hydrogen, for catalysts, primarily the aromatization reactions can be blamed while regarding the SAPO-11 catalysts the dehydrogenation isomerization reactions resulting in unsaturated branched hydrocarbons should be also caused according to the following reaction scheme (Fig. 5 ).The olefin content of pyrolysis oils was followed via their Jodine number (Fig. 8). According to the result, the total olefin content of the products was higher with Ni/SAPO-11 based catalysts than on Ni/ZSM-5 based regarding each sample. Considering, that the n-olefin content of products is decreased in the presence of catalysts and the fact that the n-olefin content of the products was lower using SAPO-11 supported catalysts than that of the ZSM-5, it is easy to see that more branched olefins can be obtained over SAPO-11 catalysts with accordance the before mentioned reaction scheme.The chlorine content of pyrolysis products is important for their future utilization, which is summarized in Fig. 6 . Catalysts can significantly affect the chlorine content of the products. It is well known that the decomposition of PVC is taken at relatively low temperature (300–350 °C), which is lower than the temperature needed to the cracking of the polyethylene CC bonds (400–420 °C) [4]. The dechlorination efficiency is increase by the increase of the temperature and pyrolysis time [13]. However higher temperature needs more energy and due to the recombination reaction, pyrolysis oils contain higher concentration of chlorinated compounds. During the pyrolysis of PVC, the CCl bond with high dipole moment begins to break, resulting in chlorine and alkyl radicals. Hydrogen may be also generated by further reactions from the alkyl radicals, which should recombine with chlorine to produce HCl. This will primarily appear in gas products. However, the chlorine radical may also be recombined with alkyl radicals, resulting in organic chlorine compounds which will initially appear in the pyrolysis oil and the residual product of pyrolysis. The problem is mainly with organic chlorine compounds in pyrolysis oil because their removals are difficult and significantly deteriorate the long-term utilization and stability of the products. Inorganic chlorine in the gas product, mostly HCl, can be easily removed with various sorption processes.Regarding pyrolysis products, 102202 ppm chlorine was in the gases, 4364 ppm was in the pyrolysis oil and 1950 ppm was in the residue without catalyst. These values were significantly changed by the using of catalysts. The distribution of chlorine among the different products and the high chlorine content in gases was confirmed by others [27]. Generally, catalysts reduced the amount of chlorine in gas products and pyrolysis oil, while the amount of chlorine in the residue fraction increased. When analysing data, it is also worth considering that chlorine is generated in the first half of the decomposition reactions, that is why, the longer takes the pyrolysis, or the higher the yield of the given fraction, their diluent effect will prevail.Comparing the two zeolites, the chlorine content of the gas products was lower, and that of the pyrolysis oil were higher in all cases produced on the mixtures of Ni/ZSM-5 catalysts containing catalyst, than that of over Ni/SAPO-11 catalysts. This was the consequence probably the fact that, owing to the higher hydrogen content of the product mixture over SAPO-11 catalysts, the chlorine radicals were easily able to combine to HCl and reduce the concentration of organic chlorine compounds in hydrocarbons. This assumption is also supported by the fact that the chlorine content of the gas products correlates with the hydrogen content. It was confirmed by others, who concluded that the chlorine from PVC correlates with hydrogen to generate hydrochloric acid (HCl) at a lower temperature using metal oxide catalysts [14]. It is also worth note that the presence of red mud and Ca(OH)2 was highly preferred in the reduction of the chlorine content of pyrolysis oil. This was observed for both ZSM-5 and SAPO-11 supported catalysts. The advanced property of the red mud for chlorine reducing in gases and pyrolysis oil was also reported by A. Lopez-Urionabarrenechea et al. They demonstrated that vast amount of chlorine moved to the solid phase during the catalytic post treating of pyrolysis oil [27].The reason for this was that due to the alkali character of both substances, chlorine-containing compounds could be chemically react with additional compounds of catalysts, resulting in a significant increase in the residual chlorine content, while the chlorine content of the gas products and pyrolysis oils was significantly reduced. The chlorine content of the pyrolysis oil decreased from 4364 ppm to 284 ppm while the chlorine content of gases changed from 81762 ppm to 48128 ppm using 211-Z, while 228 ppm chlorine was measured in pyrolysis oil using 211-S catalyst. Fig. 7 shows the appearances of the pyrolysis oil after 60 days treating and summarize the result of corrosion test. A copper plate was put into the pyrolysis oil at 20 °C till 60 days. The change in weight of the copper plate was registers in each 5th day. According to results, increasing tendencies were found in case of both catalyst free and catalyst supported pyrolysis. The change in the weight of copper plate was 0.458%/mm2 without catalysts, which value could be significantly decreased by catalysts. It is also well shown, that lower weight loss was found in case of pyrolysis oils obtained by the using of SAPO-11 supported catalysts. Especially the presence of Ca(OH)2 and red mud rich catalysts showed high efficiency in better corrosion properties. E.g. the weight loss of copper plate was 0.055%/mm2 (121-Z) and 0.039%/mm2 (211-Z) or 0.039%/mm2 (121-S) or 0.025%/mm2 (211-S). These results are supported by the chlorine content of pyrolysis oils shown in Fig. 6. The lower the chlorine content of pyrolysis oil, the less the weight loss of the copper plate. Regarding the appearances of the pyrolysis oil, they were dark after the treating excluded the using of 121-Z, 211-Z, 121-S and 211-S catalysts. In those cases the colours of the oil only slightly become darker, and they kept yellowish transparent liquid.That result was confirmed also by TAN during the aging accelerated test (Fig. 8). The TAN of pyrolysis oil obtained by catalyst free pyrolysis was 5.9 mg KOH/mg sample. However pyrolysis oil had less acidic components by the using of catalysts, especially SAPO-11 based; 2.7–4.3 mg KOH/mg sample in case of ZSM-5 based catalysts, while 2.1–4.2 mg KOH/mg sample using SAPO-11 catalysts. On the other hand the TAN increased in less degree when catalyst was used for pyrolysis, especially with high Ca(OH)2 and red mud concentration. Without catalyst the change in TAN was 70% (from 5.9 mg KOH/g to 19.4mgKOH/g). The increasing in TAN was less using catalysts, especially in case of SAPO-11, because the TAN increased with 36–57% and 29–48% using Ni/ZSM-5 and Ni/SAPO-11 containing catalysts, respectively. Presumably acidic chlorine containing compounds are transformed from the chlorinated organic components during the aging, which can be neutralized by KOH.The change in the main properties of the pyrolysis oil as function of time is key property for their long term application. An accelerated aging test was performed to investigate the longer term utilization of pyrolysis oils and the effect of the catalysts to the properties. Results are summarized in Fig. 8.Regarding the density and viscosity slight increasing tendencies were found both without and with catalysts. However, catalysts had advantageous effect to both properties. The density and viscosity of the pyrolysis oil was 0.851 g/cm3 and 1.905 mm2/s without catalysts, respectively. The density could be increased by 1.99%, while the viscosity by 5.49% without catalysts. It is clear, that the catalysts can increase both the density and viscosity increasing of pyrolysis oil at the end of treating. It is also important to note, that lower density and viscosity of pyrolysis oils was measured by the using of catalysts. The density of pyrolysis oil obtained by absence of catalysts increased to 0.868 g/cm3 at 7th day of the treating at 80 °C. Comparing, when catalysts were used, the pyrolysis oil density was 0.773–796 g/cm3 without treating, which increased to 0.791–0.817 g/cm3 during the accelerated aging. It means, that the density can increased by 2.03–2.68% using Ni/ZSM-5 catalysts, while 2.57–3.06% using Ni/SAPO-11 catalysts. Similar phenomena was concluded regarding the viscosity, however the increasing ratio was higher: 5.95–9.32% in case of Ni/ZSM-5 catalyst mixtures and 9.15–11.44% by Ni/SAPO-11 based catalysts. It is important to remark, that same order of catalysts was found regarding the density and viscosity increasing in case of both Ni/ZSM-5 and Ni/SAPO-11 catalysts: 121, 211, 111 and 211. That order is the same, as it was earlier mentioned for increasing in the concentrations of unsaturated hydrocarbons. The higher concentration of unsaturated hydrocarbons is the cause for higher increasing in both density and viscosity using of SAPO-11 catalysts.The change in olefin content was also followed via the FTIR spectra of the products at 990, 955, 910 and 890 cm−1 wavenumbers. Pyrolysis oils show typical hydrocarbon FTIR spectra (Fig. 9 ). There were infrared activity between 2800 and 3000 cm−1 (CH2 , CH3 symmetric and asymmetric vibration bands), between 1250 and 1500 cm−1 (asymmetric and symmetric deformation stretching of CH3 groups). Peaks with low intensity at 1615 cm−1 was attributed to the aromatic hydrocarbons (CC streching vibration), while around 610 and 690 cm−1 refer to the chlorinated hydrocarbons CCl streching vibration, which was the less in case of catalyst mixtures with high red mud and Ca(OH)2 content. Bands in the range of 850–1000 cm−1 refer to the olefins and mowing vibration at 720 cm−1 was caused by CH2 group. In our current work the 850-1000 cm−1 range was investigated. There are four peaks in that range: at 890 cm−1 (vinylidene), at 910 and 990 cm−1 (vinyl) or 950 cm−1 (vinylene). In general, the infrared signal at 910 and 990 cm−1 decreased, while at 950 cm−1 increased by the using of catalysts, compared to catalyst free case, which refers to the rearrangement of hydrocarbon skeletal. Fig. 9 summarizes the change in peak areas during the accelerated aging. As it seems, the peak areas of the FTIR spectra of pyrolysis oils only slightly decreased (less than 5%), which is similar result as it was demonstrated regarding density or viscosity. It is important to note, that the change of vinyl type olefins was higher than that of the other two types, furthermore the catalyst use during the pyrolysis can also decrease the change of the integrated peak areas. Results refers, that vinyl type olefins had greater role during the aging than the other two. Glancing the result it can be also concluded, that the relative change in integrated peak areas was typically between 5 and 7 days using catalyst mixture with Ni/ZSM-5, whereas between 1 and 3 days in case of SAPO-11 supported catalysts. It means that the aging of pyrolysis oils obtained by ZSM-5 supported catalyst takes later than that of SAPO-11 catalysts.The concentration of unsaturated hydrocarbons during the aging was followed via Jodine number, which refers to the amount of the Jodine needed for the saturation of CC bonds in mass unit of the sample. It is important observation, that pyrolysis oils by higher unsaturated branched hydrocarbons resulted faster aging. However it is also worth to note, that only slight and not significant decreased was found, which compliant result with changing in density or viscosity is. Jodine numbers are slightly decreased as function of treating days. E.g. it changed from 149 to 144 regarding pyrolysis oil obtained from catalyst free pyrolysis, which means 3.6% decreasing after the 7th day. Ni/ZSM-5 based catalysts could not affected the Jodine number of pyrolysis oils (3.3–3.9% decreasing), however the concentration of unsaturated hydrocarbons decreased a bit greater extent using Ni/SAPO-11 based catalysts (4.3–6.1%). Results well demonstrate that the highest decreasing in unsaturated hydrocarbons was found by the using of 1:1:2 catalyst ratio both in case of Ni/ZSM-5 and Ni/SAPO-11 catalysts. Fig. 10 summarized the oligomer-polymer phase separated from the pyrolysis oils during the aging test, which well supported the infrared results. Without catalyst nearly linear relationship was found regarding the polymerized olefins as function of treating days. At the 7th day 0.15% of the pyrolysis oil could be separated by filtration. That value was significantly lower by the using of catalysts: 0.05–0.69% in case of Ni/ZSM-5 based catalysts, and 0.06–0.078% using Ni/SAPO-11 based catalysts. It means that the catalysts can reduced the amount of the separated fraction. Presumably, it could be explained by the difference in the distribution of the hydrocarbon structure of unsaturated compounds. In accordance with the FTIR result, in case of pyrolysis oils by ZSM-5 catalyst showed faster change during the end of the treating time (after 4th days), while the polymerization reactions took place rather during the first half of the accelerated aging (before 3rd day). It is important to remark, that the polymerized fraction is accumulated in the bottom of sample holder, and the pyrolysis oil kept transparent above that part.In this work the thermo-catalytic pyrolysis of real municipal plastic waste, the longer term stability and corrosion properties of pyrolysis products were investigated. It was found, that the PVC containing raw material could be converted mainly into pyrolysis oil, however the composition of catalysts had a notable effect to the gas:pyrolysis oil ratio. High synthetic zeolite containing catalyst mixtures can drastically increase the gas yield. Comparing the Ni/ZSM-5 and Ni/SAPO-11 catalyst mixtures, due to the larger pore areas and higher Si/Al ratio, the ZSM-5 containing compositions resulted higher yield in gases, than SAPO-11. SEM result well demonstrated that larger agglomerates were found in residues obtained from thermo-catalytic pyrolysis using SAPO-11 based catalyst, than ZSM-5. Regarding gases, catalysts can isomerize the main carbon frame (especially using catalysts with high concentration of Ni/ZSM-5 and Ni/SAPO-11), and promote the production of unsaturated hydrocarbons; however the n-olefin and n-paraffin ratio slightly increased. The synergetic property of Ni ion-exchanged catalyst and red mud to hydrogen production was also demonstrated. High concentration of red mud led to the highest proportion of hydrogen in gases: 121-Z and 121-S; furthermore more hydrogen was obtained using SAPO-11 based catalysts. In accordance with hydrogen content, the concentration of unsaturated branched hydrocarbons, especially isobutene, trans-but-2-en and cis-but-2-en were also high in case of SAPO-11 containing catalyst. Pyrolysis oils contain C5 C30 hydrocarbons, while the aromatic, branched and unsaturated hydrocarbon content increased by the use of catalysts. ZSM-5 based catalysts show high efficiency in aromatization reaction. For conclusion the stability tests, the density and viscosity are only slightly increased, while the olefin content (especially vinyl type) slightly decreased during the accelerated aging at 80 °C till 7 days, demonstrating, that the high vinylene olefin content of given fractions is not crucial regarding the stability. Catalysts can decrease the ratios of changing comparing to the catalyst free thermal pyrolysis. However, the aging of pyrolysis oils obtained by Ni/ZSM-5 containing catalyst takes later than that of Ni/SAPO-11. Contrary, acidic components can cause more significant change in both TAN or even during the corrosion test. Both the TAN and relative change in copper plate weight loss during the corrosion test were lower using catalysts during the pyrolysis, especially SAPO-11 based. Regarding the chlorinated compounds, however a single step and not stepwise process was used, the chlorine could be transformed mainly into gaseous products and the chlorine concentration in pyrolysis oils can further decreased by the using of catalysts. Especially the presence of red mud and Ca(OH)2 was highly preferred for the reduction of the chlorine content of pyrolysis oil, and catalyst mixtures with Ni/SAPO-11 showed better properties for chlorine reduction than Ni/ZSM-5 based.The authors acknowledge the Horizon 2020, Marie Curie Research and Innovation Staff Exchange (RISE) (MSCA-RISE-2014 (Flexi-pyrocat, No.: 643322)). The authors also acknowledge the financial support of Széchenyi 2020 under the EFOP-3.6.1-16-2016-00015 and Economic Development and Innovation Operative Program of Hungary, GINOP-2.3.2-15-2016-00053: Development of liquid fuels having high hydrogen content in the molecule (contribution to sustainable mobility).

This work is focussing to the thermo-catalytic batch pyrolysis of contaminated real municipal plastic waste using different catalyst mixtures in their different ratios: Ni/ZSM-5, red mud, Ca(OH)2 and Ni/SAPO-11, red mud, Ca(OH)2. The effect of the catalysts to the pyrolysis oil properties and the in-situ upgrading (especially the storage, transportation and corrosion stability) of pyrolysis oil was investigated. High concentration of Ni/ZSM-5 and Ni/SAPO-11 zeolites in catalyst mixtures can increase the yield of gases and pyrolysis oil, the concentration of aromatics or the hydrogen content in gases; however the presence of red mud in higher content can further increase the hydrogen concentration. ZSM-5 based catalysts showed higher efficiency in aromatization reactions. An accelerated aging test at 80 °C till 1 week was performed to investigate the storage and transportation stability of pyrolysis oils. Only slight increase was found in the density and viscosity, on the other hand there was a bit greater increase using SAPO-11 based catalysts than ZSM-5. The change in the olefin content was followed via bromine number and FTIR spectra of pyrolysis oil, which resulted ∼3% and ∼4% decreasing using Ni/ZSM-5 and Ni/SAPO-11 containing catalyst mixtures. Regarding acidic components, they significantly increased by aging time, while the high red mud and/or Ca(OH)2 in catalyst mixtures had notable benefit, because they can drastically decrease the concentration of chlorinated compounds, which led to less weight loss during corrosion test using copper plate till 60 days at 20 °C.

Hydrogen (H2) has significant advantages as an energy vector compared to petroleum or other conventional fossil fuels, although currently there are problems that must be solved, associated with its production, storage, and transportation [1]. Formic acid (FA) is a non-toxic renewable biomass material that can be used as an ideal liquid hydrogen carrier and achieve efficient hydrogen production. The development and utilization of FA fuel cells, using it as a hydrogen storage medium, is an effective method to solve the problem of energy depletion and environmental degradation [1,2]. However, the side reaction in the dehydration process of FA causes the catalyst to be poisoned resulting in a decrease in catalytic performance [3–5]. The development of high-performance FA dehydrogenation catalysts is of great significance for promoting the commercial application of these fuel cells.FA is produced by chemical methods such as the hydrolysis of methyl formate, but it is also obtained in equimolar proportions together with levulinic acid, by hydrolysis of cellulose raw materials derived from biomass. Currently, with the increased interest in the production of levulinic acid and other valuable chemicals from biomass, it is important to develop processes to use the derived formic acid, since otherwise, it constitutes a waste material [1]. In this direction, the interest in the use of the decomposition reaction of FA to produce H2 has increased remarkably. Therefore, the challenge is to produce pure H2, with minimum CO content, at the lowest possible temperature. This demand can be achieved through the careful choice of the catalyst and the reaction conditions, which is why many research efforts are currently being devoted to this line.The production of H2 from FA using homogeneous and heterogeneous catalysts has been studied in aqueous [3] and vapor phases [2,3], but in most cases, formulations based on noble metals, such as Rh, Pt, Ru, Au, Ag, and Pd supported on C, Al2O3 and SiO2 have been investigated [1–3]. For the vapor phase reaction, Solymosi et al. [2] found the following order of activity on a set of carbon-supported noble metals: Ir > Pt > Rh > Pd > Ru.On the other hand, in the case of non-noble metals, molybdenum carbide has attracted great interest but it showed substantially lower activity even at high temperatures [6]. In an earlier study, we reported the catalytic performance of molybdenum carbide supported on carbon with promising results [7], though high temperatures are required for the catalyst synthesis. Moreover, the catalytic activities of supported Cu and Ni catalysts have been measured in the FA decomposition reaction, proving to be active at relatively higher temperatures (> 220 °C) than noble metal catalysts [8–10]. These authors study also the catalytic decomposition of FA on Ni, and Ni-Cu alloy powders and report an improved selectivity toward dehydrogenation reaction although lower rate when Cu is added to Ni. More recently, Pechenkin et al. described 100% conversion and 98% yield to H2 using 10%CuO-5%CeO2/γ-Al2O3 at 200 °C [11]. In this line, we have shown a Ni/SiO2 catalyst doped with 19.3 wt% of Ca which gives 100% conversion of formic acid at 160 °C, with a 92% selectivity to hydrogen [12]. In addition, we recently reported that the bimetallic Ni-Cu system supported on carbon has better catalytic performance than the monometallic Ni or Cu catalysts [13]. Bulushev and coworkers [14] reported an improved catalytic performance for the hydrogen production from formic acid over Ni catalysts supported on carbon doped with nitrogen. These authors also suggested the Ni single atoms stabilized on the pyridinic nitrogen sites are responsible of the improved behavior of these Ni catalysts [15]. In addition, we have demonstrated that the catalytic activity and selectivity of the bimetallic Ni-Cu system is enhanced if the carbon support is doped with N-pyrrolic heteroatoms [16].In supported Pd catalysts, the temperature required for vapor phase FA decomposition can be reduced to less than 80 °C by the addition of K2CO3 [1]. The difference lies in the initial stages of the reaction since FA reacts with potassium ions to give formate species dissolved in the formic acid/condensed water solution in the catalyst pores [5] increasing the overall activity of the process. In a similar way, studies of adsorption and decomposition of FA on potassium modified Cu(110) reveal that the modification of the copper surface with potassium is accompanied by a decrease in the temperatures of HCOOH decomposition [17]. In a similar way, added cesium on Cu(110) increases formate species production during FA adsorption and accelerates its decomposition [18].Hence, in this work, the synthesis of Ni-Cu bimetallic catalysts supported on a non-porous high surface area graphite to be used in the decomposition reaction of FA in the vapor phase was carried out. The effect of doping on support with alkali metals (Li, K and Na) is studied, using in all cases an atomic ratio of alkali metal to active metal (Ni, Cu or Ni-Cu) equal to unity. The catalysts were characterized by X-ray diffraction (XRD), temperature-programmed reduction (TPR) and transmission electron microscopy (TEM). In addition, the programmed temperature surface reaction (TPSR) of FA was studied, analyzing the gases released at the outlet of the reactor employing the mass spectrometer, and attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy of the used catalysts was also performed.Nickel and copper monometallic catalysts and also bimetallic catalysts were synthesized. The technique employed to impregnate the metals was incipient wetness impregnation. Commercial high surface area graphite (H400; SBET = 399 m2/g) was obtained from Timcal Graphite. In all cases, the total metal concentration was 5% by weight, with 2.5% of each metal in bimetallic catalysts. The salts Ni(NO3)2·6H2O (Alpha Aesar) and Cu(NO3)2·3H2O (Sigma Aldrich) were used as precursors. The samples were dried at 100 °C. The following salts were used for doping with alkaline elements the graphite support: Na2CO3·H2O (Alpha Aesar), KNO3 (Panreac) and LiNO3 (Sigma Aldrich). The alkaline loadings, Li (0.6% w / w), Na (2% w / w) and K (3.4% w / w) were added so that the resulting catalysts contain the same atomic charge of alkaline metal and active metal. To incorporate the dopants, the incipient wetness impregnation method was also used. These materials were dried in an oven for 12 h at 100 °C and treated in He at 500 °C for 3 h. Subsequently, the metallic charge of Ni and Cu with 2.5% of each metal, was incorporated as previously indicated.The crystalline structure of the samples was examined by X-ray diffraction (XRD) using an X ́Pert Pro PANalytical. The temperature-programmed reduction (TPR) experiments were carried out in a conventional fixed-bed flow reactor and the effluent gases were continuously monitored by a thermal conductivity detector (Konik TCD); the samples were heated up in a 5% H2/Ar stream with a rate of 10 °C/min up to 675 °C. Transmission electron microscopy (TEM) images of the reduced catalysts were acquired using a JEOL 2100F field emission gun electron microscope equipped with an energy dispersive X-ray (EDX) detector. The fresh sample was reduced at 300 or 400 °C for 1 h in a pure H2 stream. The catalysts were reduced ex-situ and a He flow passivation procedure was carried out at room temperature. The used samples were measured after the catalytic test and a passivation procedure to room temperature in an inert atmosphere. The samples were dispersed in ethanol and mounted on the appropriate grid for the TEM microscope. The particle size was determined by counting at least 300 particles.The catalytic activity measurements for the FA decomposition in the vapor phase were carried out in a conventional fixed-bed flow reactor. The copper and bimetallic catalysts were pretreated in H2 flux at 300 °C for 1 h and then cooled in N2 flux at the reaction temperature. The nickel catalyst was pretreated at 400 °C for 1 h in H2 flux and then cooled in N2 flux at the reaction temperature. A mixture of FA diluted with N2 was fed to the reactor using a saturator-condenser at 15 °C (HCOOH concentration equal to 6%, with a flow of 25 ml min-1). For all the experiments, 75 mg of catalyst were charged to obtain a ratio of W (weight of catalyst)/ F (total flux) equal to 5 10-5 g h ml-1. The reactants and products were analyzed by gas chromatography (Varian 3400) fitted with a 60/80 Carboxen TM 1000 column and a thermal conductivity detector (TCD). At each temperature, a few measurements were performed in order to ensure that steady-state activity was reached. During the test, the unique products determined were CO, CO2 and H2. The total conversion of formic acid was determined as the sum of CO and CO2 concentrations related to the initial concentration of FA. In addition, the CO2 selectivity was calculated as the CO2 concentration related to the sum of CO and CO2 concentrations. The catalysts were studied in two heating cycles. The stability of the catalyst was evaluated over 14 h at a selected temperature. Catalytic tests with the supports proved that conversion was negligible.The temperature-programmed surface reaction (TPSR) measurements were carried out in conventional dynamic vacuum equipment coupled to a quadrupole mass spectrometer (SRS RGA-200). The catalysts were reduced before experiments in hydrogen flow at 300 or 400 °C and were degassed under high vacuum at the same temperature. The adsorption was then carried out using a 40 Torr pulse of HCOOH at 40 °C. Once the gas phase was evacuated, the desorption step was carried out at a programmed temperature, analyzing the gases released employing the mass spectrometer. The evolution of signals assigned at H2, N2, H2O, CO, HCOOH, CH2O, O2, and CO2 (m/e = 2, 14, 17/18, 28, 29/46, 30, 32 and 44, respectively) was followed as a function of temperature. Calibration of the relative intensity of the H2 and CO2 signals, m/e equal to 2 and 44, respectively, was performed. In a conventional vacuum equipment system, a certain number of moles of H2 and CO2 was admitted and a recirculation pump was employing for the mixture of the gases and then this stream was analyzed by the mass spectrometer.Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectra were acquired using a Bruker Vector 22 spectrometer equipped with a germanium crystal. For the ATR experiments, the SiC was carefully separated from the solid catalyst after the catalytic tests (stability tests for 14 h on stream) and then, was placed the sample on the crystal and collected the spectrum. A total of 256 scans with a resolution of 4 cm-1 were collected to measure each used catalyst.Several articles have been widely investigated for CO2 and H2 reactions supported by noble metals (Rh, Ru, Ir, Pt and Pd) and non-noble metals (Ni, Co, Cu and Fe) [1–6,19–31]. The metal and promoters, redox properties, acid/base features and surface area of the support, the metal particle size and metal-support interactions are the key factors for obtaining a good activity/stability balance on the catalysts.The hydrogen production from FA has been studied fundamentally on catalysts based on noble metals [1–23]. The strength of this work is to achieve high conversion and selectivity employing non-noble metals such as Ni and Cu. A graphitic material with high specific surface was used as a support and it was modified by adding alkali metals to change its surface basicity and thus favor the adsorption of HCOOH and its decomposition. The supports were modified with Na, Li and K; the percentages used were such that the number of dopant atoms is equal to the amount of active phase present in the catalyst.The nomenclature of the catalysts only indicates the metal, Ni or Cu, in the monometallic and Ni-Cu in the bimetallic, because in all cases the support was H400 graphite material, and those in which an alkaline was added as /A (where A = Na, Li or K). The loadings were indicated in the experimental session.The reducibility of the supported catalysts was studied by temperature-programmed reduction (TPR). This technique is a powerful tool for the study of the behavior of metal precursors and obtaining the strength of the oxide-support interaction. Fig. 1 shows the TPR profiles of the catalysts in the temperature range of 40–675 °C. The copper monometallic catalyst profile shows a single peak at 226 °C associated with the reduction of highly dispersed Cu2+ [27,32].The nickel monometallic catalyst profile presents a reduction peak at 263 °C associated with easily reducible Ni2+ species and another peak at 313 °C associated with the reduction of Ni2+ bulk. Furthermore, a high-temperature peak, at about 416 °C, is asymmetric and wide is attributed to remaining Ni2+ particles and also to gasification of support carbon atoms in the vicinity of the nickel particles, which catalyze this reaction producing methane [28,33–36].The profile for the bimetallic catalyst shows peaks in intermediate temperatures to the monometallic ones. The main peak is located at 254 °C, one of less intensity at 318 °C and another broad peak in the region between 400 and 600 °C with a center at 522 °C. The lower temperature peak would be associated with the reduction of Cu2+ and Ni2+ particles while the other peaks could indicate the existence of small Ni2+ particles which reduce at higher temperature and metal catalyzed methanation of carbon atoms around metallic particles [30,37].On the other hand, doping with alkaline oxides slows down the reduction process of the Ni and Cu particles, causing a displacement of the TPR profiles to higher temperatures.The catalyst doped with Li has a profile similar to the bimetallic without dopant shifted to higher temperatures. The main peak is located at 298 °C, one of less intensity at 375 °C and another broad peak in the region between 400 and 600 °C with a center at 525 °C. The catalyst doped with Na has a similar profile, too; however, a high-temperature peak less marked. The Na doped catalyst profile shifted to higher temperatures, a peak at 340, another with less intensity at 407 and the third at 506 °C completing the profile at 600 °C.The catalyst doped with K has a different TPR profile, has a broad peak in the region between 200 and 425 °C with three positions marked at 292, 317 and 350 °C. In this catalyst, the proximity of the maximums could indicate a narrower particle sizes distribution. According to the results of the TPR experiments, it can be observed that when modifying the supports with alkali metals, the metal-support interactions were modified, displacing the reduction profiles at higher temperatures. The decrease in the reducibility of the samples which causes a shift to higher reduction temperature was observed on Ni-based catalyst doped with different contents of Na [38], on Cu catalysts modified with different metal oxides (MgO, BaO, ZnO and MnO) [39] and on Cu-Ni system doped with MnO [30].Considering these TPR results, before the catalytic test and analysis with characterization techniques, the catalysts were pre-treated in hydrogen flow at 300 °C, except for the Ni monometallic solid, which was pre-treated in hydrogen flow at 400 °C for 1 h.The catalysts were measured by X-ray diffraction to observe the presence of crystalline phases. The H400 support diffractogram ( Fig. 2) exhibits the characteristic diffraction pattern of graphitic materials, with a pronounced peak at 2θ = 26° due to reflection from the basal plane (002). The characteristic peaks of (100), (101), (004) and (110) crystal planes are also present [32,40].In the diffractograms of the reduced catalysts, it is observed that there are no changes in the graphitic crystalline structure after the incorporation of the metallic phases by incipient wetness impregnation. In reduced Cu monometallic catalyst, three peaks at 43.0°, 50.5° and 74.3° assigned to the characteristic diffraction peaks of metallic copper phase (JCPDS 65-9743) [28], corresponding to (111), (200), (220) plane phase, respectively. Furthermore, a broad peak at 2θ = 36.3°, 42.3° (over a graphite peak) and another at approximately 2θ = 61° are observed, indicative of the Cu2O crystal structure (JCPDS 01-078-2076) probably due to the atmosphere exposure between the pretreatment and the XRD experiment. In reduced Ni monometallic catalyst, three broad peaks at 37.2°, 43.3° and 62.9° assigned to the characteristic diffraction peaks of NiO phase (JCPDS 71-1179) [41].In the diffractogram of the reduced bimetallic Ni-Cu catalyst, no characteristic signals of Ni or Cu are detected (Fig. 2), probably due to the smaller particle sizes obtained in this solid.When the catalysts were doped with alkali metals, three peaks at 43.0°, 50.5° and 74.3° assigned to the characteristic diffraction peaks of metallic copper corresponding to (111), (200), (220) plane phase, respectively (JCPDS 65-9743) are observed [28]. The relative intensity of these signals concerning those of the support is greater for the catalyst with Li > K > Na. In the diffractogram of the catalyst doped with Na, signals are corresponding to sodium carbonate that remained without decomposing in the treatment at 500 °C with He flux (JCPDS 37-0451) [42]. No signals corresponding to Ni-Cu bimetallic nanoparticles were detected in the catalysts evidencing that these have sizes of 4.5 nm or less, which is the detection threshold of the XRD technique.The distribution, shape and size of the particles were examined by transmission electronic microscopy. TEM images were taken with different magnifications of the samples reduced in H2 at 300 °C (400 °C for the Ni monometallic catalyst). In Fig. 3, TEM images of the Cu and Ni monometallic, undoped, Na and K-doped bimetallic catalysts are shown. It can be seen that the particles of both metals are evenly distributed on the support. To estimate the mean size, 300 particles were measured, being 4.7, 4.9, 4.2, 4.3 and 5.3 nm for the Cu, Ni, Ni-Cu and Ni-Cu/Na and Ni-Cu/K catalyst, respectively. The doping with Na and K did not considerably modify the average size of the Ni-Cu bimetallic particles, but it did slightly modify the distribution of the particles (see histograms in Fig. 3). Fig. 4 shows the images obtained for the reduced Ni-Cu/K catalyst in the STEM mode for the select area for the EDX mapping of copper (yellow), nickel (light blue), potassium (magenta) and oxygen (red). These images showed Cu particles of heterogeneous size, while Ni and K appears well dispersed. The observation of larger copper agglomerates is in agreement with the XRD results (Fig. 2). Nevertheless, the EDX analysis revealed that Ni, Cu and K particles are evenly distributed on graphitic material and coincide in occupying the same space on the support, evidencing the formation of nickel-copper particles dispersed over the K doped HSAG.The decomposition of formic acid is via dual-path mechanisms: Selective dehydrogenation: HCOOH ( g ) → H 2 + CO 2 ∆ G 298 = − 48.4 kJ mol − 1 Undesirable dehydration:   HCOOH ( g ) → H 2 O + CO ∆ G 298 = − 28.5 kJ mol − 1 Therefore, high activity and selective catalysts are desirable for the generation of hydrogen from the decomposition of HCOOH. The catalytic activity of all catalysts was evaluated in a fix bed reactor with a mass/flow ratio (W/F) equal to 5 10-5 g h ml-1 in the temperature range of 60–200 °C to build light-off curves. Fig. 5 shows the conversion of formic acid as a function of the reaction temperature for the series of mono and bimetallic Ni and Cu catalysts. Previous to the catalytic test, the catalysts were pre-treated in hydrogen flow at 300 °C, except for the Ni monometallic solid, which was pre-treated in hydrogen flow at 400 °C for 1 h, considering the TPR results.It is important to note that the comparison of the catalytic activity was measured when the Ni-Cu/Na catalyst was reduced at 300 and 400 °C and no marked difference was found (Fig. S1-supplementary information). The conversion values as a function of time for the two experiments were also included, observing that they are almost coincident in the 14 h tested at 120 °C (Fig. S2-supplementary information). Therefore, it is concluded that the choice of the pretreatment temperature was adequate.The light-off curves were performed following the same procedure in all the samples. After reducing the catalysts in H2 flux, it was cooled in a stream of N2 to 60 °C, and then the reaction mixture was fed with a concentration of HCOOH of 6% in N2. After completing the curve from 0% to 100% (1st evaluation-Fig. 5), the temperature was lowered to leave it isothermal and measure the stability of the samples ( Fig. 6). After 14 h on stream, the temperature was decreased and the complete light-off curve (2nd evaluation-Fig. 5) was again measured. Table 1 compares the reaction temperature values for which the catalysts reach 50% conversion of FA. It can be seen that comparing monometallic catalysts, Ni reached the conversion value at a lower temperature (138 °C) than the Cu catalyst. However, the high selectivity to hydrogen production of the Cu catalyst is notable. Moreover, comparison with previous results obtained over a Ni/SiO2 catalyst [13], which gives 50% conversion at 180 °C with selectivities to H2 of 91%, demonstrates the superior performance of Ni catalyst supported on graphite for the hydrogen production from formic acid. A composition equal to 2.5% Ni and 2.5% Cu was chosen for the bimetallic catalysts, due to the catalytic performance of the monometallic catalysts supported on HSAG-400, seeking that the solids have the highest Ni activity and at the same time the high selectivity towards H2 provided by Cu. These results are in agreement with the DFT study of Herron et al. [43] which predicts the formation of CO2 and H2 on Cu(111) and Cu(100) while on Ni(111) and Ni(100) dehydration products CO and H2O are expected. Moreover, experiments have shown the most preferable pathway for HCOOH dissociation on the stepped Ni surface was HCOOH dehydrogenation to give COOH followed by dehydroxylation to form CO [44].In the case of the Ni-Cu bimetallic catalyst, it is observed that the activity and selectivity were intermediate, reaching 50% conversion at 145 °C, but with a selectivity of 97% towards H2. It is worth noting the low concentration of CO (3%) in the reactor outlet stream, which is associated with a parallel dehydration reaction of HCOOH.This work has also addressed the study of doping with alkali metals (Li, Na and K) with an atomic ratio of metal/alkaline equal to 1 to promote the basicity and the catalytic activity. Doping with K and Na shows a marked effect on the temperature at which it reaches 50% conversion of HCOOH, which in both cases was significantly lower than that of the undoped catalyst, which reveals a positive effect on the catalytic activity in these multi-component solids that make them competitive with noble metals [5]. However, the doping with Li worsened the behavior of the material, since only at the temperature of 160 °C is the 50% conversion reached, this temperature being even higher than that of the less active monometallic (Cu catalyst). Therefore, among the alkali metals studied only potassium and sodium gave promotion to the bimetallic Ni-Cu catalyst, the order of the activities measured for samples being K > Na > undoped > Li (Table 1, Fig. 5). This finding is somewhat different to that found over Pd/C catalyst for what all the alkali metal species gave promotion [1]. The doped Ni-Cu catalysts have shown high selectivity values towards hydrogen (95–97%) and the selectivities were maintained at high conversions.The specific catalyst-mass based reaction rates obtained at 100 °C and the turnover frequency (TOF) calculated per surface metal atom are compiled in Table 2. Also, the apparent activation energies calculated from the Arrhenius plots for all the catalysts studied are given.In Table 2 it is seen that the reaction rate and TOF value at 100 °C for the Cu catalyst are lower than (approximately half of) those obtained for the supported Ni, while the Cu-Ni bimetallic gives intermediate values. This order in catalytic activity agrees with that found in an earlier work for Ni, Cu and Ni-Cu alloy powders [10]. It is important to note that the TOF of the Ni-Cu catalyst is significantly increased when Na and specially K is added, since the TOF of the Ni-Cu is doubled for Ni-Cu/K catalyst. Thus, the TOF for the Ni-Cu/K catalyst is 0.0113 s-1 at 100 °C, value of reaction rate considerably higher than those reported for Ni [16] and Ni-Cu [17] supported on nitrogen doped carbon materials. However, it is very close to those reported in the literature for noble metals [1,45]. Jia et al. report a TOF value of 0.013 s-1 at 80 °C for a Pd/C catalyst, although this is in a great extent increased when K is added to the catalyst.Interesting that the values of apparent activation energies for the two monometallic catalysts were similar, and also in the case of the undoped bimetallic. For catalysts modified with Li, Na and K, the apparent activation energy values were lower, the lowest value being obtained for the K-doped catalyst. It should be note that in this work the apparent activation energy values were lower than those reported in an earlier work for the decomposition of formic acid over Ni powder without support and Ni catalysts supported on silica and alumina, as well as for the decomposition of Ni formate [11,46]. Also, for similar catalysts, Ni supported on carbon, the obtained values were higher in the range of 100 ± 10 kJ mol-1 than those calculated for the present catalysts [15,16].On the other hand, the catalysts were relatively stable under the conditions tested, although it can be seen in Fig. 5 that the points corresponding to the 2nd evaluation are below those obtained in the first, probably due to a restructuring of the material at the reached temperature (130–190 °C) and with conversion levels close to 100%.In the case of catalyst doping with Li (Fig. 6) a lower activity is observed in the first hours on stream followed by a slight decay of the conversion for the temperature (145 °C) at which the stability test was carried out for this catalyst. It should be noted the marked positive effect on the conversions achieved doping the catalysts with K and Na in the whole range of temperature. The doping of K in Pd catalysts supported over SiO2, Al2O3 and activated carbon has been previously reported [5]. These authors observed a significant effect of improvement in the catalytic behavior of noble metal for the formic acid decomposition. As a reaction mechanism they proposed, as a first step, the formation of a phase containing liquid formic acid condensed in the pores of the catalyst and this phase provides a reservoir for the formation of formate ions with the participation of K+ ions; that later decompose to form CO2 and H2. In our materials, since the support is a non-porous material, condensation of formic acid is not likely to occur in pores, however, formates or oxyhydroxide phase could form in the alkali metal in the doped catalysts, these species being the reaction intermediates. In addition, the promotion effect of alkali metals increases down in the group (Li < Na < K) when the formation of these species is boosted [47].Temperature programmed surface reaction (TPSR) is a powerful technique to determine the surface chemistry of bulk metal, supported metal and bulk metal oxide on supported catalysts. It can provide both qualitative and quantitative analysis of the surface active sites present on the catalyst surface, the reaction mechanisms, and kinetics occurring on the catalyst surface by using chemical probe molecules such as alcohols, carboxylates, or specific acidic-basic reacting gases. In the present paper, the experiments were carried out by adsorbing the HCOOH molecule (reactive under study) and monitoring the gas outlet flux with a mass spectrometer during desorption at programmed temperature experiments.The TPSR experiments were carried out to understand the differences in the catalytic performance. The catalysts were reduced before experiments in hydrogen flow at 300 or 400 °C and were degassed in a high vacuum at the same temperature for 1 h. The adsorption was carried out using a pulse of 40 Torr of HCOOH at 40 °C. The TPSR experiments for the Ni and Cu based catalysts are shown in Figs. 7 and 8. As above stated the acid formic decomposition may proceed through either dehydrogenation giving CO2 and H2 and dehydration producing H2O and CO. So, the evolution of signals assigned to H2, N2, H2O, CO, HCOOH, CH2O, O2, and CO2 (m/e = 2, 14, 17/18, 28, 29/46, 30, 32 and 44, respectively) was followed as a function of temperature. The evolution of the desorbed masses of H2, CO2 and HCOOH was plotted as a function of temperature to make the comparison clearer in Figures. The small amount of CO (m/e = 28) desorption is not plotted because the mass 28 is also secondary of CO2 which is the major product of decomposition and the determination of real CO production is prone to considerable error.The TPSR experiments for the Ni, Cu and Ni-Cu catalysts are shown in Fig. 7. At lower temperatures (< 100 °C) the desorption of the unreacted HCOOH is observed and above 50 °C the decomposition process begins to produce H2 and CO2. This behavior was similar for the three samples compared in this figure.For the Ni catalyst, in Fig. 7a the H2 and CO2 profiles show in addition to the 80 °C peak a maximum at 125 °C, then an increase mainly in CO2 signal with another maximum at 200 °C. After that, both signals remain at zero until the region of 300–400 °C where the decomposition of the fraction of retained FA in the catalyst occurs. Previous studies using spectroscopic techniques and temperature programmed desorption (TPD) of FA adsorption over Ni(111) surfaces [48] reported the formation of bidentate formates at low temperature which transformed to monodentate formates and decomposed in the 25–225 °C temperature range, producing CO2 at 100 °C and also CO at 157 °C.For the Cu catalyst, Fig. 7b, the HCOOH signal had two steps marked of desorption at 80 and 140 °C. In addition, at that temperature of 140 °C, a maximum of CO2 production is observed and also in the H2 signal at 150 °C. The adsorption and decomposition of formic acid on clean Cu(110) has been previously studied by means of thermal desorption mass spectroscopy [14,49]. On the Cu(110) surface the formate species formed is stable up to 127 °C, but decomposes to simultaneously evolve H2 and CO2 in TDS peaks at 190–200 °C. However, the temperature for HCOOH decomposition on Cu metallic powder has been shown to be lower than on Cu(110) surface [50]. In addition, the reaction is structure sensitive on Cu catalysts since Cu(100) and Cu(211) bind HCOO much more strongly than Cu(111) and have varied barriers for the likely rate determining step, formate species dehydrogenation [51].For the bimetallic catalyst, Fig. 7c, the H2 and CO2 profiles are not the results of adding the profiles obtained for the monometallic catalysts. The low-temperature region is similar to that mentioned for monometallics, however, the CO2 and H2 signals show a maximum at 145 °C and another increase in the region of 160 and 250 °C. For this catalyst, a remarkable coincidence is observed in the profiles of both gaseous products.In Fig. 8, the undoped bimetallic catalyst and the catalysts doped with K, Li and Na are compared. The objective of adding an alkali metal to the Ni-Cu bimetallic catalyst was to promote the catalytic activity by increasing its basicity. These alkali metals could also favor the formation of stable carbonates at high temperatures [14].In the three cases, the alkali/active metal ratio was maintained to compare the effect caused by each one. It is possible to observe that the profiles have similar shapes although with different relative intensities (Fig. 8b–d). In addition, for all the catalysts was observed that at a lower temperature (< 120 °C) desorption of the unreacted HCOOH is observed and above 50 °C the decomposition process begins to produce H2 and CO2. Relative to the non-alkali modified Ni-Cu catalyst the intensity of the CO2 and H2 profiles for the alkali modified catalysts have considerably increased which is related with the promotion by the alkali of the formate species formation [13].The potassium-doped catalyst shows a first maximum of the product signals at 105 °C, this being the lowest temperature observed for all the catalysts under study. This could explain the higher activity observed for this catalyst.The sodium-doped catalyst shows the first maximum at 135 °C where a greater intensity of the CO2 signal and retention of H2 at that temperature was observed. This behavior could be related to the sodium precursor salt (sodium carbonate) employed in this solid.The lithium-doped catalyst shows a first signal at 145 °C of both products. It can be seen, that the Ni and Cu monometallic, undoped bimetallic and Li-doped catalysts had lower adsorption of HCOOH and subsequent lower production of H2 and CO2 (Figs. 7a–c and 8b). This could be related to the lower catalytic activity observed for these catalysts.The H2/CO2 ratio was determined from the HCOOH TPD profiles (Table 1) using a calibration of the relative intensity of the H2 and CO2 signals, m/e equal to 2 and 44, respectively. The monometallic catalysts present values of the H2/CO2 intensity ratio less than 1 (0.43 and 0.49), this could indicate that the fraction of hydrogen not released after decomposition is retained, forming hydride or formate species. For the bimetallic non-doped catalyst, a desorbed equimolecular ratio is observed, indicating that neither H2 nor CO2 are retained on the catalyst. The catalysts doped with K, Na and Li present values of H2/CO2 ratios greater than 1, indicating stronger adsorption of CO2, probably forming carbonate species [14]. If the samples doped with Na and K are compared, it can be seen that in the first one, a greater amount of CO2 (H2/CO2 = 0.5) was desorbed by the decomposition of formic acid at a lower temperature (100–150 °C range, Fig. 8c). For the other catalyst, Ni-Cu/K, the production of H2 and CO2 was equimolar as corresponds to the formic acid decomposition reaction at a lower temperature (100–150 °C range, Fig. 8d). The sample doped with Na presents greater desorption of H2 at a higher temperature. This indicates in the case of Na greater stability of species, for example, formate or bicarbonate type, that store hydrogen in this solid. The presence of formate, bicarbonate and carbonate species in the catalysts used on reaction was confirmed by attenuated total reflectance (ATR) ( Fig. 9).The spectra of the used catalysts show signals assigned to bridged carbonates (ν(C O): 1730–1640 cm-1; νas(COO): 1285–1280 cm-1; νs(COO): 1020–1000 cm-1), bidentate carbonates (ν(C O): 1670–1530 cm-1; νas(COO): 1270–1220 cm-1; νs(COO): 1030–980 cm-1), monodentate carbonates (νas( CO 3 2 − ): 1530–1470 cm-1; νs( CO 3 2 − ): 1370–1300 cm-1; ν(C-O): 1080–1040 cm-1), carbonites (νas(CO2): 1495–1478 cm-1; νs(CO2): 890–717 cm-1) and formates species (νas(COO-): 1605–1540 cm-1; νs(COO-): 1370–1345 cm-1) [44,52–54]. It is important to note that the signals associated with the formate species present a higher relative intensity for the Na-doped catalyst, being consistent with what was observed in the formic acid TPSR experiments. Probably the lower stability of these species in the catalyst doped with potassium is the factor that determines their greater capacity to promote the catalytic activity.The synthesis of Ni-Cu bimetallic catalysts supported on a non-porous high surface area graphite was performed. The catalysts were employed in the decomposition reaction of formic acid in the vapor phase. The effect of doping on support with alkali metals (Li, K and Na) was studied, using in all cases an atomic ratio of alkali metal to active metal (Ni, Cu or Ni-Cu) equal to unity.The bimetallic Ni-Cu catalyst has an intermediate behavior of monometallic, with high catalytic activity for the decomposition of formic acid and high selectivity for the production of hydrogen (97%). The comparative study of the promotion of the Ni-Cu catalyst with the alkali metals Li, Na and K shows that the catalyst doped with K has the best behavior in the formic acid decomposition reaction. TPRS experiments show that formate, bicarbonate or carbonate species decompose at different temperatures depending on the alkali metal present in the catalyst and the formation/decomposition of these species turns out to be an important factor in promoting catalytic activity. The sample doped with Na presents greater desorption of H2 at a higher temperature and a higher relative intensity of the formate species was observed in the used catalyst by ATR. The potassium-doped catalyst shows the maximum production of H2 and CO2 equimolar at 105 °C, being the lowest temperature observed for all the catalysts under study. This could explain the higher activity observed for this catalyst.The bimetallic catalyst doped with K showed 100% conversion of formic acid at 130 °C with a 95% of selectivity to hydrogen. Also, all the tested materials were promising for their application since they showed catalytic behaviors close to those of noble metals reported in the literature. B.M. Faroldi: Conceived and designed the experiments, Investigation, Writing – original draft. J.M. Conesa: Investigation, collaborated with the catalytic and TPRS experiments. A. Guerrero-Ruiz: Writing – review & editing. I. Rodríguez-Ramos: conceived and designed the experiments, Project administration, Writing – review & editing, Funding acquisition, B.F. and I.R.R. conceived and designed the experiments. All authors discussed the results and contributed to the manuscript.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors wish to acknowledge the financial support received from the Spanish Agencia Estatal de Investigación (AEI) and EU (FEDER) (projects CTQ2017-89443-C3-1-R, CTQ2017-89443-C3-3-R, PID2020-119160RB-C21 and PID2020-119160RB-C22). B. Faroldi thanks CONICET for Postdoctoral External Fellowship Program.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.apcata.2021.118419. Supplementary material .

Ni, Cu and Ni-Cu catalysts supported on high surface area graphite were synthesized by incipient wet impregnation. Also, the effect of doping the graphite support with alkali oxides (Li, Na and K) was studied. The catalysts were tested in the formic acid decomposition reaction to produce hydrogen. The bimetallic Ni-Cu catalyst doped with K showed the best catalytic performance with 100% conversion of formic acid at 130 °C and a 95% of selectivity to hydrogen. The turnover frequency (TOF) of the catalysts follows the order: Ni-Cu/K > NiCu/Na > Ni-Cu > Ni-Cu/Li. While the order for the apparent activation energy values is: Ni-Cu > Ni-Cu/Li > Ni-Cu/Na > Ni-Cu/K. The mechanism of the reaction is approached by programmed temperature surface reaction (TPSR) experiments and attenuated total reflectance (ATR). The greater catalytic activity of the Ni-Cu catalyst doped with potassium is ascribed to the lower stability of the formate, bicarbonate and carbonate species on its surface.

World production of waste plastic grows year by year, as a consequence of the huge demand for plastic materials in every commercial field [1]. However, a significant proportion of waste plastics end up into the waste stream leading to many environmental problems. Plastics in the ocean are of increasing concern due to their persistence and effects on the oceans, wildlife and potentially, humans [2]. The cumulative quantity of waste plastic is predicted to be nearly 250 million tonnes per year by 2025. Therefore, there is an urgent need to develop more effective methods to process waste plastics and improve its utilization efficiency.Chemical recycling processes such as pyrolysis are an effective option to recover energy from waste plastic. A wide distribution of products including gas, chemicals, chars and other products can be obtained from pyrolysis of plastics [3,4]. Furthermore, pyrolysis with a subsequent catalytic steam reforming process enables the conversion of plastic into more valuable gases such as hydrogen [5]. A 60 g/h scale continuous tank reactor for plastic pyrolysis followed by the catalytic packed-bed reactor for steam reforming was designed by Park [6] and Namioka [7] for the hydrogen-rich gas production from waste polypropylene and polystyrene, while the optimum operating conditions were also studied. Erkiaga et al. [8] compared the products from pyrolysis (500 °C)-steam reforming of HDPE with those from a gasification (900 °C)-steam reforming system by the same authors [9]. Results show that the former one produced a high H2 yield of 81.5% of the maximum stoichiometric value, which was a little bit lower than the later one (83%) but enabling a more energy-efficient technology for plastics utilization. The pyrolysis and in-line steam reforming of waste plastics has been reviewed by Lopez et al. [10], and they reported that more than 30 wt.% of H2 yield with up to 70 vol.% of concentration could be obtained.Catalysts can assist in chain-scission reactions and the breakage of chemical bonds during the pyrolysis-steam reforming process, allowing the decomposition of plastics to occur at a lower temperature and shorten the reaction time. Different types of catalyst such as olivine [11], Ru [6], Fe [12], and Ni [13,14] catalysts have been investigated for gaseous products from the pyrolysis-reforming of waste plastics. Because of the high activation ability of CC and CH bonds on the Ni metal surface as well as the relatively low cost, Ni based catalysts have been a preferred choice in the process [15]. There has been much reported work in the literature devoted to the selection of the optimum loading content, promoters and preparation method of Ni based catalysts for the pyrolysis-reforming of plastics. By varying the flow rate of reduction gas and metal addition to the Ni catalyst, Mazumder et al. [16] found that the acid–base properties, metal dispersion and crystal size of catalyst can be greatly improved. Wu and Williams [17] suggested that the increase in Ni loading could improve hydrogen production from polypropylene, and Mg modified Ni catalyst showed better coke resistance than the non-modified catalysts. Other promoters such as Ce and Zr were also explored, and the improvement in catalyst intrinsic activity was ascribed to the enhancement of water adsorption/dissociation [18]. In order to obtain a higher Ni dispersion, some novel assisted methods were developed to relieve the diffusion resistance of Ni into the inner structure of catalyst. For example, ethylene glycol [19,20] and ethylenediaminetetraacetic acid [21] assisted impregnation methods were used to prepare Ni catalysts with good stability and activity for hydrocarbons reforming.Catalyst synthesis methods, in particular, the metal loading method, is a curial factor to be considered for catalyst activity. The physical structure and chemical characterizations of the catalyst, including the porosity, reducibility and stability would be closely related to the preparation process [22,23]. Impregnation (or incipient wetness) is the most common method for catalyst preparation, because of the simple procedure and the flexibility to include different catalyst promoters. Co-precipitated Ni catalyst was designed to minimize catalyst deactivation and promote hydrogen production from waste hydrocarbons [24,25]. Meanwhile, sol-gel prepared catalysts have attracted more attention recently [26]. The reinforced impact on Ni dispersion with average size of 20–24 nm was found by using a sol-gel method, leading to superior catalyst activity towards methane reforming [27]. Some reports have compared different catalyst preparation methods, for example, Bibela et al. [28] used a Ni-Ce/Mg-Al catalyst for steam reforming of bio-oil, and found that the wetness impregnated catalyst showed higher carbon conversion than a catalyst prepared via co-precipitation at increasing pH. A sol-gel prepared and promoted Ni/Al2O3 catalyst was reported to benefit the metal-support interaction with better particle size uniformity than an impregnated catalyst [27,29]. Around twice the hydrogen yield was produced from steam reforming of ethanol with a Ni/SiO2 prepared catalyst using a sol-gel method compared with that by an impregnation method [30].It is known that the co-precipitation, impregnation and sol-gel methods have been adopted as suitable metal loading alternatives for catalyst synthesis. However, the published literatures concerning the comparison of Ni catalyst made by these three methods for catalytic thermal processing of waste plastics are limited. Considering this, the aim of this present work was to investigate Ni/Al2O3 catalysts prepared via co-precipitation, impregnation and sol-gel methods for the pyrolysis-steam reforming of waste plastics. The catalyst activity was evaluated in terms of the hydrogen and carbon monoxide production, as well as the catalyst coke formation. In addition, it has been shown that different plastics show different pyrolysis behaviour, producing different product hydrocarbons, which may effect the catalytic steam reforming process and catalyst coke formation and product distributions [31,32]. Therefore, the influence of the type of plastic feedstock on the product selectivity and catalyst activity was also investigated.This work follows on from our previous reports [4,12,33] which investigated the influence of different types of catalyst and process parameters on the pyrolysis catalytic steam reforming of waste plastics in relation to hydrogen productionThree different waste plastics, high density polyethylene (HDPE), polypropylene (PP) and polystyrene (PS), which are the most common plastic wastes worldwide, were supplied by Regain Polymers Limited, Castleford, UK. Plastics were collected from real-word waste plastics and mechanically recycled to produce 2–3 mm spheres. The ultimate analysis of the plastic wastes were determined using a Vario Micro Element Analyser, and the results are shown in Table 1 . The proximate analyses including moisture, volatiles and ash content of waste plastics were conducted according to ASTM standards E790, E897 and E830, respectively. Briefly, the moisture content was determined by placing 1 g of plastic uniformly in a sample boat in an oven at 105 °C for 1 h. The measurement of volatiles content was operated by using a sealed crucible containing 1 g of plastic in an electric furnace at 950 °C for 7 min, while the ash content was obtained by placing 1 g of plastic in a sample boat in air at 550 °C for 1 h. Results were summarized in Table 1. As the plastics used in this work were from real-world applications instead of the pure polymers, some additives may have been present in the samples. For example, oxygen was detected for the elemental analysis, whereas it would not be present in the pure polymer. Waste HDPE was observed to have the highest ash content of 4.98 wt.%, while the other two plastics show little ash content.The Ni/Al2O3 catalysts prepared using co-precipitation method (Ni/Al-Co), impregnation method (Ni/Al-Im) and sol-gel method (Ni/Al-Sg) were tested to catalyse the pyrolysis-reforming of waste plastics. Ni/Al-Im was obtained by a conventional wet impregnation method. 10 g γ-Al2O3 and 5.503 g Ni(NO3)2·6H2O (corresponding to Ni loading of 10 wt.%) were mixed in deionized water. The mixture was then stirred using a magnetic stirring apparatus at 100 °C until it turned into slurry. The precursor was dried overnight and calcined at 750 °C for 3 h. The prepared Ni/Al-Co catalyst involved mixing the metallic nitrates of 7.43 g Ni(NO3)2·6H2O and 99.34 g Al(NO3)3·9H2O (Sigma-Aldrich) together with 150 ml deionized water, so that a 10 wt.% Ni loading was obtained. The solution was kept at 40 °C with moderate stirring, then the precursor was precipitated with NH4(OH) dropwise until the final pH of around 8 was achieved. The precipitates were filtered and washed with deionized water and then dried at 105 °C overnight, followed by calcination at 750 °C in air for 3 h. The Ni/Al-Sg catalyst with the same Ni loading of 10 wt.% was prepared by a simple sol-gel method. 20 g of Aluminium tri-sec-butoxide (ATB, Sigma-Aldrich, 97%) was firstly dissolved into 150 ml absolute ethanol (>99.5%, Merck) and stirred for 2.5 h at 50 °C. 2.210 g of Ni(NO3)2·6H2O was dissolved in 8 ml deionised water separately to form the Ni precursor. Then the Ni solution was pipetted into the support solution while maintaining stirring at 75 °C for 0.5 h. 1 M HNO3 was added into above solution until the pH of 4.8 was obtained. After drying at 105 °C overnight, the precursors were calcined at 450 °C in air for 3 h.All of the catalysts were ground and sieved with a size range between 50 and 212 μm. The catalysts used in this work were reduced in 5 vol.% H2 (balanced with N2) atmosphere at 800 °C for 1 h before each experiment.A schematic diagram of the pyrolysis-catalytic steam reforming reactor system for waste plastics is shown in Fig. 1 [33]. The experimental system consisted essentially of a continuous steam injection system using a water syringe pump, a nitrogen gas supply system, a two-stage stainless tube reactor, a gaseous product condensing system using dry ice, and gas measurement system. The reactor has two separate heating zones, i.e. first stage plastic pyrolysis reactor of 200 mm height and 40 mm i.d; second stage catalytic reactor of 300 mm height and 22 mm i.d. The real temperatures of two zones were monitored by thermocouples placing in the middle of each reactor and controlled separately. The calibration of the reactor temperature was performed before this set of experiments, and the temperature described in this paper was given as the real one. For each experiment, 0.5 g of catalyst was loaded into the second stage where the temperature was maintained at 800 °C. High purity nitrogen was supplied as the inert carrier gas. 1 g of plastics were placed in the first stage and then heated from room temperature to 500 °C at 40 °C min−1, and the evolved volatiles passed into the catalyst reactor for reforming. Water was injected into the second stage with a flow rate of 6 g h−1. After the reforming process, the condensable liquids were collected into condensers while the non-condensable gases were collected into a 25 l Tedlar™ gas sample bag off-line gas chromatography (GC) measurement. Each experiment was repeated to ensure the reliability of the results.The gas products were separated and quantified by packed column GCs. A Varian 3380 GC packed with 60–80 mesh molecular sieve, coupled with thermal conductivity detector (TCD) was used to analyse permanent gases (H2, O2, N2, CO). CO2 was determined by another Varian 3380 GC/TCD. Argon was used as the carrier gas for both GCs. Hydrocarbons (C1 to C4) were analysed using a different Varian 3380 GC/FID coupled with a HayeSep 80–100 mesh molecular sieve column and using nitrogen as carrier gas. Each gas compound mass yield was calculated combining the flow rate of nitrogen and its composition obtained from the GC.The yield of non-reacted pyrolysis oil was calculated as the mass difference between fresh and used condenser system in relation to the total weight of plastic and steam input. Coke yield was determined from the temperature programmed oxidation analysis of the spent catalyst. Residue yield was measured as the mass difference between fresh and the used whole reactor system in relation to the total weight of plastic and steam input. Mass balance was therefore calculated as the sum of gas, liquid and residue obtained in relation to the total plastic and steam input.X-ray diffraction (XRD) analysis of the fresh catalysts was carried out using a Bruker D8 instrument with Cu Kα radiation operated at 40 kV and 40 mA. In order to explore the distribution of active sites on the catalysts, the Debye-Scherrer equation was used to obtain the average crystal size from the XRD results. The porous properties of the fresh catalysts were determined using a Nova 2200e instrument. Around 0.2 g of each sample was degassed at 300 °C for 2 h prior to the analysis. The specific surface area was calculated using Brunauer, Emmett and Teller (BET) method. The total pore volume was determined at a relative pressure P/P0 of 0.99, and the pore distribution was obtained from the desorption isotherms via the BJH method. In order to determine the actual loading of nickel in the catalyst, a Optima 5300DV (Perkin Elmer Inc.) inductively coupled plasma optical emission spectrometer(ICP-OES)was used. About 25 mg of catalyst was previously dissolved in acidic solution, followed by diluting with deionized water to 50 ml in preparation for analysis.The morphologies of the fresh prepared catalysts and the coke deposited on the used catalysts were investigated using a Hitachi SU8230 scanning electron microscope (SEM), which was operated at 2 kV and working distance of 3 mm. An energy dispersive X-ray spectroscope (EDXS) was connected to the SEM to study the elemental distribution. A FEI Helios G4 CX Dual Beam SEM with precise focused ion beam (FIB) was used to analyse the cross-section of the prepared catalysts. Before the analysis, the catalyst was coated with platinum in order to protect the sample during the sectioning process. Fresh catalysts were further examined at a higher magnification by a high-resolution transmission electron microscope (TEM, FEI Tecnai TF20) coupled with a connected EDXS for microstructure and elemental distribution. For the TEM analysis preparation, samples were initially dispersed well in methanol using an ultrasonic apparatus, and were pipetted on to a carbon film coated copper grid. The coke deposited on the surface of catalyst was characterized by temperature programmed oxidation (TPO) with a Shimadzu TGA 50. For each TPO analysis, around 25 mg of spent catalyst was heated from room temperature to 800 °C in an air atmosphere (100 ml min−1) at a heating rate of 15 °C min−1 and a holding time of 10 min at 800 °C.The XRD patterns of the fresh catalysts are shown in Fig. 2 . The Ni/Al2O3 catalyst prepared by the impregnation method produced sharp peaks compared to the other fresh catalysts. The easily identified peaks centred at 2θ = 44.5, 51.9 and 76.4° corresponding to the (111), (200) and (220) plane respectively, confirmed the presence of Ni (JCPDS: 01-087-0712) in the cubic form. The aluminium oxides at 37.6, 45.8, 66.8° were also determined (00-029-0063). As there was no NiO detected from the XRD results, it demonstrates that the nickel catalyst precursors had been completely reduced into active compounds (Ni) before each experiment. According to the Scherrer equation, the average crystallite size of Ni based on the main peak at around 2θ at 44.5° was determined to be 26.17, 52.28 and 19.69 nm for the Ni/Al-Co, Ni/Al-Im and Ni/Al-Sg catalysts, respectively. This indicates that a higher Ni dispersion and smaller Ni particles were found for the catalyst prepared by the sol-gel method compared with impregnation and co-precipitation. Table 2 summarizes the BET surface areas and pore size properties of the fresh nickel Ni/Al2O3 catalysts. The Ni/Al-Co and Ni/Al-Im catalyst showed surface areas of 192.24 and 146.41 m2 g−1, respectively. The Ni catalyst produced via the sol-gel method showed a higher surface area of 305.21 m2  g−1 compared to the catalysts obtained by impregnation or co-precipitation. The Ni/Al-Sg catalyst also gave the highest pore volume of 0.915 ml g−1 while Ni/Al-Im generated the lowest. However, the average pore size of these three catalysts were similar, at around 6.6 nm. Therefore, it indicates that the Ni catalyst prepared by the sol-gel method gives a more porous structure compared to the other two methods. The adsorption/desorption isotherms and pore size distribution of fresh catalysts are shown in Fig. 3 . All of the physisorption isotherm types for the three catalysts appear to be type IV according to the IUPAC classification [34]. From the pore size distributions, the Ni/Al2O3 catalyst prepared by the sol-gel method shows a quite narrow pore size distribution, while the impregnated prepared catalyst shows a broad distribution. This indicates that compared with Ni/Al-Co and Ni/Al-Im catalyst, the Ni/Al-Sg catalyst produces a more uniform porous structure, and most pores are with a size of around 6.64 nm. Therefore, it may be concluded that a mesostructured Ni/Al2O3 catalyst can be obtained by the sol-gel preparation method. The results of the real nickel loading from ICP-OES analysis was also listed in Table 2. It can be seen the real content of Ni in the co-precipitated and sol-gel catalyst was a little lower than the designed value, while it was excellent agreement for the impregnated Ni/Al-Im catalyst. In summary, the active Ni sites were successfully loaded into the catalyst by different preparation method.The morphologies and the distribution of active metallic Ni for the fresh catalysts were determined by SEM-EDX analysis, as shown in Fig. 4 . Compared with the Ni/Al-Co catalyst shown in Fig. 4, which shows a flat surface, the catalyst particles of Ni/Al-Im observed were irregular. The nickel catalyst prepared by the sol-gel method seems to be composed of many small particles in a loose structure. The Ni EDX mapping showed a uniform distribution of Ni particles in the catalysts. In order to investigate the inner structure of the fresh catalysts, the cross-sectional morphologies of catalyst particles were examined by FIB/SEM. From Fig. 5 , the Ni/Al-Co catalyst which has a relatively low surface area (Table 2) shows a tight structure, whereas the Ni/Al-Sg catalyst shows a porous inner structure. The observations agree well with the porosity results that show the Ni/Al-Sg catalyst generates a higher surface area and higher pore volume compared with the other catalysts. This type of structure was reported to benefit Ni penetration inside the catalyst particles, and further promote the catalyst activity [33].The fresh catalysts were further examined under high magnification by TEM and the results are shown in Fig. 6 . The images show obvious dark spots, which were ascribed to the presence of metallic Ni. As can be seen, all the Ni particles were well dispersed, and hardly any agglomeration was seen. Statistical analysis of the Ni particle size distributions of the three TEM images was carried out by ImageJ software, and the results are shown in Fig. 6(d)–(f). More than 95 percent of the Ni particles present were of a size less than 50 nm. The Ni/Al catalyst prepared by the sol-gel method showed the narrowest size distribution, with the smallest average particle size of 15.40 nm. Both Ni/Al-Co and Ni/Al-Im catalysts have ta size distribution concentrated at 15∼30 nm, but they show larger average particle size of 28.91 and 29.60 nm, respectively. Therefore, the sol-gel prepared catalyst exhibited the highest homogeneity and smallest active metal size among the three catalysts, which is in good agreement with the results from XRD analysis. The EDX mappings of the sol-gel synthesized Ni/Al catalyst shown in Fig. 6(g) also demonstrate that both the Ni and Al were uniformly distributed inside the catalyst.The use of different nickel catalysts prepared via co-precipitation, impregnation and sol-gel methods for the pyrolysis-catalytic steam reforming of waste polyethylene was investigated in this section. The results of syngas production and gas composition are summarized in Table 3 . The mass balance of all the experiments in this paper were calculated to be in the range of 92 to 98 wt.%. In addition, results from the repeated trials show that the standard deviations of the hydrogen and carbon monoxide yield were 0.26 and 0.29 mmol g−1 plastic respectively. For the volumetric gas concentrations, the standard deviation was 0.26% for H2 and 0.08% for CO respectively. These data indicates the reliability of the experimental procedure. From Table 3, the highest hydrogen yield of 60.26 mmol g-1 plastic was obtained with the Ni/Al catalyst prepared by the sol-gel method, followed by that prepared by the impregnation method. The lowest hydrogen yield of 43.07 mmol -1 plastic was obtained with the Ni/Al-Co catalyst. The production of carbon monoxide has the same trend as that of hydrogen yield. Syngas production achieved its maximum with the Ni/Al-Sg catalyst, that is, per unit mass of the polyethylene can yield 83.28 mmol of syngas. The gas composition is also shown in Table 3. It can be observed that the concentration of H2 and CO2 were steadily increased with the catalyst order: Ni/Al-Co < Ni/Al-Im < Ni/Al-Sg, while the content of CH4, CO and C2-C4 were decreased correspondingly. During the pyrolysis-reforming of waste plastic, the thermal decomposition of plastic occurs in the pyrolysis stage as Eq. (1). The pyrolysis volatiles were then steam reformed by the catalyst to produce more valuable gases like hydrogen and carbon monoxide (Eqs. (2) and (3)). As the CH4 and C2–C4 concentrations from Ni/Al-Sg were rather lower, while H2 and CO yields were significantly higher than those from the other two catalysts, it can be concluded that the steam reforming of hydrocarbons (Eq. (2)) was greatly promoted in the presence of the Ni/Al-Sg catalyst. In addition, the ratio of H2 to CO, which can reveal the degree of waster gas shift reaction Eq. (3), achieved its maximum of 2.62 with the Ni/Al-Sg catalyst. Therefore, the nickel catalyst prepared by sol-gel method displayed the highest activity to both hydrocarbons reforming and water gas shift reactions among the three catalysts investigated. (1) CxHyOz → (CH4 + H2 + C2-4 + CO + …) + Tar + Char (2) CxHy + H2O → CO + H2 (3) CO + H2O ⇔ CO2 + H2 Temperature programmed oxidation was used to investigate the coke deposition on the used catalyst. As shown in Fig. 7 , the oxidation process involved three main stages: the removal of water in the range of 100∼300 °C, the oxidation of Ni from 300 to 450 °C, and carbonaceous coke combustion from 450 °C onwards, which were also observed in our previous studies [35]. The amount of coke was calculated based on the weight loss of spent catalyst from 450 °C (when Ni had finished oxidization and coke started to combust) to 800 °C, and the results are shown in Table 3. It can be observed that, during the pyrolysis-catalytic steam reforming of waste polyethylene, the Ni/Al-Sg catalyst produced the highest coke yield of 7.41 wt.% among the three catalysts, but displayed the highest catalyst activity for syngas production. In addition, the Ni/Al-Co catalyst which generated the lowest hydrogen yield produced the least coke formation. It should be noted that the catalytic volatiles thermal cracking Eq. (4) may also be involved during this process. The results regarding syngas production and coke yield with the three catalysts indicate that the Ni/Al-Sg catalyst showed high catalytic activity for both reforming reactions and volatiles thermal cracking reactions. The derivative weight loss thermograms in Fig. 7 showed two distinct peaks at temperatures around 530 and 650 °C. It has been reported that the oxidation peak at lower temperature was related to amorphous coke, while the peak at higher temperature is linked to graphitic filamentous coke oxidation [12,36]. The coke deposited on the Ni/Al-Sg catalyst appears to be mainly in the form of filamentous carbon, which was also confirmed in the SEM morphology analysis shown in Fig. 8 (c). While for the Ni/Al-Co catalyst SEM results shown in Fig. 8(a), more coke deposits without any regular shapes were observed. The larger production of amorphous coke on the spent Ni/Al-Co catalyst compared with the other two catalysts could also be responsible for the lower syngas production, since the amorphous coke was considered to be more detrimental to catalyst activity than the filamentous carbons. In addition, compared with the SEM results of the fresh catalysts shown in Fig. 4, the morphologies of the three nickel catalysts did not change significantly. For example, the catalyst prepared by the sol-gel method, maintained its loose structure after the reforming process, indicating the good thermal stability of the catalyst. (4) CxHy → C + H2 Polypropylene was also investigated for the pyrolysis-catalytic steam reforming process for hydrogen production in the presence of the three different Ni/Al catalysts to produce more gases. The gas productions and concentrations are shown in Table 4 . The Ni/Al-Sg catalyst displayed the most efficient catalytic activity in terms of the steam reforming of the polypropylene, as the gas yield was 144.03 wt.% which was much higher than the other two catalysts. In addition, much higher hydrogen yield (67.00 mmol g−1 plastic) and carbon monoxide yield (29.98 mmol g−1 plastic) were obtained by using the Ni/Al-Sg catalyst. The syngas production from PP with the Ni/Al-Sg catalyst was slightly higher than was observed from HDPE, and this phenomenon can also be found with the Ni/Al-Co and Ni/Al-Im catalysts. This may be due to the higher hydrogen and carbon content and lower ash content of PP compared with HDPE (Table 1), and which suggests more effective hydrocarbons participation in the reforming reactions to obtain more syngas. The nickel catalyst prepared by the co-precipitation method showed the least activity for the reforming process, producing 46.05 mmol H2 g−1 plastic and 20.39 mmol CO g−1 plastic. The composition of the gases from waste polypropylene were mainly composed of H2, CH4, CO, C2-4 hydrocarbons and CO2. The concentration of H2 and CO with Ni/Al-Sg achieved 59.38 and 26.57 vol.%, respectively. The CH4 and C2-4 gases content with the Ni/Al-Sg catalyst were lower than the case with the other two catalysts, which also indicates the higher catalytic activity of the catalyst made by the sol-gel method.The amount and the type of coke deposition on the three catalysts from the pyrolysis-steam reforming of polypropylene were determined by TPO analysis, as shown in Fig. 9 . The amount of carbonaceous coke was calculated and the results are shown in Table 4. The Ni/Al-Co and Ni/Al-Im catalysts produced around 5 to 6 wt.% of coke, lower than the case of the Ni/Al-Sg catalyst which showed a 8.49 wt.% coke yield. From the derivative weight loss results, the peak associated with amorphous carbon was much larger than that of the filamentous carbon with the Ni/Al-Co catalyst. However, for the Ni/Al-Im and Ni/Al-Sg catalysts, produced more filamentous carbons. This phenomenon was also observed with HDPE. It can be deduced that the nickel catalyst prepared by impregnation and sol-gel methods favour the production of filamentous carbonaceous coke from the pyrolysis-steam reforming of waste plastics. The presence of both amorphous carbon and filamentous carbon with the Ni/Al-Co and Ni/Al-Im catalyst were further confirmed by the SEM images shown in Fig. 10 (a) and (b). Fig. 10(c) shows that the deposits on the used Ni/Al-Sg catalyst were predominantly filamentous carbon. Furthermore, there was a dense covering of carbon on the catalyst no matter which type of catalyst was used. The amount of carbon deposits from SEM images seems to be larger for PP than the case for HDPE, which was consistent with the TPO results.Wu and Williams [13] used an incipient wetness method (similar to the impregnation method in this work) prepared Ni/Al2O3 for steam gasification of PP. A potential H2 yield of 26.7 wt.% was obtained, with the gaseous product containing 56.3 vol.% of H2 and 20.0 vol.% of CO. The syngas production can be calculated as 77.62 mmol g−1 plastic, which was close to the yield obtained in this study (75.77 mmol g−1 plastic). However, the coke deposition (11.2 wt.%) was higher it from this study (5.34 wt.%), which might be due to the lower surface area (90 m2/g) compared with Ni/Al-Im used here (146.41 m2/g). High hydrogen yields of 21.9 g g−1 PP and 52 wt.% (potential value) were obtained by the same authors from polypropylene with co-precipitation prepared Ni-Mg-Al (Ni:Mg:Al ratio = 1:1:1, 800 °C, 4.74 g h−1 steam) [37] and co-impregnated Ni/CeO2/Al2O3 catalyst (Ni 10 wt.%, CeO2 20 wt.%, 900 °C) [14], respectively. It should be noted that the Ni catalysts in these literatures were prepared either at a higher loading or with promoter added, otherwise using at higher catalysis temperature. It also suggests that hydrogen production can be promoted by the use of effective catalyst promoters or by regulation of operational parameters. Czernik and French [38] concluded that many common plastics can be converted into hydrogen by thermo-catalytic process with a microscale reactor interfaced with molecular beam mass spectrometer. A bench-scale plastic pyrolysis-reforming system was also carried out by them using PP as a representative polymer, while 20.5 g/h H2 was generated with a 60 g/h of PP feeding rate.The product distributions in terms of gas yield and composition from the pyrolysis-catalytic steam reforming of waste polystyrene with the different catalysts are displayed in Table 5 . The Ni/Al-Co catalyst produced a H2 yield of 51.31 mmol g−1 plastic which was a little lower than the yield of 55.04 mmol H2 g−1 plastic with the Ni/Al-Im catalyst. Among the three catalysts, the Ni/Al-Sg catalyst produced the maximum H2 yield and CO yield per mass of plastic feedstock, which was also observed with HDPE and PP. However, compared with HDPE and PP, PS shows a comparatively higher yield of CO, with values up to 36.10 mmol g−1 plastic with the sol-gel prepared catalyst. Most of the concentrations of CO and CO2 obtained from PS were also larger than the corresponding data from PP or HDPE. It may due to the higher content of elemental carbon in the feedstock. In addition, as the coke yield produced using PS was lower than from PP or HDPE except from those with Ni/Al-Co, it suggests most of the carbon in PS was converted into gas product by participating in catalytic steam reforming reactions Eq. (2) or the water gas shift reaction Eq. (3). The hydrogen content of the product gases fluctuated slightly in the range of 57.90 and 59.32 vol.% depending on the different catalyst applied. The hydrocarbons in the final gas product were relatively low for PS whichever type of catalyst was used, and the concentration of C2-C4 was less than 1.10 vol.%. The water gas shift reaction Eq. (3) was promoted by the Ni/Al-Co catalyst as the H2/CO ratio was higher than with the other catalysts.TPO analysis of the used catalysts from the pyrolysis-catalytic steam reforming of polystyrene was also carried out to characterize the carbonaceous coke deposition on the catalyst, as shown in Fig. 11 . The results of the calculated amount of coke produced are shown in Table 5. The Ni/Al-Sg catalyst produced the highest coke yield of 6.14 wt.% even though it produced the largest hydrogen production amongst the three catalysts. It suggests that both steam reforming and decomposition of hydrocarbons Eq. (3) and Eq. (4) were significantly facilitated with the Ni/Al-Sg catalyst during the pyrolysis-steam reforming of waste polystyrene. As for the type of carbon deposits, overlapping derivative weight loss peaks were observed with the Ni/Al-Im and Ni/Al-Sg catalysts, indicating that both amorphous and filamentous coke were produced. This is in agreement with the morphologies observed by SEM images shown in Fig. 12 . The Ni/Al catalyst prepared by co-precipitation displayed the deposits in a great proportion of amorphous form, and the derivative TPO peak at lower temperature was more significant than that at higher oxidation temperature.The yield of hydrogen and carbon monoxide from pyrolysis-steam reforming of waste plastics varied with the catalyst preparation method used. Overall, despite the difference in the feedstock, the sol-gel prepared nickel catalyst produced the highest syngas production, while the co-precipitation prepared catalyst produced the lowest syngas production among the three catalysts investigated. In addition, the maximum carbonaceous coke deposition on the catalyst was also obtained with the Ni/Al-Sg catalyst. This suggests that both the hydrocarbon reforming reactions and the hydrocarbon thermal decomposition reactions were promoted more in the presence of the sol-gel prepared catalyst. For example, the largest production of H2 and CO was obtained with the Ni/Al-Sg catalyst with waste polypropylene at 67.00 mmol H2 g−1 catalyst and 29.98 mmol CO g−1 catalyst and also the highest coke yield of 8.49 wt.%. This is in agreement with previous results from Efika et al. [39] that a sol–gel prepared NiO/SiO2 catalyst generated higher syngas yield than the catalyst made by an incipient wetness method, and the former one also appeared to have more carbon formation on its surface. Although the syngas production achieved the maximum at the presence of Ni/Al-Sg, it should still be noted that the CO content was relatively high. It may be related to the high reforming temperature which was unfavourable to the Reaction (3) due to the exothermic nature of the reaction [40]. Furthermore, a dual functional Ni catalyst with both catalysis and CO2 sorption, for example, a sol-gel prepared Ni/Al catalyst coupled with CaO, was suggested for further study, in order to promote the WGS reaction for higher H2 yield [37,41].The catalytic performance in terms of hydrogen yield and CO production was also influenced by physicochemical characteristics e.g. the porosity, and the type of coke deposited. In particular, the increase in the surface area and pore volume could not only improve the dispersion of metal ions, but also facilitate the interaction of reactant molecules with the catalyst internal surface [42]. In addition, the catalyst was generally deactivated by two types of carbonaceous coke, amorphous (or monoatomic) and filamentous carbon. The filamentous carbon was found to have little influence on the catalytic activities, while the amorphous carbon has been reported to be more detrimental to catalyst activity [13]. Furthermore, these two factors are associated with each other, as Li et al. [43] have suggested that the catalyst activity can be improved by uniform Ni dispersion, while uneven distribution and large Ni particles are the main reason for the formation of non-filamentous coke which leads to the loss of catalyst activity.In this work, the sol-gel prepared Ni catalyst showed a high surface area and uniform Ni dispersion, as evidenced from the BET and TEM results. Furthermore, the coke obtained is in the filamentous form (from TPO and SEM results). Therefore, the Ni/Al-Sg prepared catalyst presents an excellent catalytic performance towards syngas production for the pyrolysis-catalytic steam reforming of waste plastics. However, for the co-precipitation prepared Ni catalyst, the catalyst coke deposits were found to be of the monoatomic or amorphous type, though it showed a higher surface area than the Ni/Al-Im catalyst. It suggests that a high activity at the initial reaction stage may occur, but it experienced a rapid deactivation by detrimental coke deposition.The hydrogen and carbon monoxide yield from the pyrolysis-catalytic steam reforming of PP was higher than that observed for HDPE, no matter which catalyst was applied, indicating more syngas production can be obtained per mass of PP compared to HDPE in this work. This may be due to the fact that PP had higher H and C elemental contents compared to HDPE (Table 1), while the ash content of HDPE was relatively higher. In addition, PS was found to produce the highest CO and syngas yield among the three plastics. Barbarias et al. [44] investigated the valorisation of PP, PE, PS and PET for hydrogen production by pyrolysis-catalytic steam reforming. The pyrolysis volatiles at 500 °C were identified, and results show that nearly 100 wt.% of PS was converted into volatiles with 70.6 wt.% of styrene, while more than 65 wt.% of wax were obtained from polyolefins. Therefore, the higher syngas production from PS in this study may due to the fact that more styrene from PS pyrolysis instead of the wax from polyolefins were introduced into the steam reforming stage. They concluded that the H2 yields from PS was lower than those from polyolefins, while the H2 production from PS in this study was comparative even higher than those from HDPE and PP. Around 38, 35 and 30 wt.% of hydrogen yield were achieved by the same authors [44,45] at 16.7 gcat min g−1 plastic of space time, 700 °C from HDPE, PP and PS respectively. The difference between those values and the yield in this work were attributed to the different reactor system as well as the operational parameters.However, in this study, it is still difficult to evaluate the ability of each plastic for H2 and CO production in relation to C or H elemental content. Therefore, CO conversion Eq. (5) and H2 conversion Eq. (6) were calculated, to reveal the degree of C or H in the gas product. These two indicators essentially reflect the reforming ability of each plastic by catalyst towards H2 or CO. In addition, the coke conversion was also calculated as Eq. (7). (5) CO conversion (wt.%) = (C content in CO gas) / (C content in raw plastic) (6) H2 conversion (wt.%) = (H content in H2 gas) / (H content in raw plastic) (7) Coke conversion (wt.%) = (Coke yield per unit mass of plastic)/ (C content in raw plastic) The results of these indicators with the Ni/Al-Sg catalyst was taken as an example in relation to the different plastics and the results presented in Fig. 13 (the results with the other two catalysts are not presented here, but they show a similar trend). From Fig. 13, the ability for H2 production of HDPE and PP was rather close, as the H2 conversion obtained was around 95 wt.%. However, the H2 conversion was significantly increased in the presence of PS, with the highest conversion of 145.11 wt.%. The conversion of over 100 percent was due to the production of H2 from H2O. The CO conversion was gradually increased with the order: HDPE < PP < PS and suggests that the steam reforming reaction of hydrocarbons (Eq. (2)) was more favourable with PS, generating more H2 and CO. The maximum syngas production of 98.36 mmol g−1 plastic was obtained using PS with the Ni/Al-Sg catalyst. In regard to the gas compositions from different plastics, the molar ratio of H2/CO achieved was in the range of 1.72 to 2.62, and it was relatively higher from the polyolefin plastics. Therefore, there should be potential in industrial applications in that the H2 to CO ratio can be tuned to meet the desired ratio by adjusting the mixed proportion of different plastics.From the TPO results related to catalyst carbon coke deposition, PP generated the highest coke yield of 8.48 wt.%, but from Fig. 13, it can be seen that the calculated carbon conversion was in order of PP>HDPE > PS, even though PS has more C content in the feedstock. The results suggest that the coke formation by decomposition of hydrocarbons Eq. (4) was more favourable in the presence of PP. Wu and Williams [37] also found that PP generated the highest coke deposition on used Ni catalysts when the catalyst temperature was 800 °C with a water flow rate of 4.74 g h−1, compared with HDPE and PS. Also, PS produced relatively lower coke yield among the three plastics under variable process conditions. A similar trend was also reported by Acomb et al. [46], when exploring the pyrolysis-gasification of LDPE, PP and PS, as higher residue yields were obtained from LDPE and PP. Furthermore, the reforming temperature in this work was 800 °C, and Namioka et al. [7] also found that the coke deposition of PP was more apparent than that of PS at higher reforming temperatures (>903 K).In relation to the type of carbon deposition on the catalyst, the results show that the carbon was mainly in the form of the filamentous type from waste HDPE and PP (Fig. 7) and Fig. 9), while more amorphous carbon was produced from PS (Fig. 11). This phenomenon was especially evident for the Ni/Al-Im and Ni/Al-Sg catalysts, which generated both types of carbon. This can be explained by the difference in the gas composition, as Angeli et al. [47] suggested that the increase in the C-number in the mixed gases favoured the formation of filamentous carbon, and Ochoa et al. [48] reported that the carbonization of adsorbed coke to form multi-walled filamentous carbon can be promoted by the reaction of CH4 dehydrogenation. In this work, HDPE and PP produced a higher content of C1-C4 hydrocarbon gases compared with PS (comparing Tables 3–5), resulting in the two polyolefin plastics producing more filamentous carbon on the used catalyst.The Ni/Al catalyst prepared by the sol-gel method generated higher H2 and CO yields from waste plastics than the catalysts prepared by co-precipitation and impregnation, due to the higher surface area and fine nickel particle size with uniform dispersion.The Ni/Al-Co catalyst prepared by co-precipitation produced the least syngas yield among three catalyst preparation methods investigated. From the TPO results, the type of carbon deposited on the Ni/Al-Co catalyst was mainly amorphous type carbon while it was in filamentous form for the impregnation (Ni/Al-Im) and sol-gel (Ni/Al-Sg) prepared catalysts.Thermal decomposition reactions were more favoured with olefin type plastics (HDPE and PP) to produce higher hydrogen and coke, whereas the steam reforming reactions were more significant with polystyrene. The maximum H2 yield of 67.00 mmol g1 plastic was obtained from pyrolysis-catalytic steam reforming of waste polypropylene with more hydrocarbons in the product gases, while waste polystyrene generated the highest syngas yield of 98.36 mmol g1 plastic with more oxygen-containing gases in the produced gases.The authors wish to express their sincere thanks for the financial support from the National Natural Science Foundation of China (51622604) and the Foundation of State Key Laboratory of Coal Combustion (FSKLCCB1805). The experiment was also assisted by the Analysis Laboratory in the School of Chemical and Process Engineering at the University of Leeds and Analytical and Testing Center in Huazhong University of Science & Technology (Wuhan, China). This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowski-Curie grant agreement No. 643322 (FLEXI-PYROCAT).

Three Ni/Al2O3 catalysts prepared by co-precipitation, impregnation and sol-gel methods were investigated for the pyrolysis-steam reforming of waste plastics. The influence of Ni loading method on the physicochemical properties and the catalytic activity towards hydrogen and carbon monoxide production were studied. Three different plastic feedstocks were used, high density polyethylene (HDPE), polypropylene (PP) and polystyrene (PS), and compared in relation to syngas production. Results showed that the overall performance of the Ni catalyst prepared by different synthesis method was found to be correlated with the porosity, metal dispersion and the type of coke deposits on the catalyst. The porosity of the catalyst and Ni dispersion were significantly improved using the sol-gel method, producing a catalyst surface area of 305.21 m2/g and average Ni particle size of 15.40 nm, leading to the highest activity among the three catalysts investigated. The least effective catalytic performance was found with the co-precipitation prepared catalyst which was due to the uniform Ni dispersion and the amorphous coke deposits on the catalyst. In regarding to the type of plastic, polypropylene experienced more decomposition reactions at the conditions investigated, resulting in higher hydrogen and coke yield. However, the catalytic steam reforming ability was more evident with polystyrene, producing more hydrogen from the feedstock and converting more carbon into carbon monoxide gases. Overall the maximum syngas production was achieved from polystyrene in the presence of the sol-gel prepared Ni/Al2O3 catalyst, with production of 62.26 mmol H2 g−1 plastic and 36.10 mmol CO g−1 plastic.

No data was used for the research described in the article.Greenhouse gases such as CO2 and CH4 absorb and emit radiant energy, causing global warming [1]. The concentration of CO2 in the atmosphere has been rising despite nature’s effort to curb it via the carbon cycle. Human activities have substantially contributed to this, with a 45 % increase since the age of the industrial revolution, from 280 ppm to 419 ppm in 2022 [2]. In addition, CH4 emissions resulting from processing oil and gas extraction and agriculture increase. It will become increasingly imperative to convert or eliminate these greenhouse gases to lower their atmospheric concentrations.The catalytic dry reforming of methane is a promising technology that could utilize these greenhouse gases. It combines these two molecules with a catalyst to produce synthesis gas (H2 and CO): (1) C H 4 + C O 2 ↔ 2 H 2 + 2 C O Δ H 298 K = 247 k J / m o l Thus, CH4 and CO2 emissions could potentially be reduced to some extent [3–6]. Alternatively, synthesis gas is a crucial feedstock for producing chemicals and fuels via Fischer-Tropsch synthesis, methanol, and dimethyl ether [7,8].DRM is accompanied by secondary reactions such as the reverse water–gas shift reaction, methane decomposition and CO disproportionation: (2) CO 2 + H 2 ↔ H 2 O + C O Δ H 298 K = 41 k J / m o l (3) C H 4 ↔ C + 2 H 2 Δ H 298 K = 75 k J / m o l (4) 2 C O ↔ C + C O 2 Δ H 298 K = - 172 k J / m o l Catalyst selection must consider the endothermic nature of the reaction, which necessitates a high temperature and may result in the formation of carbon deposits according to equations 3 and 4 [9,10]. Many researchers tried to improve the reaction parameters such as temperature, WHSV, CH4 and CO2 ratio to enhance the DRM efficiency. Although high temperature is favorable for endothermic reactions, it may cause metal agglomeration and sintering, while carbon deposits block active metal sites, resulting in the deactivation of the catalyst [11]. Appropriate choice of the active metal, support, promoter, structure and methods for preparation and activation are considered as tools to enhance the activity of DRM [12]. Strong metal-support interaction is considered one of the key properties to provide a stable and active catalyst by proposing high surface area, highly dispersed small active metal particles [12,13]. Several nobel and transition metals (e.g., Pt, Ru, Rh, Co and Ni) have been evaluated as active phases in DRM [14,15]. Except for Ni, which is inexpensive and has excellent catalytic performance [16], their promising results are constrained by their high costs. However, Ni is also rendered inactive by sintering and coking. Researchers have proposed numerous methods for improving the performance of Ni-based catalysts. Synthesis methods were shown to improve acidity and alkalinity and redox character properties of Ni active metal sites for DRM [17]. Ni supported Ce catalysts prepared using three different methods: microemulsion, sol–gel and auto-combustion. The last showed the best catalytic performance in DRM due to the presence of small monoclinic domains of NiO dispersed and strongly stabilized on the partially reduced ceria surface [18–20]. On the other hand, preparing Ni on CeO2 with the precipitation method enhanced the gasification of coke due to the contribution of oxygen in CeO2 support and prevented Ni sintering due to metal-support interaction (MSI) [21]. Furthermore, other factors can play a role: i) improving the active metal dispersion by suitable high-surface-area supports [22–24] or ii) using supports such as SiO2-Al2O3, ZrO2-Al2O3, MgAl2O4 [25], iii) the use of promoters to increase the MSI effect [26], and iv) the introduction of active sites capable of activating CO2 for carbon removal via the reverse Boudouard reaction (back reaction to eq. (4)) [27].It is well known that the rate of CO2 reforming of methane depends on the dispersion and nature of metal clusters. Regardless of thermodynamic and transport effects, the carbon formation rate increases as Ni crystal diameter increases [28]. Herein, we will focus on the use of mesoporous supports as interesting results have been reported with their application [29–31]. Taking advantage of confinement in such well-defined pore systems could be an alternative strategy for stabilizing small Ni particles.Mesoporous silica supports, such as SBA-15 and SBA-16, are excellent candidates for grafting Ni particles due to their uniform pore diameters, which could inhibit Ni particle agglomeration. Despite being used at elevated temperatures, the thick walls and well-arranged 2D hexagonal pore structures of these supports remain unchanged [7,32]. Furthermore, Ni can be easily supported on mesoporous silica by immobilizing Ni nanoparticles within the pores using the conventional impregnation method [33].Despite the excellent dispersion of Ni particles on the support achieved by this approach, Ni particle sintering can still occur in DRM at around 600 °C [34,35]. To improve the confinement of Ni particles, the addition of other active metals such as Co could be a solution [36]. Ni-Co-based catalysts are well known in the literature that their alloy and proper ratio of Co and Ni can affect the stability of the catalyst against coke formation. It was found that only Co supported on TiO2 and/or ZrO2 deactivates due to the oxidation of Co [36,37]. XAS showed the presence of NiCo alloy, which produced a negligible amount of deposited carbon, likely due to balancing the oxidative (CO2, H2O) and reductive (CH4, CO, H2) molecules leading to a long-life catalyst [38]. Another method was to support Ni on mesoporous materials such as SBA-15 and to modify them with Mg, Sc, and La oxides as promoters [33]. The evaluation of these catalysts produced intriguing findings. The CH4 conversion was increased by 28 and 26 %, respectively, for the Mg and Sc promoted catalysts compared to the parent catalyst, which reached about 53 % over 6 h on stream at 700 °C. The performance of these catalysts was attributed to the good dispersion of the active metals on the support enhanced by the added promoters. Also, La3+ and Sm3+ were added to Cu-doped to be used as supports for 5 wt.% Ni. Although Sm3+ showed a higher number of basic sites compared to La3+, it was found that La3+ modified Ni/Cu-CeO2 catalyst produced 13 times less carbon deposits due to the formation of small particles of Ni and NiCu alloy. Furthermore, the larger mobility of surface/subsurface lattice oxygen on La3+modified catalyst facilitated carbon removal [39]. The type and the amount of these surface/subsurface oxygen species depend on the nature of the dopants used, as has been reported in the literature [33,40]. The effects of promotors were studied, and the addition of Co to Ni showed higher activity compared to Ca addition, which was also not stable [41]. Furthermore, the method of preparation has an impact on the activity of the catalyst. Ni supported on Ce-Zr using surfactant was not active in DRM compared to coprecipitated material due to the covering the Ni active sites with Ce-Zr. However, the addition of Ce to Ni-ZrO2 was beneficial in terms of decreasing the amount of filamentous coke [42]. In the current work, we have examined the role of Sc in the catalyst system when using single active metal Ni over Sc-SBA-15 support, thereby avoiding the use of expensive Co. Consequently, mesoporous support SBA-15 was doped with different lower scandium loadings (0.5, 1, and 3 wt.%) for optimization. Thereafter, the catalysts were tested in DRM with various feed compositions at different temperatures and evaluated based on CO2, CH4 conversion, H2/CO ratio, and analysis of carbon deposits.SBA-15 was prepared by a method described in one of our previous publications [43]. Both promoter (0.5, 1, 3 wt.% of Sc) and active metal (5 wt.% of Ni) were incorporated into the structure of the support by sequential wet impregnation. At first, calculated amounts of Sc(NO3)3 to set Sc loadings of 0.5, 1, and 3 wt.% were dissolved in water (50 mg Sc/g solution). This was impregnated onto the support, dried overnight (120 °C) and calcined in air at 600 °C for 3 h. The specified amount of Ni(NO3)2·6H2O, equivalent to 5 wt.% of Ni, was dissolved in 30 ml of water. This solution was subsequently impregnated onto the Sc-modified support. The resulting samples were then dried overnight and calcined in the air as done before. The catalyst precursors were designated as Ni-xSc/SBA-15 (x = 0.5, 1, 3 wt.%).The catalysts were tested in DRM in a stainless steel fixed-bed tubular reactor with an internal diameter of 0.94 cm and length of 30 cm (PID Eng & Tech micro activity reactor, Madrid, Spain). A load of 0.1 g of the catalyst powder (not diluted with inert) was placed over glass wool inside the reactor together with a thermocouple that was in contact with the catalyst bed to monitor the temperature. Before starting the reaction, the catalyst was activated in H2 atmosphere (30 ml/min) at 700 °C for one hour. Thereafter, N2 with a flow of 20 ml/min was used to purge the reactor of any residual H2. To ensure no remnant of H2 was present, GC runs were taken while the temperature was increased to 750 °C under flowing of N2. The reaction was performed at 1 bar, CH4/CO2/N2 feed ratio of 3:3:1 (respectively corresponding to 30, 30 and 10 ml/min volumetric flow rate), and 70 ml/min total flow rate and a space velocity of 42,000 ml/(gcat·h). The feed and product were analyzed by an online gas chromatograph (Shimadzu GC-2014) equipped with a combination of molecular sieve and Porapak Q columns and thermal conductivity detector (TCD). The calculation of conversion is based on the following relations: (5) C H 4 c o n v e r s i o n = m o l e o f C H 4 , i n - m o l e o f C H 4 , o u t m o l e o f C H 4 , i n X 100 (6) C O 2 c o n v e r s i o n = m o l e o f C O 2 , i n - m o l e o f C O 2 , o u t m o l e o f C O 2 , i n X 100 (7) H 2 C O = m o l e o f H 2 p r o d u c e d m o l e o f C O p r o d u c e d The N2 adsorption–desorption technique was used to determine the specific surface area of the samples using Tristar II 3020 (Micromeritics, Norcross, GA, USA). For the analysis, 0.2–0.3 g of the sample was taken and outgassed at 200 °C for 3 h. Adsorption-desorption isotherms were recorded at liquid N2 temperature of −196 °C.X-ray powder diffraction (XRD) was done with a Miniflex diffractometer, (Rigaku Corporation, The Woodlands, TX, USA) with a Cu K radiation source and a nickel filter. The device was operated at 40 kV and 40 mA at a step size of 0.01°. The 2θ scanning range adopted for recording the diffraction patterns was 1–3° for low-angle analysis and 10–80° for wide-angle. The Joint Committee on Powder Diffraction Standards (JCPDS) database was used to identify the different phases from the diffractogram.The reducibility of the fresh catalysts was determined with AutoChem II (Micromeritics), where 0.075 g of sample was loaded into the sample tube holder. Samples were heated under pure Ar at 150 °C for 30 min, followed by cooling to room temperature. Thereafter, the temperature was raised to 900 °C at 10 K/min under a flow of 10 %H2/Ar at 40 ml/min. A TCD was used to monitor the H2 content at the outlet. Temperature-programmed CO2 desorption (CO2-TPD) was measured using automatic chemisorption equipment (Micromeritics AutoChem II 2920) with a TCD. A 70 mg sample was heated at 200 °C for 1 h under helium (He) flow to remove adsorbed components and then cooled. Then, CO2 adsorption was carried out at 50 °C for 60 min in a He/CO2 gas mixture (90:10 vol ratio) at 30 ml/min. Afterwards, the temperature was raised to 800 °C at 10 K/min while the TCD recorded the CO2 desorption signals.X-ray photoelectron spectroscopy (XPS) measurements were carried out with an ELS5000 spectrometer (Omicron Nanotechnology) with a monochromatic artificial intelligence source. A beam of X-rays was released from a flood gun having about 400 μm radius. The samples were scanned at different ranges of 395–415 eV, 526–540 eV and 840–900 eV at energy steps of 5, 2, and 10 eV, respectively. The pass energy used was 200 eV.Thermogravimetric analysis (TGA) was performed on a TGA-51 device (Shimadzu, Kyoto, Japan). The spent catalysts (about 10 mg) were recovered after stopping the reaction at ambient temperature and heated from room temperature to 1000 °C with 20 K/min under airflow at 50 ml/min.A high-performance Transmission Electron Microscope (TEM) analysis of the catalysts was made with a JEOL JEM-2100F operated at 120 kV accelerating voltage to study the morphology of the carbon deposits. Fig. 1 demonstrates that the bare support and the fresh samples have very similar N2 adsorption–desorption isotherms of type IV with H1 type hysteresis loops due to capillary condensation at relative pressure above 0.5, according to IUPAC classification [44]. This is typical of mesoporous materials. At a relative pressure above 0.7, a sharp rise in the adsorption isotherm is observed for all samples, indicating that the well-ordered hexagonal framework of the support was maintained after metal impregnation [45]. When Sc and Ni were added to the SBA-15 support, the surface area and pore volume decreased because of pore blockage (Table 1 ). It is not surprising that the surface area decreased as the wt.% of promoter increased. All samples have an average pore diameter of approximately 6.5 nm, classifying them as mesoporous materials.Small-angle XRD analysis (Fig. 2 A) shows three reflections appearing at 0.84, 1.45, and 1.68° in the diffractograms, representing the (100), (110), and (200) planes. These reflections are typical for the 2-D hexagonal symmetry (space group p6mm) of the support SBA-15 [46]. The addition of Ni reduced the reflection intensity of the (100) plane, which completely vanished on adding 1% Sc.For the wide-angle diffraction patterns depicted in Figs. 2B-D, a diffraction line corresponding to amorphous silica from the support common appeared at around 22.7° [47]. Fig. 2B compares the diffraction patterns of SBA-15 support with those of fresh 5Ni/SBA-15 catalyst. This clearly shows the presence of the crystalline NiO phase at the typical diffraction angles 37.2, 43.2, 62.8, 75.3, and 79.3° (JCPDS 78–0643). These reflections belong to the corresponding crystallographic planes (111), (200), (220), (311), and (222) of NiO [48]. No additional peak was observed for samples containing Sc in addition to Ni (Fig. 2C, 2D). The only noticeable difference is a slight reduction in signal intensity representing the silica phase. These findings indicate that the support structure is highly stable.The reduction behavior of the support and the fresh calcined catalysts is shown in Fig. 3 . The support exhibited no reaction with H2. The small peaks that appeared between 150 and 200 °C could be assigned to impurities present in the support as this is common to all the samples. The temperature at which the reduction peaks for Ni2+ to Ni0 appear is dependent on the interaction strength between NiO species and the support. The peaks below 400 °C are associated with the reduction of easily accessible or weakly bound NiO species, whereas the peaks above 500 °C represent NiO species having medium-strength interaction with the support [49]. Increasing the Sc loading reinforces the interaction between NiO species and the support by influencing their distribution and strength. Initially, the intensity of the weakly bound NiO species decreases with increasing Sc loading, while the intensity of the moderately bound NiO species becomes prominent. Secondly, there are shifts in the reduction temperature towards higher values with Sc content for both types of NiO species. The peaks above 700 °C represent the NiO species with a strong interaction with the support but are only noticeable at the highest Sc loading. A similar observation has been reported in studies that utilized Sc as a promoter [50,51].The CO2-TPD was performed with fresh 5Ni-xSc/SBA-15 samples to study their basic properties (Fig. 4 ). The curves in Fig. 4 (CO2-TPD) can be classified into weak, medium, and strong basicity regions, attributed to surface hydroxyl, surface oxygen anion and bulk oxygen anion/oxygen vacancy sites, respectively. The strength of the basic sites was classified as weak (less than 300 °C), medium (300 – 400 °C), strong (400 – 650 °C), and very strong (> 650 °C) depending on the CO2 desorption temperature [52]. These catalysts are characterized by medium basic sites as the CO2 desorption temperature maxima appear at around 270 °C. Compared to 5Ni/SBA-15, the incorporation of Sc increased the number of basic sites as the intensity of the peaks increased on increasing the amount of Sc.The XPS analysis of the fresh catalysts is shown in Fig. 5 . The Ni 2p region, shown in Fig. 5a, exhibited a complex structure of the NiO doublet, with Ni 2p3/2 peak at 855 eV, Ni 2p1/2 at 873 eV, and pronounced satellite features at about 862 and 880 eV for the scandium-free sample [53]. The entire Ni 2p spectrum was slightly shifted to higher binding energies (by ∼ 1.5 eV) when adding 0.5 wt.% scandium. This shift, in turn, decreased with higher Sc loading. In the Sc 2p1/2 region, a characteristic doublet at 403.3–407.8 eV with a splitting of the two peaks by ∼ 4.5 eV was observed (Fig. 5b), assigned to oxides and hydroxides of scandium, respectively [54]. The O 1 s region (Fig. 5c) showed a main peak at about 532.8 eV, generated by SiO2 structures of the SBA-15 support. Because of Si 2p (not shown here), O 1 s and Sc 2p spectra exhibited rather stable binding energies in contrast to Ni 2p. A shift in the BE of O 1 s upon increasing the Sc content was observed, supposing that the added scandium seems to interact mainly with the nickel.Prior to the activity measurements with catalysts, blank tests were performed without active Ni metal using the same operating conditions. Table S1 displays the CH4 and CO2 conversions of the catalysts in the absence of Ni. The CH4 and CO2 conversions are below 2 % for the bare SBA-15 and in the presence of different Sc wt.%.The catalysts 5Ni-xSc/SBA-15 (x = 0, 0.5, 1, 3 wt.%) were tested at 750 °C, with a total feed flow rate of 70 ml/min and a space velocity of 42,000 ml/(gcat·h). The initial CH4 conversion was 76% without Sc, while it increased to 78% upon adding 0.5 wt.% Sc (Fig. 6 A, 6B). Increasing the Sc loading above 0.5 wt.% lowered the CH4 conversion to approximately 70% for both 1 and 3 wt.% of Sc. As for the H2/CO ratio in Table 2 , all the catalysts show high values (≥ 0.94) and the sample with 0.5 wt.% Sc loading led to the highest value (0.99). The deviation from equity is explained by the competing reverse water–gas shift reaction, which converts CO2 with the H2 produced via DRM to form CO and H2O. Obviously, the higher loading of Sc did not enhance the performance of the pristine catalyst 5Ni/SBA-15 [2,33]. On increasing the amount of Sc content above 0.5%, it is probable that the Sc covers the Ni active sites decreasing activity.All the catalysts showed acceptable stability with a low relative performance loss over 8 h on stream, as revealed in Table 2. 5Ni-0.5Sc/SBA-15 was the least deactivated catalyst, indicating that 0.5% Sc undoubtedly was the optimum loading.The H2-TPR profiles (section 3.3) show that the amount of NiO species decreases at low temperatures, while additional peaks appear at higher temperatures upon adding Sc. This may suggest that more NiO particles become strongly bonded to the support, making them difficult to reduce but, at the same time, more resistant to agglomeration. Besides, the XPS results confirm the increase in the binding energy (Ni 2p signal) with the Sc loading.For all the catalysts, CO2 conversion was higher than that of CH4. This suggests that CO2 was partly involved in the reverse water gas shift (RWGS) reaction, which converts CO2 with the H2 produced via DRM to form CO and H2O.The above-discussed tests were performed at the stoichiometric ratio of CH4 and CO2 (eq. 1), whereas in many technically relevant cases, the ratio is higher (CH4/CO2 = 2:1), e.g., in biogas or natural gas. Fig. 7 depicts the results obtained with the best-performing catalyst herein, 5Ni-0.5Sc/SBA-15, at a CH4/CO2 feed ratio = 2 (750 °C, space velocity = 42,000 ml/(gcat·h)).The initial CO2 conversion was around 69% and gradually decreased to 66% after 8 h on stream. Due to CH4 excess and the stoichiometry of DRM, CO2 is acting as the limiting reactant, and CH4 conversion in DRM cannot exceed 50% of the CO2 conversion unless other side reactions occur. With ongoing experiment, the CH4 conversion stabilized at around 32%. Additionally, carbon analysis of the used sample was performed to determine whether a side reaction had occurred. The corresponding TGA plot is available in the supplementary file (Fig. S3). The resulting weight loss assigned to the amount of carbon deposit is around 8 %. In addition, the measured H2/CO ratio is shown in Fig. 7B, where the values obtained during the reaction are close to unity, confirming that the contribution of side reactions is negligible.To get more insight into coke formation, an additional run was made using 5Ni-0.5Sc/SBA-15 with only CH4 and N2 as the feed at comparable conditions. The feed flow rate was set to 30 and 10 ml/min for CH4 and N2, respectively. The reaction was performed at 750 °C and 1 bar. The results are presented in the supplementary file (Fig. S1). No significant decomposition of CH4 was observed for this catalyst, as CH4 conversion was less than 0.5% over 4 h on stream.At the end of the 8 h runs with 5Ni-0.5Sc/SBA-15, the spent catalyst samples were recovered and were then subjected to thermal analysis in the air to quantify the amount of carbon deposits. The weight-loss curves are shown in Fig. 8 . The minute weight gain recorded at the beginning of the analysis could be because of the time-delayed initialization of the precision balance. After that, the sample remained quasistatic until around 450 °C, where a hump was observed for all spent catalysts. Fig. 8 reveals that Ni/SBA-15 has produced the highest amount of carbon deposits of about 8.0%, while the Ni-3Sc/SBA-15 catalyst formed the least amount of 2.5%. As the loading of Sc increases, the weight loss decreases. The catalysts promoted with 0.5 and 1 wt.% Sc demonstrated intermediate weight losses (4.0% and 3.5%, respectively). The basicity was more pronounced in 5Ni-3Sc/SBA-15, as evidenced by CO2-TPD characterization (section 3.4). This sample adsorbed more CO2, which consequently might enhance the gasification of deposited carbon at DRM conditions.A long-term test over 80 h was carried out using 5Ni-0.5Sc/SBA-15 catalyst (Fig. 9 ) at 1 bar, CH4/CO2/N2 feed ratio of 3:3:1, and a total flow rate of 70 ml/min. The initial conversions of CH4 and CO2 reached 78 and 84 %, respectively. The catalyst suffered only 14% activity loss over the complete time on stream to reach final conversions of 67.2 and 74% for CH4 and CO2, respectively. The loss in activity might be due to the partial oxidation of the catalyst in the presence of CO2. As for the H2/CO ratio, a value of about 0.97 was recorded at the start and 0.91 at the end (Fig. 9). The ratio below unity suggests the occurrence of reverse water gas shift reaction. This agrees with the observed higher CO2 conversion relative to that of CH4.This catalyst has good stability for many reasons, possibly due to the well-balanced reduction-CO2 adsorption cycle, which plays a role in the coke gasification, as discussed in section 3.6. It is expected that the catalyst will still be active if the rate of carbon deposition does not outweigh the rate of carbon gasification. On increasing, the amount of Sc content above 0.5%, it is probably that the Sc is covering the Ni active sites leading to the decrease in the activity [33]. This can be seen in the TPR profiles of the samples by observing the first peak, which decreased as the Sc loadings increased. Fig. 10 presents the TEM images of both fresh and spent samples of 5Ni/SBA-15, and 5Ni-0.5Sc/SBA-15 used to investigate the sintering of the Ni metal particles as well as the location and appearance of carbon deposits on the catalysts.TEM images confirm the hexagonal structure of the SBA-15 pores in the fresh Sc-free catalyst, as highlighted in the zoomed area in Fig. 10A. This agrees with the XRD analysis. For 5Ni/SBA-15 (Fig. 10A), the Ni particles appear to be partly confined within the pores of the support and partly localized at the mouth of the pores. As for the fresh 0.5% Sc promoted catalyst in Fig. 10C, the Ni particles are mainly dispersed within the pores except for some sparse particles appearing at the entrance of the pores. In addition to that, the SBA-15 support obviously retained its structure upon adding Sc and Ni. Thus, the addition of Sc facilitates the dispersion of Ni particles within the pores of the support.The particle size distributions as recorded by TEM for fresh and spent samples of 5Ni/SBA-15 and 5Ni-0.5Sc/SBA-15 are shown in Fig. 11 . In the fresh 5Ni/SBA-15, the Ni particle size ranges from 2 to 20 nm, while the spent sample shows much larger Ni particles up to 40 nm. This is accompanied by a slight shift in mean particle size from 6.5 nm to 9.6 nm at the end of the reaction (Fig. 11 A, B). Carbon deposits are present in two morphologies: large filaments based on carbon nanotubes and larger Ni particles encapsulated in carbon spheres are visible (inset in Fig. 10B). Fig. 10D shows that the carbon deposits found on spent 5Ni-0.5Sc/SBA-15 are mostly multi-walled carbon nanotubes of varying diameters. Some encapsulated carbon could also be spotted, like for the Sc-free catalyst. The addition of Sc obviously assisted in preserving the structure of the support.As for the fresh and spent 5Ni-0.5Sc/SBA-15 catalyst (Fig. 11C, D), the mean particle size (2.5–13.5 nm) remained almost the same after reaction (2.5–15 nm). Compared to unpromoted samples, the presence of 0.5% Sc loading inhibits the agglomeration of active Ni metal particles.DRM was performed using Ni-xSc/SBA-15 catalysts (x = 0.5, 1, 3 wt.%). 5Ni-0.5Sc/SBA-15 catalysts outperformed and showed the highest CH4 and CO2 conversions of 78 and 86%, respectively, and the highest stability compared to the other catalysts. This can be explained based on the H2-TPR analysis, which showed that adding Sc strengthened the bonding between Ni and the support as found by shifting the reduction temperature on increasing the Sc loading leading to the stability of the catalyst. The XRD analysis revealed the presence of NiO on the as-prepared calcined catalysts, whereas the absence of reflections for Sc-containing phases indicated its high dispersion even at a 3 wt.% loading. The XPS studies displayed that the binding energy of Ni with the support increased by about 1.5 eV upon adding only 0.5 wt.% Sc and O 1s also showed that on increasing Sc loading above 0.5%, the interaction of Ni and Sc increases which might lead to the deactivation of the catalyst. Additionally, the CO2-TPD indicated that all catalysts have medium basic sites, and increasing the amount of Sc increases the basicity with an optimum of 0.5% Sc. The catalyst was stable over 80 h showing a loss of 8%.One of the reasons for catalyst deactivation besides coking is its partial oxidation with CO2. This was particularly observed for the sample with a 3 wt.% Sc load, which formed the least amount of carbon at 2.8%, despite exhibiting more pronounced deactivation. The TEM images of the spent catalyst 5Ni-0.5Sc/SBA-15 displayed that the Ni particle size remained unchanged after the reaction. Therefore, a catalytic test of 5Ni-0.5Sc/SBA-15 catalyst using a CH4/CO2 feed ratio = 2 to simulate biogas feed was performed. The results indicated that dry reforming reaction was predominant under these conditions. However, this study was limited as the catalytic tests were measured only at one set point reaction temperature. It would be interesting to perform the catalytic tests at different measurements to calculate kinetic energy. Additionally, no investigation of the pressure effect. Also, no variation of contact time, e.g., the effect of conversion on coking. It is recommended to apply catalytic tests soon to investigate the previous issues.The views and opinions expressed in this paper do not necessarily reflect those of the European Commission or the Special EU Programs Body (SEUPB). Ahmed S. Al-Fatesh: Conceptualization, Methodology, Writing – review & editing, Supervision. Samsudeen O. Kasim: Conceptualization, Methodology, Writing – review & editing, Supervision. Ahmed A. Ibrahim: Conceptualization, Methodology, Writing – original draft. Ahmed I. Osman: Conceptualization, Methodology, Writing – review & editing, Supervision. Ahmed E. Abasaeed: Conceptualization, Methodology, Writing – original draft. Hanan Atia: Writing – reviewing, Supervision, Editing. Udo Armbruster: Writing – reviewing, Supervision, Editing. Leone Frusteri: Data curation, Formal analysis. Abdulrahman bin Jumah: Writing – original draft. Yousef Mohammed Alanazi: Writing – original draft. Anis H. Fakeeha: Writing – reviewing, Supervision, Editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to sincerely appreciate to Researchers Supporting Project number (RSP-2021/368), King Saud University, Riyadh, Saudi Arabia. Dr Ahmed I. Osman wishes to acknowledge the support of The Bryden Centre project (Project ID VA5048). The Bryden Centre project is supported by the European Union’s INTERREG VA Programme, managed by the Special EU Programmes Body (SEUPB).Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuel.2022.125523.The following are the Supplementary data to this article: Supplementary data 1

This study investigated the performance of supported Ni catalysts in the utilization of greenhouse gases like CO2 and CH4 via dry reforming. The support SBA-15 was impregnated first with Sc at different loadings (0.5, 1, and 3 wt.%) and then with Ni (5 wt.%). The catalysts were first tested up to 8 h on stream with stoichiometric feed as well as methane in excess. The as-prepared catalysts were characterized using BET, XRD, TPR, CO2-TPD, XPS, TGA, and TEM. This is in accordance with the surface area measurement, XRD, and TEM data. The Ni added to Sc-SBA-15 appeared to interact with both the support and Sc as the intensity and reduction temperature of the Sc promoted catalysts increased relatively to the unpromoted sample depending on Sc content. The catalyst with 0.5 wt.% Sc loading led to the highest conversion and the lowest relative activity loss. The CH4 and CO2 conversions, on average, were 78 and 86 %, respectively, at the end of the runs at 750 °C. The final H2/CO ratio was 0.99, which is a good value compared to many literature catalysts. This catalyst also showed relatively constant CH4 and CO2 conversions over 80 h on stream. Increasing the Sc loading above 0.5 wt.% was not beneficial in terms of activity.

No data was used for the research described in the article.Extensive use of fossil fuels along has conducted to a rapid and continuous increase of CO2 concentration in the atmosphere [1], up to 420 ppm in April 2022, while preindustrial levels were of 278 ppm (in ca. 1750) [2]. Therefore, there is an increasing interest in CO2 capture and conversion processes that can contribute to revert this trend.CO2 capture consists of selectively removing the CO2 from ambient air or industrial process streams to produce a concentrated stream of CO2 that can be transported to the storage site [3]. Among the technologies developed for CO2 capture, chemical and physical absorption stand out, since they are the most mature and used at commercial scale [4]. Usually, CO2 is absorbed by aqueous solutions of amines or NaOH [5,6]. However, CO2 capture and storage presents high costs due to the desorption and compression steps prior to transportation [5]. The utilization of this captured CO2 to produce value-added chemicals may compensate the costs associated to its capture [7]. Nevertheless, the industrial use of CO2 as a raw material is still limited due to its high chemical stability [8].There are different alternatives to produce chemicals and fuels from CO2, including hydrogenation [9], electrochemical reactions [10] and photochemical reactions [11]. However, these technologies still show low yields and high cost and, therefore, further development is necessary [12]. Hydrogen is the main reductant used in these processes. Although nowadays most of the hydrogen is still produced by energy intensive processes, such as the endothermic steam methane reforming (SMR) [13], it is expected that in the near future abundant green hydrogen, produced by water hydrolysis, will be available, since pilot to commercial-scale plants are rapidly developing [14]. In-situ hydrogen production methods that can yield a more reactive reductant at a lower cost also are of high interest.High-temperature water (HTW) has emerged as an alternative hydrogen donor and reaction media due to the fact that it presents fewer and weaker hydrogen bonds, lower dielectric constant and a higher isothermal compressibility than water at room temperature. Moreover, its use is preferred over organic solvents because it is an environmentally friendly solvent [15]. The dissociation of water with a metal under hydrothermal conditions is an alternative to direct use of hydrogen in the reduction of CO2. Studies on the abiotic synthesis of organics indicate the feasibility of H2O dissociation and production of organics by CO2 reduction using metals under hydrothermal conditions [16–18]. The formation of long-chain hydrocarbons by hydrothermal reduction has also been recently demonstrated [19]. Moreover, organics, such as CH4 [20] and other hydrocarbons [21] have been found in hydrothermal oceanic vents which may indicate the leading role that reactions such as serpentinization of magnesium-and iron-rich rocks to produce H2 may have had in the origin of life on the Earth.For the industrial development of this technology, to find an active catalyst for CO2 conversion under mild conditions is crucial for using CO2 as a raw material in the production of chemical and fuels. Among other types, such as homogeneous and biological catalysts, heterogeneous catalysts show some advantages such as high stability and easy separation from reactants and products. Thus, they are usually preferred for their use in industrial applications [22].Different products can be synthesized from CO2 using metal reductants and catalysts in hydrothermal conditions. For example, CH4 was produced with a yield of 98 % from NaHCO3 using Raney Ni nanoparticles as catalyst [23], and acetate was obtained with yields in the range of 10 % using cobalt-based catalysts [24]. In addition to CH4, methanol can also be formed using CO2 as the raw material [25,26]. The production of formic acid by the reduction of CO2 captured as NaHCO3 under hydrothermal conditions has been previously studied to optimize the reaction parameters, particularly the reductant:catalyst:NaHCO3 molar ratio, the reaction temperature and time and the amount of water employed. For instance, using a combination of Ni as catalyst and Fe as the reductant with a ratio 1:1, a yield to formic acid of 15.6 % was reached after 2 h of reaction at 300 ºC [27,28]. Higher yields to formic acid of 63.6 % were found when using a combination of Fe and Cu with a molar ratio 6:6:1 of Fe/Cu/NaHCO3 at the same reaction time and temperature [29]. The performance of Fe reductant without catalyst was also investigated, yielding 92 % of formic acid when employing high proportions of Fe [30]. Zn can also be used as reductant for the hydrothermal reduction of NaHCO3, with yields between 64 % and 78 % at 300 and 325 ºC respectively [31,32], values that could be increased with Ni catalysts [33]. In the case of Al, the yield to formic acid obtained after 2 h of reaction at 300 ºC was also 64 % [34]. Besides the high yields obtained in the hydrothermal reduction of CO2 using metals, it presents solutions to two of the challenges of CO2 reduction processes presented above: the reactivity of CO2 captured in basic solutions, such as HCO3 -, is higher than that of gaseous CO2, and the reaction can take place in the same aqueous media where CO2 is captured by NaOH, without intermediate separation or purification processes, thus avoiding the related energy consumption and processing costs.The main product obtained in most of these studies is formic acid. Formic acid can be used as preservative and insecticide, as a reducing agent, or as carbon source in synthetic chemical industries [35]. The dehydrogenation of formic acid to produce hydrogen is a fast and easily controllable process and therefore, in the past years, formic acid has also gained great attention as a hydrogen storage vector.The hydrothermal reduction of CO2 therefore presents promising advantages in terms of integration with capture processes, selectivity and yield, but the harsh pressure and, particularly, temperature conditions required to carry it out still are a concern, since these conditions have a direct impact on the cost of the process and on the stability of the base (e.g. amine) used to capture CO2 and, therefore, the possibility to recycle it. It is therefore of great interest to reduce the required operating temperature, while maintaining the performance of the process. With this purpose, in this work a large number of combinations of metal reductants and catalysts are tested systematically. In comparison with previous studies, in which tested temperatures were above 250 ºC, in this work operating temperatures are reduced down to 200 ºC. Moreover, a kinetic model under the optimum reaction conditions is provided.NaHCO3 (100 %) was purchased from COFARCAS (Spain). The reductants employed included Zn powder (< 150 µm, 99.995 % metal basis) and Fe powder (≥ 99 %), both provided by Sigma Aldrich (Spain), and granular Al (< 1 mm, 99.7 %) from Panreac (Spain). The catalysts used encompassed Cu powder (<425 µm, 99.5 % metal basis), Ni powder (< 150 µm, 99.99 %), Fe3O4 powder (50–100 nm, 97 % metal basis) and Pd/C (5 % Pd content) acquired from Sigma Aldrich (Spain), as well as Fe2O3 powder (< 5 µm, ≥ 96 %) from Panreac (Spain). A standard reagent formic acid (puriss. ∼98 %) from Sigma Aldrich was used for obtaining the calibration curves. All reductants and catalysts were used without further treatment or purification.Aqueous solutions of NaHCO3 were used as the CO2 source. NaHCO3 solutions were prepared with MilliQ water at a concentration of 0.5 M. Experiments were conducted in batch reactors (length: 16 cm, o.d.: ½”, wall thickness: 0.083″) made of SS 316 stainless steel tubing with an internal volume of 9 mL.Each reactor was loaded with the selected reductant in a molar ratio reductant:CO2 of 5:1 and the catalyst in a molar ratio catalyst:CO2 of 2:1, except in the case of Pd/C catalyst, were due to limitations with respect to the volume of solids that could be loaded in the reactor while maintaining an efficient stirring, a catalyst:CO2 mass ratio of 0.25:1, corresponding to a Pd: CO2 molar ratio of 0.005:1, was used. Thereafter, the reactor was filled with NaHCO3 solution up to approximately 40 % of the volume of the reactor. The closed reactors were placed in an Al2O3 sand fluidized heating bed preheated at the target temperature (200, 250 or 300 ºC) to ensure a rapid heating, which required between 3 and 5 min at the temperature range of 200–300 ºC. After the reaction time was completed, reactors were introduced in a cold water bath to quench the reaction. Liquid samples were collected and the solid reductants and catalysts were separated by vacuum filtration and dried at 105 ºC overnight. To ensure reproducibility, each reaction was conducted at least twice and the standard deviation between the results of the repeated experiments was calculated.Liquid samples were analyzed by HPLC (Waters, Alliance separation module e2695) attached to a RI detector (Waters, 2414 module) using a Rezex-ROA-Organic Acid (8 %, pore size 300 ×7.8 mm) purchased from Phenomenex. Prior to analysis, all the samples were filtered through a 0.22 µm filter. The HPLC method consisted of passing a mobile phase of 25 mM of H2SO4 with a flow rate of 0.5 mL/min during 30 min. The temperatures of the column and the detector were set at 40 and 30 ºC, respectively. Each sample was analyzed twice to ensure reproducibility of the HPLC operation.The yield of formate was calculated according to Eq. 1: (1) Y Formate = C Formate , f C NaHCO 3 , i × 100 where C Formate,f is the molar concentration of formate obtained at the end of the reaction calculated by calibration curves in HPLC analysis, and C NaHCO3,i is the initial concentration of the NaHCO3 aqueous solution, fixed at 0.5 M.XRD patterns were recorded using a Bruker D8 Discover A25 diffractometer attached to a LynxEye detector operated at a voltage of 40 kV and a current of 30 mA. Data were collected at room temperature in the 2θ range from 5 to 70º with a step size of 0.020º using Cu Kα radiation (λ = 1.5418 Å). Database PDF-2-ICDD 2020 was used to analyze the XRD patterns collected.N2 gas adsorption was used to determine the surface area of the fresh Pd/C catalyst. The catalyst surface area was measured according to BET method using a ASAP™ 2420 Micromeritics Accelerated Surface Area and Porosimetry System. 0.1133 g of fresh Pd/C was introduced in the sampling tube and after degassing it under vacuum overnight, and the N2 isotherms were recorded at − 196 ºC.In a previous work of this research group [35], it was stated that in this type of reactions the equilibrium formic acid/formate is mostly shifted to formate due to the reaction pH, which is alkaline. Therefore, the reactions considered are presented in R1-R4. (R1) 2 H 2 O l + Al s → 1.5 H 2 aq + AlO ( OH ) ( s ) (R2) NaHC O 3 + H 2 → Na + + HCOO − + H 2 O (R3) HCO O − aq + H 2 O l → HC O 3 − aq + H 2 ( aq ) (R4) HCO O − aq → CO g + O H − ( aq ) In this previous work, it was observed that the reaction R1 of production of hydrogen was much faster than bicarbonate reduction [35]. Thus, this first step was not considered in the simplified model, assuming that H2 is instantaneously formed. With R2, bicarbonate is reduced to formate using hydrogen. This formate is decomposed to CO2 (R3) and to CO (R4) [36]. As the conversion of the decomposition of HCOO- to CO (R4) is at least an order of magnitude lower than the conversions of R3, the formation of CO was also not taken into account for simplification purposes [36]. As equilibrium calculation presented in a formed work [35] resulted that most CO2 was dissolved in aqueous solutions as bicarbonate, in R3 the product is HCO3 - instead of gaseous CO2.Having into account these simplifications, the global reaction taken account in the model is R5, presented as a pseudo-equilibrium reaction. (R5) NaHC O 3 + H 2 ↔ Na + + HCOO − + H 2 O Whose kinetics follow Eq. 1: (1) dC NaHC O 3 dt = 1 S cat − k 1 ∙ C NaHC O 3 m ∙ C H 2 n + k 2 ∙ C HCOO − p Where S cat is the surface area of the catalyst, k 1 and k 2 are the kinetic constants of bicarbonate and formate decompositions, direct and inverse reactions, respectively; m , n and p the order of the reaction respect to each compound, C NaHC O 3 is the concentration of bicarbonate, C HCO O − is the concentration of formate and C H 2 is the concentration of H2.To study the kinetics of the reaction, experiments were conducted at different times, specifically 15, 30, 60, 90 and 120 min. At these times, the concentration of NaHCO3 and HCOO- were quantified by HPLC analysis. It is important to highlight that the concentration of H2 ( C H 2 ) is considered constant due to the fact that it is assumed that the metal reductant is completely oxidized and H2 is released very quickly, according to the results obtained in a previous work [35]. As the mole ratio of reductant to NaHCO3 employed is 5:1, the amount of H2 formed is highly in excess in comparison to NaHCO3. Moreover, H2 is present in both gas and liquid phase. The concentration of H2 in the liquid phase is determined by its solubility at the pressure and temperature of the reaction media. To calculate the solubility of H2 the next steps were followed: (1) As aforementioned, it is assumed that the reductant is completely oxidized. Therefore, the amount of H2 formed depends on the redox reaction of the reductant and water. In the case of Al reductant, 1.5 mol of H2 is formed per each mol of Al, according to Reaction R1. Considering this, it can be assumed that the concentration of H2 remains constant during the reaction since the molar ratio H2:CO2 is 7.5, and therefore, NaHCO3 is the limiting reactant. (2) The volume that the H2 occupies is 5.4 mL, this is, the total volume of the reactor (9 mL) minus the volume of the solution of NaHCO3 added (3.6 mL). (3) Using the ideal gas equation, the pressure of H2 is calculated in that volume at the reaction temperature. (4) Using data generated with the Predictive Soave Redlich Kwong (PSRK) equation [37], a model to calculate the molar fraction of H2 dissolved in H2O is developed. The model is valid from pressures from 50 to 150 atm. For simplification of the calculations, the results of this thermodynamic model were correlated according to Eqs. 2–4 for 200, 250 and 300 ºC respectively. Model fit showed a R 2 = 1.00 , R 2 = 0.998 and R 2 = 0.988 for the temperatures 200, 250 and 300 ºC respectively. (2) x H 2 = 5.834 ∙ 10 − 8 P 2 + 2.004 ∙ 10 − 5 P − 3.0689 ∙ 10 − 4 (3) x H 2 = 2.375 ∙ 10 − 7 P 2 − 1.084 ∙ 10 − 5 P + 1.154 ∙ 10 − 3 (4) x H 2 = 5.748 ∙ 10 − 7 P 2 − 7.5346 ∙ 10 − 5 P + 2.906 ∙ 10 − 3 where x H 2 is the molar fraction of H2 in H2O and P the pressure in atm. (5) The number of mole of H2O at reaction conditions is calculated taken into the account its density at the reaction temperature and pressure. The density of the H2O was calculated with the MS Excel Add-In Water97v13.xla [38]. (6) The solubility of H2, and thus, the amount of H2 in water, can be calculated using the model developed in step 4 and the amount of H2O calculated in step 5. (7) Once the amount of H2 in H2O is known, the H2 remaining in the gas phase is recalculated. (8) With this new value of the H2 in gas phase, its pressure is calculated according to step 3. Steps 4, 5, 6 and 7 are iterated until the pressure calculated in step 7 converges to the one used in step 3. (9) Once the values of the H2 pressure calculated in steps 3 and 7 are equal, the amount of H2 in H2O is given by steps 4 and 5. As aforementioned, it is assumed that the reductant is completely oxidized. Therefore, the amount of H2 formed depends on the redox reaction of the reductant and water. In the case of Al reductant, 1.5 mol of H2 is formed per each mol of Al, according to Reaction R1. Considering this, it can be assumed that the concentration of H2 remains constant during the reaction since the molar ratio H2:CO2 is 7.5, and therefore, NaHCO3 is the limiting reactant.The volume that the H2 occupies is 5.4 mL, this is, the total volume of the reactor (9 mL) minus the volume of the solution of NaHCO3 added (3.6 mL).Using the ideal gas equation, the pressure of H2 is calculated in that volume at the reaction temperature.Using data generated with the Predictive Soave Redlich Kwong (PSRK) equation [37], a model to calculate the molar fraction of H2 dissolved in H2O is developed. The model is valid from pressures from 50 to 150 atm. For simplification of the calculations, the results of this thermodynamic model were correlated according to Eqs. 2–4 for 200, 250 and 300 ºC respectively. Model fit showed a R 2 = 1.00 , R 2 = 0.998 and R 2 = 0.988 for the temperatures 200, 250 and 300 ºC respectively. (2) x H 2 = 5.834 ∙ 10 − 8 P 2 + 2.004 ∙ 10 − 5 P − 3.0689 ∙ 10 − 4 (3) x H 2 = 2.375 ∙ 10 − 7 P 2 − 1.084 ∙ 10 − 5 P + 1.154 ∙ 10 − 3 (4) x H 2 = 5.748 ∙ 10 − 7 P 2 − 7.5346 ∙ 10 − 5 P + 2.906 ∙ 10 − 3 where x H 2 is the molar fraction of H2 in H2O and P the pressure in atm.The number of mole of H2O at reaction conditions is calculated taken into the account its density at the reaction temperature and pressure. The density of the H2O was calculated with the MS Excel Add-In Water97v13.xla [38].The solubility of H2, and thus, the amount of H2 in water, can be calculated using the model developed in step 4 and the amount of H2O calculated in step 5.Once the amount of H2 in H2O is known, the H2 remaining in the gas phase is recalculated.With this new value of the H2 in gas phase, its pressure is calculated according to step 3. Steps 4, 5, 6 and 7 are iterated until the pressure calculated in step 7 converges to the one used in step 3.Once the values of the H2 pressure calculated in steps 3 and 7 are equal, the amount of H2 in H2O is given by steps 4 and 5.Once the initial concentrations of NaHCO3, H2 and HCOO- are known, the system is modeled considered it as a discontinuous stirred tank reactor, solving the mass balances to formic acid, hydrogen and bicarbonate. In addition to Eq. 1, Eqs. 5 and 6 were also solved, using the Euler numerical method: (5) C NaHCO 3 t = C NaHCO 3 t − 1 + d C NaHCO 3 ( t − 1 ) dt ∙ dt (6) C HCO O − t = C HCO O − t − 1 − d C HCO O − ( t − 1 ) dt ∙ dt (1) The differential equations are solved using the Euler method with a time step of 10 s. The values of the orders of the reaction, this is m, n and p simulated were 1, 2 and 0.5. All the combinations of these values were simulated in order to optimize. (2) With the fixed reactions orders, the values k1 and k2 were optimized in order to minimize the objective Eq. 7 using the function Solver of Excel: (7) AVERAGE ∑ t = 15 min t = 120 min ABS C NaHCO 3 _ MODEL − C NaHCO 3 _ EXPERIMENTAL C NaHCO 3 _ EXPERIMENTAL = 0 where C NaHCO 3 _MODEL is the concentrration of NaHCO3 obtained with the model and C NaHCO 3 _EXPERIMENTAL is the value of NaHCO3 experimentally obtained. (3) The combination of reaction orders selected was the one where the sum of the average absolute error for the concentration of NaHCO3 plus the average absolute error of the concentration of formate was smaller. The differential equations are solved using the Euler method with a time step of 10 s. The values of the orders of the reaction, this is m, n and p simulated were 1, 2 and 0.5. All the combinations of these values were simulated in order to optimize.With the fixed reactions orders, the values k1 and k2 were optimized in order to minimize the objective Eq. 7 using the function Solver of Excel: (7) AVERAGE ∑ t = 15 min t = 120 min ABS C NaHCO 3 _ MODEL − C NaHCO 3 _ EXPERIMENTAL C NaHCO 3 _ EXPERIMENTAL = 0 where C NaHCO 3 _MODEL is the concentrration of NaHCO3 obtained with the model and C NaHCO 3 _EXPERIMENTAL is the value of NaHCO3 experimentally obtained.The combination of reaction orders selected was the one where the sum of the average absolute error for the concentration of NaHCO3 plus the average absolute error of the concentration of formate was smaller.The kinetic constants were correlated at three different temperatures: 200, 250 and 300 ºC. The values of the constant were adjusted to Arrhenius equation to easily calculate the effect of the temperature (Eq. 8). (8) k T = A ∙ e − E a RT where k ( T ) is the rate constant as a function of the temperature, A is the pre-exponential factor, E a the activation energy, R is the gas constant and T is the absolute the reaction temperature.This work studied the influence of different combinations of reductants and catalysts to reduce CO2 to formate under hydrothermal conditions. As aforementioned, NaHCO3 was used as the carbon source, being it the product resulting from the capture of CO2 with basic NaOH solutions. The reductants employed included Zn, Al, and Fe and the catalysts were Cu, Ni, Fe2O3, Fe3O4 and Pd/C. Three different reaction temperatures were also explored, specifically 200, 250 and 300 ºC. Temperatures above 300 ºC were not tested since the decomposition of formic acid into CO2 and H2 at high temperatures under hydrothermal conditions is high [36].When using Al or Zn as reductants, the selectivity of liquid-phase products towards formate was 100 %. In the case of using iron as reductant, a peak corresponding to another unidentified product was found apart form the peak of formate. Sample chromatograms are provided as Supplementary Information (Fig. S1). Formate yields are plotted in Fig. 1 together with the error bars calculated from the results of repeated experiments. The standard deviation of the two replicates was lower than 1.5 % in most of the cases and, therefore, some of the error bars cannot be appreciated in Fig. 1 due to the axis scale. Table 1 compiles the yields obtained with the different combinations of metal reductants, catalysts and temperature.It is clear from Fig. 1 that the temperature has a positive effect on the yield to formate. The higher the temperature, the higher the yield and, in general, the change observed in the yield is greater from 250 to 300 ºC than from 200 to 250 ºC, except in the case of Pd/C and Ni catalysts (Fig. 1b a 1c respectively), where the yield at 300 ºC was only slightly higher than that at 250 ºC when using Zn as the reductant. In contrast, with Al and Fe reductants, the yield to formate decreased moderately from 250 to 300 ºC.The highest formate yield reached a value of 57 %. It was observed at 250 ºC when the reductant was Al and the catalyst Pd/C. A similar yield of 55 % was obtained at 300 ºC with Cu catalyst, again with Al reductant. Interestingly, a comparable yield of 52 % was detected at 300 ºC with Zn reductant in the absence of catalyst. Indeed, in the absence of catalyst, Zn showed the best performance at the three temperatures evaluated, yielding 5.4 % and 12 % of formate at 200 and 250ºC respectively. In contrast, at 200 ºC, the yields obtained with Al and Fe and without catalyst were negligible and at 250 ºC the yield with Al was of 5.6 % while in the case of Fe was only 1 %. The feasibility of using Zn reductant to produce formic acid from NaHCO3 under hydrothermal conditions was previously demonstrated [32], where the intermediate Zn-H, obtained by the oxidation of Zn by HTW, may have a leading role by acting as the active hydrogen source in CO2 hydrogenation [39].With Fe reductant, the performance of Fe2O3 catalyst at 200 and 250 ºC is practically constant, yielding 3 % and 4 % of formate, respectively. The same trend can be observed with Fe3O4 catalyst, where the yield at both temperatures was lower than 1 %. However, at 300 ºC, Fe3O4 showed a better performance than Fe2O3 catalyst for Fe reductant yielding 30 % of formate, while the yield with Fe2O3 was 24 %. In general, in the temperature range studied, Fe reductant showed the lowest yields, while Al reductant exhibited the best performance. However, in the absence of catalyst with Fe reductant at 300 ºC, the yield obtained was higher than in the case of Al powder reductant, reaching a value of 45 %. This yield is comparable to that observed by Duo et al. [30], who determined a yield of approximately 50 % with Fe, although they used a low NaHCO3 concentration of 2 mmol/L. Duo et al. [30] significantly increased the yield of formic acid to more than 90 % by increasing the amount of Fe powder employed. The best performance of Fe reductant detected by Duo et al. [30] can be explained by the particle size of the reductant and the more diluted reaction conditions, along with the horizontally shaken of the reactor which may have enhanced mixing and heat transfer, favoring the reaction. However, for practical applications, higher reactant concentrations are desirable to increase throughput.It is very well-known that Cu catalyst is very selective to methanol when reacting with CO2 [25]. However, methanol was not observed in this work, probably due to the alkaline pH of the reaction media, since Huo et al. [25] used HCl to acidify the media to produce methanol. Fig. 2 presents SEM micrographs of selected solid samples after reaction experiments, while Figs. 3–5 presents the corresponding XRD patterns. XRD was employed to investigate changes in the phases present in both the reductants and selected catalysts after reaction. Fig. 3 shows the changes in the phases of the three reductants employed after 120 min of reaction at 300 ºC. Reference diffractograms of the original, unoxidized metals can be retrieved from [40].As it is clear from Fig. 3, the only reductant completely oxidized after reacting during 120 min at 300 ºC was Zn. This result is in agreement with the works of Jin et al. and Roman-Gonzalez et al. [32,34] who demonstrated that under hydrothermal conditions, Zn was almost completely oxidized to ZnO after 10 min of reaction time. In the case of Al reductant, both crystal phases are present, Al and AlO(OH). However, Yao et al. [33] concluded that the oxidation of Al under hydrothermal conditions in the presence on NaHCO3 was completed after 30 and 90 min of reaction time. This disagreement in the results may be explained by the different particle size of Al powder employed. In Fig. 2b it is shown that Fe was oxidized to Fe3O4 under the reaction conditions, but not completely, since typical peaks of Fe crystal phases, specifically at 2θ of 44.8 and 65º, are still present.XRD analysis of Fe reductant combined with Fe2O3 and Fe3O4 catalysts was also conducted. The XRD patterns obtained are shown in Fig. 4. It can be seen in Fig. 4 that Fe2O3 is only present when it was used as catalyst (Fig. 4a) and therefore, under reaction conditions Fe reductant is only oxidized to Fe3O4 (Fig. 4b). The same conclusion can be reached by looking at Fig. 2b. Duo et al. [30] also stated that under hydrothermal conditions, Fe may oxidized to Fe3O4 rather than Fe2O3. Interestingly, it seems that the presence of Fe3O4 in the media promotes in some extent the oxidation of Fe, because after reaction only one a small characteristic peak of Fe phase appeared at 44.5º (Fig. 4b), while in the case of using Fe2O3 as catalyst (Fig. 4a) or just Fe as reductant (Fig. 3b), two characteristic phase peaks of Fe are detected, specifically at 44.5º and 65º and with high intensities.The evolution of the crystal phases of Pd/C catalyst and Al reductant at different reaction times was also investigated. The results are shown in Fig. 4. The XRD patterns at different times shown no apparent differences, as can be seen from Fig. 5. No matter long or short reaction times, Al reductant was not completely oxidized and typical Al crystal phase at 2θ of 38º, 44º and 65º are still detected after 120 min of reaction time at 300 ºC. On the other hand, Pd crystal phase could not be detected in the diffractogram, as a consequence of the low concentration of Pd in the Pd/C catalyst (5 %wt) and the low proportion of the catalyst in the total sample (0.018 g of catalyst vs 0.24 g of Al).The kinetic behavior of the reaction was investigated at three different temperatures according to the method explained in Section 2.5. The best adjustment to the experimental data at 250 ºC and 300 ºC was for a first order reaction respect to all components. Therefore, the units of the kinetic constants obtained are expressed in m−2s−1. Figures from 5 to 7 show the model for a pseudo-first order reaction respect to all components and the experimental data for temperatures of 200, 250 and 300 ºC respectively, while Table 2 presents the average and maximum deviations between experiments and calculations (Eq. 7) obtained at each temperature.As it can be seen from Figs. 6 to 8, the model correctly describes the experimental results at the three tested temperatures, with average deviations in the concentration of NaHCO3 ranging from 1.5 % to 3.6 %, slightly increasing with temperature. At 250 ºC it appears that the experimental point at 120 min differs significantly from the model prediction. Errors in the calculation of the concentration of formate are slightly higher, ranging from 4.9 % to 5.9 %. Again, the experimental point at 120 min is the one which shows more variation respect to the model. These deviations of the model with respect to experiments are satisfactory since they are comparable to uncertainties in experimental results, reported in section3.1; indeed, inspection of Figs. 6 to 8 indicate that deviations can be attributed to a large extent to scatter of some experimental data, with the model correctly reproducing the global trends of variation of the results. Figs. 6 to 8 indicate that at least 95 % of the equilibrium concentration for both bicarbonate conversion and formate production are reached in the first 30 min of the reaction. Therefore, the reaction can be stopped after 30 min, reaching comparable yields with respect to 120 min reaction for this specific set of reaction parameters, this is, Pd/C catalyst and Al reductant at 250 ºC. Table 3 summarizes the values of the kinetic constants, both for bicarbonate decomposition (k1) and formate formation (k2) at 200, 250 and 300 ºC. As one might expect, both kinetic constants increased at higher temperatures. The improvement observed from 250 to 300 ºC was more significant than from 200 to 250 ºC. Fig. 9.The values of the activation energy ( E a / R ) and the pre-exponential factor ( A ) calculated by Eq. 3 using the data obtained in Fig. 8 are shown in Table 4. The value of R2 are also included.The E a for the formation of HCOO- from NaHCO3 in hydrothermal media is 49.8 kJ/mol. The activation energy of the hydrogenation of CO2 into formic acid over a Cu/ZnO/Al2O3 catalyst has been previously calculated resulting in a value of 21.4 kJ/mol [41]. Higher values of the activation energy indicate that the formation of the formate requires more energy input. Therefore, the formation of formate from NaHCO3 under hydrothermal conditions needs more energy than when the production takes place by the hydrogenation of CO2. The isothermal decomposition of NaHCO3 into Na2CO3, CO2 and H2O presented an E a of 94.3 kJ/mol [42] under nitrogen atmosphere which is a higher value in comparison to the decomposition of NaHCO3 under hydrothermal calculated in this work which was 82.6 kJ/mol.In this work, the hydrothermal conversion of CO2 dissolved in aqueous solutions as NaHCO3 was optimized at lower temperatures, with imply lower costs and milder conditions for the reuse of bases used for CO2 capture, considering a number of combinations of reductants and catalysts, demonstrating the possibility of enhancing the yield to formate at lower temperatures by selecting the appropriate combination of reductant and catalyst. Moreover, the reductants and catalysts tested are in general abundant and commercially available materials. The highest yield to formate observed was 57 % using Al as the reductant and Pd/C as the catalyst at 250 ºC. Using Al as reductant allowed to reach yields to formate higher than 50 % at 250 ºC. In the case of the other reductants tested (Zn and Fe), yields higher than 50 % were only observed when the temperature was 300 ºC. This improvement caused by the temperature was greater in the case of Fe reductant and Fe oxides catalysts, where the yields increased from 4 % at 250 ºC to 24 % and 31 % at 300 ºC for Fe2O3 and Fe3O4 respectively. Pd/C catalyst also showed higher yields to formate in comparison with the other catalysts tested at low temperatures with both Al and Zn reductants. In the case of Fe reductant, the yield with Pd/C was practically constant in the temperature range investigated.Furthermore, a kinetic model was developed to describe the reduction of bicarbonate using Al as reductant and Pd as Catalyst. The simplified model model presents the system as a pseudo-equilibrium between formate formation and destruction. The model is able to reproduce the resulting concentrations, with average deviations with respect to experimental data ranging from 1.5 % to 5.9 %, and a maximum deviation lower than 10 %, and correctly predicts the variation of the performance of the reaction with the operating temperature.The present study shows not only the potential of reducing CO2 emissions by using it as a C1 building block, but also a sustainable alternative to produce value-added chemicals such as formic acid. Laura Quintana-Gómez: Methodology, Investigation, Writing – original draft. Pablo Martínez-Álvarez: Investigation. José J. Segovia: Conceptualization, Methodology. Ángel Martín: Conceptualization. Methodology. Resources. Writing – review & editing. Visualization. Supervision. Writing – review & editing. María Dolores Bermejo: Conceptualization, Methodology, Resources, Writing – Review & Editing, Supervision, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This project has been funded by the Ministry of Science and Universities through project RTI2018-097456-B-I00 and by the Junta de Castilla y León through project by FEDER FUNDS under the BioEcoUVa Strategic Program (CLU-2019-04). Authors also thank Jesús Salvador Azpeleta Izquierdo (Universidad de Valladolid) for his support and assistance in XRD analysis and María Dolores Marqués Gutiérrez (Laboratorio de Sólidos Porosos Servicios Centrales de Apoyo a la Investigación –Universidad de Málaga) for her assistance in BET surface area analysis.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jcou.2022.102369. Supplementary material .

The hydrothermal reduction of CO2 captured in aqueous solutions using metal reductants is a promising novel approach that achieves high yields of conversion and high selectivity, but it presents the limitation of the high temperatures needed for the reaction to take place. In this work, experiments combining several reductant metals (Zn, Al and Fe), catalysts (Pd/C, Ni, Cu, Fe2O3 and Fe3O4) and temperatures (200, 250 and 300 ºC) were performed to optimize the process at milder temperatures. Using Al as reductant and Pd/C as catalyst, yields as high as 38 % were obtained at 200ºC, compared with the highest yield (57 %) observed at 250 ºC. Thus a significant temperature reduction can be achieved using a suitable combination of reductant and catalyst. Using this reaction system, Pd/C as catalyst and Al as reductant, an extensive set of experiments at different times and temperatures were performed in order to determine the kinetics of the process and correlate them to a mathematical model of the process. The model correctly reproduces the experimental data with average errors lower than 5.9 %. These results demonstrate the feasibility of lower the operating temperature while maintaining the performance, when using an adequate combination of catalyst and reductant.

The hydroxides and oxyhydroxides of Ni are among the most extensively studied electrocatalysts because of their high activity in alkaline media [1,2], for instance, in the electrochemical oxidation of alcohols [3,4], aldehydes [5–10], glucose [11–14], hydrogen peroxide [15], water splitting [16–20], etc. There have been various expressions about their electrocatalytic mechanism. Sometimes, the electrocatalysis towards small organic/inorganic compounds occurred by the direct electrocatalysis of Ni(OH)2 [21–24] or the direct electrocatalysis of NiOOH [25–27]. However, most of the literature describes the electrocatalytic oxidation via a mediated mechanism [2,22,28] that is based on the Ni(OH)2/NiOOH redox transition in alkaline media, where the NiOOH formed during the charging acts as an oxidant for many organic/inorganic compounds [11,40]: (1.1) Ni O H 2 + OH - - e - ↔ N i O O H + H 2 O c h a r g i n g ↔ d i s c h a r g i n g (1.2) NiOOH + o r g a n i c / i n o r g a n i c c o m p o u n d → N i O H 2 + p r o d u c t However, the catalytic behaviour of Ni(OH)2/NiOOH in alkaline solutions is known to differ depending on the fine structural details, e.g., crystal structure, composition, etc. This is the reason why so much effort has been devoted to characterize the crystal phases present on catalyst surfaces during charging and discharging [29–31]. According to Bode’s diagram [29,32], Ni(OH)2 can exist in two crystal structures, in the literature denoted as “poorly-crystallised” α-phase or “well-crystallised” β-phase [29,33]. In the β-phase, the constituents are arranged in a hexagonal close-packed (hcp) layered structure of OH− ions with Ni occupying the octahedral interstices, with individual layers bonded by weak Van der Waals forces [29,33]. The structure of α-Ni(OH)2 is still considered to be Ni(II) hydroxide, but containing a variable excess of intersheet water [29,33]. Depending on the experimental method and the experimental conditions, either the α- or β-phases can be prepared via several chemical or electrochemical approaches [22,33]. Usually, α-Ni(OH)2 is isolated as a primary precipitation product, from which β-Ni(OH)2 might be obtained by ageing or potential cycling in alkaline solutions. However, β-Ni(OH)2 obtained from α-phase always contains adsorbed foreign ions or water [29]. Furthermore, the scheme presented by Bode et al. [34] also involves two phases of oxidized materials, β- or γ-NiOOH, that can be obtained through either the chemical or electrochemical oxidation of Ni(OH)2. β-NiOOH is a relatively well-defined material that crystallises in the hexagonal system and can be regarded as being derived from β-Ni(OH)2. γ-NiOOH crystallises in the rhombohedral system and represents a whole family of compounds exhibiting large intersheet distances where H2O or alkali ions (K+ or Na+) are intercalated. The γ-NiOOH system can be regarded as being derived from either α-Ni(OH)2 or β-NiOOH, once overcharge happens [22,29].Furthermore, numerous additional phases, i.e., structurally disordered phases, have been proposed and can be ascribed as structurally disordered β-Ni(OH)2/β-NiOOH with various possible structural irregularities, e.g., stacking fault disorder, α/β-interstratification, internal mechanical stress or ionic substitution, etc. [30,31,33]. All the disordered structures can be obtained from the β-Ni(OH)2 by an intercalation process and slab gliding [35] or by electrochemical potential cycling through the Ni(OH)2/NiOOH redox peaks in KOH solutions from free Ni or NiO2 surface sites [3,22 2]: (1.3) Ni + 2 OH - - e - → N i ( O H ) 2 , o r (1.4) Ni O 2 + H 2 O → N i ( O H ) 2 Contrary to α-/β-Ni(OH)2, they exhibit an X-ray diffraction pattern with very broad peaks, which means that its structure cannot be accurately determined [35].Several studies have linked structural changes in Ni(OH)2/NiOOH-based electrodes to their electrocatalytic performance (i.e., catalytic rate and input energy/onset potential) [36]. As presented by Bode [37], the phase transformation among different Ni(OH)2 and NiOOH species could proceed during charging and discharging in alkaline media, i.e., α-/γ- or β-/β- transformations. It is generally believed that the α-/γ-phases have been considered as the less catalytically active phases since the transformation involves more than one electron transfer, i.e., 1.6–1.67, suggesting that γ-NiOOH might include Ni+4 ions [38–40]. While still uncertain, the β-/β-phases are generally considered as more active since the transformation involves a 1-electron transfer (from Ni2+ to Ni3+) [1,41,42]. However, the transformations in between crystalline active phases (α ↔ γ or β ↔ β) possess some intrinsic disadvantages, such as low stability during the electrochemical reactions [2,20,43]. Recently, structurally disordered β-Ni(OH)2/β-NiOOH-based electrocatalysts have attracted attention because of their long cycle life, fast catalytic rate, and the low input energy needed for charging and discharging in alkaline media [44–46]. Their remarkable catalytic performance in electrochemical oxidation reactions can be associated with their disordered structure, i.e., β-Ni(OH)2/β-NiOOH with the lattice distortion or surface defects, etc. [47,33].The described redox transition between Ni(OH)2/NiOOH also plays a pivotal role in the electrochemical catalytic oxidation of formaldehyde (HCHO), as the latter is produced by the partial oxidation of methanol, which has technological significance in industrial catalytic processes [48], e.g., fuel-cell technology. The literature shows that interest has been focused on Ni-based electrocatalysts for HCHO oxidation as they possess a low overpotential and good durability to promote the HCHO oxidation reaction kinetics (i.e., catalytic rate) [5,7,8,10,49]. Also, they represent a good replacement for noble-metals-based electrocatalysts (Pt [50], Ag [51], Au [38,52], etc.) that were found to suffer from poor repeatability, reproducibility and the low sensitivity generated by surface poisoning from the adsorbed intermediates, i.e., carbon monoxide [17,53,54]. Recently, Ru–Ni–Ni(OH)2/NiO multi-metallic systems have been selected to be efficient non-precious catalysts [55,56] as their synergistic effects can significantly enhance the catalytic activity and stability. However, improving the catalytic activity of single metallic systems is of great interest for developing high-performance catalysts.Herein, we tailored the amount of highly active structurally disordered β-Ni(OH)2/β-NiOOH surface species via a CV-KOH modification process on two different electrodeposited Ni-based thin films; one with the surface NiO2–α-Ni(OH) for deposition performed at pH = 2.5, and one with the surface NiO2 for deposition performed at pH = 5.5. Based on the calculations of the amount of β-Ni(OH)2/β-NiOOH it was found that the KOH-modified Ni film (pH = 5.5) contains a larger amount of structurally disordered β-Ni(OH)2/β-NiOOH surface species (55% more) than the KOH-modified Ni film (pH = 2.5) where the surface is already covered by electrodeposited α-Ni(OH)2. A higher HCHO electrocatalytic activity, i.e., the lowest onset overpotential, increased the catalytic rate exhibited by the KOH-modified Ni film (pH = 5.5) due to the presence of a larger amount of structurally disordered β-Ni(OH)2/β-NiOOH surface species, as per the requirement for only 1 electron transfer, and with this connected low energy input. The KOH-modified Ni film (pH = 2.5) also performed as an efficient HCHO catalyst but less active than the KOH-modified Ni film (pH = 5.5) due to the intrinsically different surface composition consisting of some α-Ni(OH)2 after the Ni plating. A KOH-modified Ni film (pH = 2.5) transforms to both the active α-Ni(OH)2/γ-NiOOH (that requires the transfer of 1.67 electrons) and the highly active β-Ni(OH)2/β-NiOOH (that requires a transfer of only 1 electron) redox pairs.Ni-based electrodes (i.e., Ni thin films) were electrodeposited on the surface of a conductive SiO2/Au substrate that was prepared by the Physical Vapour Deposition (PVD) of the Au (600-nm-thick layer) on the glass. Before the PVD of Au, a 30-nm-thick layer of Cr was sputtered for better adhesion of the Au. The solutions used for the cyclic voltammetry and the cathodic electrodeposition were performed in 1-mol L−1 NiSO4·6H2O (98+%, Sigma-Aldrich). The pH was adjusted in the range of 2.5–5.5 by the addition of H2SO4 (99.9%, Sigma-Aldrich). The CV measurements were carried out in the potential range from −1.2 V to +0.4 V vs. Ag/AgCl at a scan rate of 50 mV s−1. The electrodeposition of the Ni thin films was carried out by applying a constant potential of −1.0 V for 300 s. Thin films were electrodeposited from the electrolyte with a pH = 2.5 and a pH = 5.5.Electrochemical measurements were performed with a Gamry Reference 600 potentiostat/galvanostat equipped PHE 200 software. The measurements were conducted in a Teflon electrolytic cell using a standard three-electrode system at room temperature. The working electrodes were “as-deposited” Ni thin films, the reference electrode was an Ag/AgCl/3.5-mol L−1 KCl (HANA Instruments GmbH-type HI5311), and a circular platinum mesh served as the counter electrode. The obtained output currents were normalized to the electrochemically active surface area (Aecsa) determined by the oxalate method (Supplementary data). The Aecsa of the Ni thin film (pH = 2.5) and Ni thin film (pH = 5.5) were determined to be 1.72 ± 0.05 cm2 and 2.18 ± 0.02 cm2, respectively.The KOH modification of the Ni thin films (pH = 2.5 and 5.5) was performed in an electrolyte composed of 0.5-mol L−1 KOH (pellets, Sigma-Aldrich) by cyclic voltammetry (CV). The cyclic voltammetric (CV) profiles were recorded from −1.0 V to +0.6 V in the 1st cycle and from 0 V to 0.6 V in the 2nd–60th cycle at a scan rate of 200 mV s−1. The electrochemical studies of HCHO oxidation were carried out in solutions with HCHO (37% w/v, Carbo Erba) concentrations ranging from 0.1-μmol L−1 to 20-mmol L−1 by cyclic voltammetry (CV) and chrono-amperometry (CA). The CVs were performed from 0 V to +1.0 V at a scan rate of 100 mV s−1. The pH of 13.7 was adjusted with the addition of NaOH pellets (Sigma-Aldrich). The electrolytes for the KOH-modification process and HCHO-detection experiments were de-oxygenated with pure N2 (for 15 min) and used immediately for every set of CV or CA measurements.The morphology of Ni thin films (pH = 2.5 and 5.5) was observed with a field-emission-gun scanning electron microscope (FEG-SEM, JEOL-JSM 7600F equipped with EDXS analysis) operating at 10 keV. The crystallinity and phase composition of the Ni thin films (pH = 2.5 and 5.5) were determined by analysing powder X-ray diffraction patterns (XRD, Bruker, D8 ADVANCE) with Cu-Kα radiation (λ = 1.5418 Å), with a step size of 0.01 s/° and a time per step of 600 s. Fourier-transform infrared spectroscopy (FT-IR) was performed to evaluate the presence of the –O-H chemical bonds on the surface of the dried Ni thin-film samples. The FT-IR spectra of the samples were taken using a Spectrum 100 spectrometer (Perkin Elmer, USA) in the wavenumber range from 4000 to 400 cm−1.In order to select the electrodeposition potential for the Ni-based thin films, cyclic voltammetry (CV) measurements were performed. The CV curves (Fig. 1 ) on the Au/Cr/SiO2 substrate, in 1-mol L−1 NiSO4 electrolyte were recorded in the potential range from the open-circuit potential (OCP, ~ 0 V) to −1.2 V in the forward (cathodic) scan and from −1.2 V to +0.4 V in the reverse (anodic) scan at a scan rate of 50 mV s−1. From Fig. 1A and B, it is evident that the voltammogram curves depend on the electrolyte pH. In the solution with the pH of 2.5 (Fig. 1A), the first anodic current increase at −0.4 V is attributed to the reduction of H3O+ (2H3O+ + 2e− →H2 + 2H2O), cH1, due to the low pH value. This reduction reaction is followed by a plateau at E ≈ −0.8 V that indicates the reduction of nickel (Ni2+ + 2e− →Ni0), cNi, and the hydrogen-evolution region via the water reduction at E > −1.0 V (2H2O + 2e− →H2 + 2OH−), cH2.The CV profile observed in the solution with the pH of 5.5 (Fig. 1B) was only characterized with a cathodic peak (cNi) at E ≈ −0.8 V, which indicates the reduction of nickel Ni2+ to Ni0 (Ni2+ + 2e− → Ni0). Also, from E > −1.0 V onwards, the evolution of hydrogen from water reduction (cH2) is taking place according to the reaction 2H2O + 2e− → H  + 2OH−. From the CV profiles (Fig. 1A and B) it is also evident that the output current is strongly influenced by the electrolyte pH. The current density of the cathodic peak cNi increases (6x) with a pH decrease (from 5.5 to 2.5), which indicates a change in the reaction mechanism for the electrodeposition of Ni [57]. Double the maximum current of the cathodic peak cH2 was observed at a pH of 2.5 (i = 18 mA at E = −1.2 V), which indicates a more intensive hydrogen-evolution reaction [57]. In agreement with the excessive 2H2O + 2e− → H2 + 2OH− reaction at pH = 2.5 (as seen from Fig. 1, A) there is also an increase of OH− ions in the vicinity of the electrode. Taking this as a fact and due to the presence of Ni2+ ions in the electrolyte, an (electro)chemical precipitation of the poorly soluble Ni(OH)2 might occur at the electrode surface, as was explained by Hall et al. [33]. Furthermore, from Fig. 1A, the fluctuations of the anodic output current were observed at E > −1.0 V which are induced by the pronounced evolution of H2 bubbles that affect the mass-transport as they block the surface of cathode [58–60].In the reverse scan, the anodic current peak aH at E = −0.015 V (Fig. 1A and B) was observed, which is attributed to the oxidation of the adsorbed hydrogen (formed during the forward scan at cH). However, the anodic peak aNi (at E = +0.15 V) that corresponds to the oxidation of Ni0 + 2OH− → Ni(OH)2 + 2e− was observed only in Fig. 1B. The absence of the anodic peak aNi on the CV presented in Fig. 1A was expected since the formation of Ni(OH)2 via the (electro)chemical precipitation takes place in the forward scan (as described above). From the CV measurements it was concluded that different surface compositions and morphologies of the Ni thin films can be achieved by applying negative potentials ≤ −1V. For this reason, the Ni thin films were fabricated from electrolytes with a pH of 2.5 and 5.5 when applying a constant potential of −1.0 V for 300 s. Fig. 2 A–D, shows the FEG-SEM images of the Ni thin films electrodeposited from two pH electrolytes (2.5 and 5.5) when applying a constant potential of −1.0 V. Fig. 2A and B, shows a typical morphology of the Ni thin film obtained by electrodeposition from the electrolyte with a pH of 2.5. As seen in Fig. 2A, the Ni thin film electrodeposited at lower pH (pH = 2.5), i.e., as-deposited Ni thin film (pH = 2.5), is composed of small crystalline nanoparticles (50–100 nm). The lines are present due to the topography of the underlying Au substrate. The image of the as-deposited Ni thin film (pH = 2.5) obtained at lower magnification (Fig. 2B) reveals that the Au substrate could not be entirely covered by Ni (Ni2+ + 2e− → 2Ni0) due to the extensive hydrogen formation at low (highly acidic) pH. The chemical composition of electrodeposited Ni thin film (pH = 2.5) was examined by EDS and XRD (explained below). The main detected element was Ni (91–95 at. %). Also, the residual amounts of oxygen (1–5 at. %) and gold (1–5 at. %) from the underlying substrate were detected which implies that the film is thin (<1 μm), as the interaction volume with using accelerating voltage of 10 keV is below 1 μm3. Fig. 2C and D, shows the typical morphology of the Ni thin film obtained by electrodeposition from the electrolyte with a pH of 5.5. As seen in Fig. 2C, the Ni thin film electrodeposited at higher pH (pH = 5.5) is composed of larger crystallites expanded to the size of 100–300 nm, i.e., as-deposited Ni thin film (pH = 5.5). The image of the as-deposited Ni thin film (pH = 5.5) obtained at lower magnification (Fig. 2D) indicates that the Ni deposit is homogeneously covering the Au substrate since hydrogen formation is less effective due to the higher pH. The chemical composition of electrodeposited Ni thin film (pH = 5.5) was also determined by EDS and XRD (explained below). The main detected element was Ni (97–99 at. %). Also, the residual amount of oxygen was detected (1–3 at. %). In comparison to Ni thin film (pH = 2.5), the deposit is thicker as the EDS did not show the presence of underlying Au substrate (the interaction volume with using accelerating voltage of 10 keV is below 1 μm3). The larger thickness might be the reason for irregular cracks in the Ni thin film, pH = 5.5 (see, for example, Fig. 2C and D). As stated in the literature [30,31,61,62], the cracked nature of the thicker Ni films is a common problem of wet chemical deposition methods, attributed to the drying contraction caused due to tensile stress. These findings are in agreement with those presented by Boubatra et al. [57]. They explained that the intensive hydrogen formation (favoured at lower pH) influences the electrochemical conditions in the vicinity of the cathode, changes the growth/nucleation processes and thus affects the final thickness and size distribution of the crystallites, e.g., a lower pH (2.5) decreases the grain size of the Ni electrodeposit by increasing the nucleation rate, or a higher pH (5.5) increases the grain size of the Ni by decreasing the nucleation rate [57,63–65].The influence of the electrolyte pH on the surface crystal structure of as-deposited Ni thin films (pH = 2.5 and 5.5) was analysed by XRD. Fig. 3 is an XRD pattern of the A) Au/Cr/SiO2 substrate, i.e., background, B) as-deposited Ni thin film (pH = 2.5) and C) Ni thin film (pH = 5.5) observed in the 2-theta range 10–90° with the insets showing the narrow 2-theta region 10–26°, 32–43° and 70–90° for better clarity. Rough indexing of the XRD pattern of the Au/Cr/SiO2 substrate displays the presence of fcc Au with three characteristics peaks at 38.2°, 77.5° and 81.7° corresponding to the standard Bragg reflections (111), (311) and (222) (ICDD 01-089-3697) [66]. The intense peak at 38.2°represents preferential growth in the (111) direction. The presence of bcc Cr was observed at 36.5° and 81.7° (ICDD 00-006-0694) [67] and a broad hump at 2-theta 20-25° indicates the presence of SiO2 (ICDD 00-033-1161) [68,69]. The representative characteristic peaks for the substrate were also identified from the XRD patterns of the as-deposited Ni thin film: pH = 2.5 (Fig. 3B) and the Ni thin film: pH = 5.5 (Fig. 3C). Furthermore, in the XRD diffractogram of the Ni thin films (pH = 2.5 and 5.5), two characteristic peaks for nickel were observed at 44.5° and 76.5° and were assigned to the standard Bragg reflections (111) and (220) of the fcc lattice (ICDD 00-001-1258) [57]. The intense peak at 44.5° represents preferential crystallographic growth in the (111) direction. Furthermore, significant differences in the crystal structure of the Ni thin films (pH = 2.5 and 5.5) were observed at 2-theta equal to 12°, 12.5°, 34°, 34,6° and 42.3°. These reflections indicate the presence of hexagonal ‘’quasi’’ close-packed [29] α-Ni(OH)2 (ICDD 00-038-0715 and ICDD 00-022-0444) only in the case of the Ni thin film (pH = 2.5). The presence of α-Ni(OH)2 is a result of the electrochemical deposition conditions (pH of electrolyte). As already described, during the cathodic electrodeposition of the Ni thin film (pH = 2.5) from a low-pH electrolyte, some of the current is consumed by hydrogen formation. This side reaction affects the local pH due to the formation of hydrogen and water at the cathode surface. Since the cathode is held at a very negative potential (E = −1.0 V), the intensive reduction of water occurs (2H2O + 2e− → H2 + 2OH−) and thus leads to the massive production of the OH−. Due to the low solubility of Ni(OH)2, (electro)chemical precipitation immediately occurs on the surface of the cathode [33,57].However, when using the XRD, we were not able to identify the peak at 2-theta equals 19° (Fig. 3B and C, insets) that can be attributed to the rhombohedral NiO2 (ICDD 04-012-0153) or hexagonal close-packed (hcp) β-Ni(OH)2 (ICDD 04-015-5276). It was assumed that the characteristic peak at 19° in the case of the Ni thin film: pH = 2.5 indicates the presence of rhombohedral NiO2, rather than the presence of β-Ni(OH), as it was prepared via the (electro)chemical precipitation process. The literature states [29,33] that hydroxides do not tend to form “well-crystallized” structures (i.e., β-Ni(OH)2) during (electro)chemical precipitation from aqueous salt (acidic) solutions. In the case the Ni thin film (pH = 5.5), we were not able to attribute any crystal phase to the characteristic peak at 19° from the XRD (the other representative characteristic peaks for NiO2 and Ni(OH)2 are overlapping with the Au or Ni). In addition, FT-IR analysis was performed. Fig. 4 has the FT-IR spectra of the (A) as-deposited Ni thin film (pH = 2.5) and (B) as-deposited Ni thin film (pH = 5.5). In order to examine the presence of NiO2 or Ni(OH)2, the FT-IR spectra were observed in the wavelength range from 400 to 4000 cm−1. Fig. 4A shows typical FT-IR spectra of the as-deposited Ni thin film (pH = 2.5). The inset spectrum, which represents the wavelength region from 400 to 700 cm−1, indicates the presence of the Ni–O–H vibration peak at 610 cm−1. Furthermore, the band located at 3000–3600 cm−1 indicates the presence of the symmetric –O–H vibrations and the absorption peak for the –OH (hydroxyl) functional group at 1720 cm−1 [31,70]. From these results, the presence of Ni(OH)2 on the Ni thin film (pH = 2.5) was additionally confirmed. Fig. 4B shows the FT-IR spectra of the as-deposited Ni thin film (pH = 5.5) that was measured following the same procedure as the as-deposited Ni thin film: pH = 2.5 (Fig. 4A). The FT-IR analysis did not reveal any vibration bands that would be typical for Ni(OH)2 being present on the Ni thin film: pH = 5.5) (i.e., the absence of two of the most representative –O–H vibration peaks at 3000–3600 cm−1 and 1720 cm−1). The findings obtained from the FT-IR measurements (Fig. 4B) confirmed the absence of Ni(OH)2. By this observation, we concluded that the surface of Ni thin film (pH = 5.5) is covered with the NiO2 (an XRD characteristic peak at 19°, Fig. 3C) formed by surface passivation once it is exposed to air and/or water molecules [2,22,46].In order to remove the native NiO2 [2,46] and build up the catalytically more active structurally disordered β-Ni(OH)2/β-NiOOH surface species [34], potential cycling in an alkaline electrolyte (i.e., the KOH-modification process) was introduced to the Ni thin films (pH = 2.5 and 5.5). This experimental step is mandatory in the case of the Ni thin film (pH = 5.5) because a system consisting of native NiO2 without the presence of Ni(OH)2 on the surface does not show catalytic activity towards the HCHO oxidation [12,46]. Fig. 5 presents the CV profiles of (A) the as-deposited Ni thin film (pH = 2.5) and (B) the as-deposited Ni thin film (pH = 5.5) observed in 0.5-mol L−1 KOH at a scan rate 200 mV s−1. The scan rate of 200 mV s−1 was selected based on our previous research [43,46,71,72]. Cycles 1–60 are included in order to show the evolution of CVs behaviour with the increasing number of cycles. Fig. 5, A-1 and B-1, demonstrate CV profiles in the 1st cycle where the potential window was chosen from −1.0 V to +0.7 V in the anodic region (forward scan) and from 0.7 V to 0.2 V in the cathodic region (reverse scan). The first increase in the anodic current density (a1) was observed at the potential −0.7 V (1-A)/-0.5 V (1-B) and corresponded to the formation of α-Ni(OH)2. The α-Ni(OH)2 was formed according to the following reactions: Ni + 2 OH− → α-Ni(OH)2 [3,22] or NiO2 + H2O → α-Ni(OH)2 [22]. The CV profiles in the 1st cycle demonstrate that the values of the anodic peak potential (Ea1) vary between the Ni thin films (pH = 2.5 and 5.5). The shift of the a1 peak potential is a result of an electro-crystallisation process, where the electrochemical reaction (electron transfer) is accompanied by atomic rearrangements (e.g., from fcc Ni0 or rhombohedral NiO2 structures to a brucite type of ‘’quasi’’-hcp α-Ni(OH)2 structure [73]). As the chemical composition and crystal phase of the surface species vary between the as-deposited Ni thin films: pH = 2.5 and 5.5 (Figs. 3 and 4), each of these electrodes needs a certain driving force (overpotential) for electro-crystallisation [74,75]. Thus, the electro-crystallisation process is probably the main reason for a shift of the a1 peak potential in between the Ni thin films (pH = 2.5 and 5.5). Furthermore, the CV plots (Fig. 5, 1-A and 1-B) reveal a difference in the observed a1 current densities: j (a1, Ni thin film: pH = 2.5) ≈ 0.01 mA cm−2 (Fig. 5, 1-A) and j (a1, Ni thin film: pH = 5.5) ≈ 0.5 mA cm−2(Fig. 5, 1-B). This observation indicates that only a small amount of α-Ni(OH)2 is formed at a1 on the surface of the as-deposited Ni thin film (pH = 2.5), since the surface is already occupied with α-Ni(OH)2 after electrodeposition (proved by the XRD and FT-IR, Figs. 3 and 4). On the other hand, the larger increase in the anodic current peak a1 was observed in the case of Ni thin film (pH = 5.5), demonstrating the conversion of the surface Ni/NiO2 (Figs. 3 and 4) to α-Ni(OH)2.With the continuation of the potential cycling to the more positive potentials (from −0.4 V to +0.4 V) the chemical dehydration of α-Ni(OH)2 to β-Ni(OH)2 (a2) occurs [34]. Once the β-Ni(OH)2 is formed, it cannot be electrochemically reduced back to α-Ni(OH)2/NiO/Ni(metallic) as large atomic rearrangements are no longer expected due to the high electrochemical stability of the β-phase [9].After the potential cycling to the more positive potentials, the next increase in the current densities was observed at a3 (E ≈ 0.4 V for Ni thin film: pH = 2.5 and E ≈ 0.45 V for Ni thin film: pH = 5.5). The a3 represents the oxidation of the β-Ni(OH)2 (formed at a2) to β-NiOOH according to the reaction: β-Ni(OH)2 + OH− → β-NiOOH + H2O + e−. As seen from Fig. 5, A-1 (Ni thin film: pH = 2.5), the j(a3) reaches higher values than the j(a1). This means that the formation of surface β-NiOOH (i.e., α-Ni(OH)2 (formed at a1) → β-Ni(OH)2 (formed at a2) + OH− → β-NiOOH (formed at a3) + H2O + e−) can also occur through the small amount of electrodeposited α-Ni(OH)2 if dehydrated. However, in the case of the Ni thin film: pH = 5.5 (Fig. 5, 1-B), the j(a3) < j(a1) indicating the incomplete transformation of α-Ni(OH)2 (formed at a1) to β-Ni(OH)2 (formed at a2). However, once the a3 current peak (Fig. 5, A-1 and B-1) was observed, the scans were immediately reversed back to +0.0 V in order to avoid the production of molecular oxygen (4OH− → O2 + H2O + 4e−). The formation of O2 is undesirable due to the possible adsorption on the surface of the Ni thin films [4,46] and thus blocking the surface-active Ni sites [4,72]. In the cathodic region, one cathodic peak (c1) was observed (Fig. 5, A-1 and B-1). This cathodic peak (c1) at E = 0.3 V (Ni thin film: pH = 2.5) and E = 0.25 V (Ni thin film: pH = 5.5) corresponds to the reduction of β-NiOOH (formed at a3) to β-Ni(OH)2. Based on the above-described results, it was assumed that after the 1st cycle (Fig. 5, A-1 and B-1) the surface of the Ni thin film (pH = 2.5) is still (mostly) covered by α-Ni(OH)2 (formed directly during Ni electrodeposition) and by a small amount of β-Ni(OH)2 (formed at a3). The same surface composition (i.e., α-Ni(OH)2 and small/negligible amounts of β-Ni(OH)2) were ascribed to the Ni thin film (pH = 5.5); however, the α-Ni(OH)2 was formed at a1 in the 1st cycle of the KOH-modification process.Since the crystallinity of the Ni(II) hydroxides can be controlled from highly crystalline (β-phase) to structurally disordered hydroxides by tailoring the experimental conditions, we selected the continuous CV cycling in KOH. The literature indicates that continuous potential cycling over the Ni(OH)2/NiOOH redox peaks leads to the transformation of well-crystalline β- Ni(OH)2 to a large set of disordered β-Ni(II) hydroxides with a variable excess of intersheet water, stacking-fault disorder or mechanical stresses [2], which proved themselves in enhanced electrocatalytic HCHO oxidation [16,44–46,76]. Due to this, the Ni thin films (pH = 2.5 and 5.5) were KOH modified in KOH up to 60 times. Fig. 5 shows the CV profiles of the A-2) Ni thin film – pH = 2.5 and B-2) Ni thin film – pH = 5.5 observed in the potential range from 0 V to +0.6 V from the 10th to 60th cycle. From the CV plots, the anodic peak (a3) was observed at E = 0.4 V (Ni thin film: pH = 2.5) and E = 0.45 V (Ni thin films: 5.5) in the forward scan. The a3 corresponds to the formation of the β-NiOOH (most probably structurally disordered) due to the diffusion of OH− to the β-Ni(OH)2 (formed in the 1st cycle) or to the NiO2 [4,5,14] according to the reactions: β-Ni(OH)2 + OH− → β-NiOOH + H2O + e− or/and NiO2 + H2O -> β-Ni(OH)2 + OH− → β-NiOOH + e−, respectively. With increasing numbers of cycles from 10 to 60, the current densities of the anodic (a3) and cathodic (c1) peaks increase until a steady state is reached and there is no significant current-density increase for the cycles ≥ 50 ( Δ j 50–60 cycles ≈ 0). From the steady-state on, no further changes in the surface composition are expected [46]. In the reverse scan, the cathodic (c1) peak was observed for both Ni thin films and corresponds to the reduction of β-NiOOH (formed at a3 from the 10th to 50th cycle) to β-Ni(OH)2: β-NiOOH + H2O + e− → β-Ni(OH)2 + OH.The KOH-modified Ni thin films (after 60 cycles): pH = 2.5 and 5.5 (Fig. 5) were analysed by XRD in order to investigate their surface crystal structure after the KOH-modification process. Fig. 6 A has the XRD patterns of the KOH-modified (green curve) and as-deposited (grey curve) Ni thin film (pH = 2.5) observed in the 2-theta ranges 10–25° (left) and 32–36° (right). The XRD diffractogram of the KOH-modified Ni thin film – pH = 2.5 consists of hcp α-Ni(OH)2 (ICDD 00-022-0444 [29]) with the three characteristics peaks at 2-theta of 12°, 24°, 34° and 34.6° and rhombohedral NiO2 (ICDD 04-012-0153) with the characteristic peak at 19°. Also, indexing this XRD pattern reveals the presence of the substrate, i.e., SiO2 (ICDD 00-033-1161), at 20–25°. As seen from Fig. 6A, the XRD pattern of KOH-modified Ni thin films – pH = 2.5 (green curve) and as-deposited Ni thin films – pH = 2.5 (grey curve) are similar. After the KOH-modification process, only a mild increase in the intensity of the α-Ni(OH)2 characteristic peaks is observed, indicating a slight enrichment of the amount of α-Ni(OH)2 on the surface. Fig. 6B (blue curve) shows the XRD pattern of the KOH-modified Ni thin film (pH = 5.5) observed in the 2-theta ranges 10–25° (left) and 32–36° (right). In order to demonstrate the changes in the composition of surface species after the KOH-modification process, the graphs also include the XRD patterns of the as-deposited Ni thin films – pH = 5.5 (black curves). From the XRD pattern of the as-deposited and KOH-modified Ni thin film – pH = 5.5 (Fig. 6, B), the following diffraction peaks were observed: characteristic peak for rhombohedral NiO2 (ICDD 04-012-0153) at 19° and broad hump for SiO2 (i.e., substrate) at 20–25° (ICDD 00-033-1161). The significant differences between the as-deposited (black curves) and KOH-modified (blue curves) Ni thin film (pH = 5.5) were observed in the 2-theta regions 23–25° (left) and 33.5–34.5° (right). The XRD pattern of the KOH-modified Ni thin film – pH = 5.5 (blue) indicates the presence of two different electrochemically assembled hcp α-Ni(OH)2 with the characteristic peaks at 12.5°, 24.2°, 34° and 34.5° that can be attributed to the 3Ni(OH)2·2H2O (ICDD 00–022-0444), and characteristic peaks at 12°, 23.5°, 33.9°and 34.4° that can be ascribed to the Ni(OH)2·0.75H2O (ICDD 00-038-0715).Furthermore, the indexing of the XRD patterns confirmed the absence of crystalline β-Ni(OH)2 that is formed at a3 (Fig. 5, A-2 and B-2) [3]. The absence of Ni(OH)2 (formed at a3) during potential cycling, Fig. 5, A-1 and B-1) reflection in the XRD was attributed to its disordered structure that resulted from potential cycling in KOH for up to 60 cycles [2,46] and was already confirmed in our previous research by TEM and FT-IR [46,72].In order to determine the amount of the structurally disordered β-Ni(OH)2/β-NiOOH (formed at a3 during the potential cycling for up to 60 cycles) on the surface of the Ni thin films (pH = 2.5 and 5.5), the number of monolayers was calculated according to the equation (3.1): (3.1) Γ Nifilm × 1 m o n o l a y e r Γ standard where Γ standard is the number of moles per square centimetre of structurally disordered β-NiOOH [mol cm−2] for 1 monolayer, i.e., 1.06 × 10−9 mol cm−2 (obtained by Bode [14,37]). The Γ Nifilm [mol cm−2] was calculated using the equation (3.2): (3.2) Q = n F A ecsa Γ Nifilm where Q is the charge [As = C], n is the number of electrons, F is the Faraday constant (96485.33 As mol−1) and A ecsa is the electrochemically determined surface area of the Ni thin films (Supplementary data). The charge was calculated based on the area ( idE ) under the reduction peak of the β-NiOOH to β-Ni(OH)2 (c1) (3.3): (3.3) Q = ∫ i d τ = 1 υ ∫ i d E where ν is the scan rate [V s−1] and i [A] is the peak current. The reduction peak (c1) was selected as it provides the charge required to fully reduce an oxyhydroxide and does not overlap any other electrochemical processes [14,33,71]. Thus, the Ni thin film (pH = 2.5) with an A ecsa of 1.72 cm2, and a cathodic charge of 6.7 × 10−5C, the number of monolayers is calculated to be 0.4, depending on the overall electrochemically active surface area. In the case of the Ni thin film (pH = 5.5) with a A ecsa of 2.19 cm2, and a cathodic charge of 2.2 × 10−4C, the number of monolayers is calculated to be 0.9, depending on the overall electrochemically active surface area. The fact that the calculated values were determined to be < 1 means that the structurally disordered β-Ni(OH)2/β-NiOOH most probably covers small patches and not the entire electrochemically active surface area of both Ni thin films (pH = 2.5 and 5.5).Based on the XRD (Fig. 6) and the CV behaviour in KOH (Fig. 5), and the calculated amount of disordered-Ni(OH)2, a possible mechanism for the surface transformation during the KOH-modification process for Ni thin films (pH = 2.5 and 5.5) is as follows. The mechanism for the surface transformation of the Ni thin film – pH = 2.5 is presented in Fig. 7 . As can be seen from Fig. 7a, the surface of the Ni thin film is covered by electrodeposited α-Ni(OH)2 before the KOH-modification process. When the electrodeposited Ni thin film (pH = 2.5) is placed in the KOH solution and exposed to prolonged cycling over the oxidation and reduction potential range (Fig. 7b), a small increase in the amount of α-Ni(OH)2 (formed at a1 in the 1st cycle, Fig. 5, A-1) is observed (proved by XRD, Fig. 6A). As the conditioning continues by potential cycling (up to 60 cycles), the amount of structurally disordered β-Ni(OH)2/β-NiOOH surface species (formed at a3, Fig. 5, A-2) does not increase significantly (Fig. 7c), as we observed a small current-density increase ( Δ j (a3, up to cycle 60) < 0.07 mA cm−2) (Fig. 5, A-2). In addition, the calculated value of 0.4 for the monolayer confirmed that the amount of structurally disordered-Ni(OH)2 is small compared to the α-phase after the KOH-modification process.For the Ni thin film – pH = 5.5, the KOH-modification process influences the surface composition of the as-deposited Ni thin film (pH = 5.5) since the surface of the electrodeposited film is covered with Ni/NiO2 (Fig. 8 a). The changes take place in the 1st cycle (Fig. 5, A-1) at a1 (the current–density increase ≈ 0.5 mA cm−2) where α-Ni(OH)2 is formed by the diffusion of OH− into Ni/NiO2. Also, the potential cycling in KOH more than 60 times (Fig. 8b) increases the amount of structurally disordered β-Ni(OH)2/β-NiOOH surface species due to the larger increase in the current density at a3 (Fig. 5, A-2: Δ j (a3, up to cycle 60) < 0.25 mA cm−2) and the calculations, 0.9 monolayer. From these observations, it was concluded that the electrochemically active surface of the Ni thin film (pH = 5.5) is mostly composed of structurally disordered β-Ni(OH)2/β-NiOOH surface species (Fig. 8c).A formaldehyde oxidation ability of all the Ni thin films was investigated via cyclic voltammetry. Fig. 9 a shows the CV profiles of as-deposited (dot curve) and KOH modified (solid curve) Ni thin films (pH = 2.5), both observed in 1-mmol L−1 HCHO and 0.1-mol L−1 NaOH at a scan rate of 100 mV s−1. From the CV profiles of the as-deposited and KOH-modified Ni thin film (pH = 2.5), a well-defined anodic peak (aHCHO) was observed at E = 0.75 V and E = 0.70 V, respectively. Also, a rapid increase of the anodic current density (aoxygen) at E > 0.80 V that correspond to the oxygen-evolution reaction. The aHCHO current increase (jas-deposited  = 0.05 mA cm−2 and jKOH-modified  = 0.07 mA cm−2) is a result of two parallel reactions [6–10,46] (mediated electron-transfer mechanism): oxidation of Ni(OH)2 (i.e., α-/structurally disordered-β-Ni(OH)2 + OH− → γ-/structurally disordered-β-NiOOH + e−) and the formation of “new” Ni(OH)2 (i.e., γ-/structurally disordered-β-NiOOH + HCHO → α/β-structurally disordered-Ni(OH)2 + CO2 + H2O). In the cathodic region, a cathodic peak (c1) was observed at E = 0.68 V for the as-deposited Ni thin film: pH = 2.5 and at E = 0.4 V for the KOH-modified Ni thin film: pH = 2.5. The cathodic current increase (c1) was attributed to the reversible transformation of the γ-/structurally disordered β-NiOOH to the α/β-structurally disordered Ni(OH)2. Fig. 9b presents the CV profiles of the as-deposited (dot curve) and KOH-modified (solid curve) Ni thin films (pH = 5.5), both observed in 1-mmol L-1 HCHO and 0.1-mol L−1 NaOH at a scan rate of 100 mV s−1. The CV response for the as-deposited Ni thin film: pH = 5.5 (dot curve) shows a complete electrode’s inactivity towards HCHO due to the absence of the HCHO oxidation peak (aHCHO). An almost negligible electrocatalytic activity for the HCHO oxidation of the as-deposited Ni thin film (pH = 5.5) can be ascribed to the presence of the native NiO2 on the surface of the film (proved by XRD, FT-IR). However, the CV profile of the KOH-modified Ni thin film: pH = 5.5 (solid curve) revealed the presence of an anodic peak (aHCHO) at E = 0.67 V, j = 0.24 mA cm−2 and a cathodic peak (c1) at E = 0.15 V, j = −0.23 mA cm−2. This result shows that the KOH-modified Ni thin film (pH = 5.5) is also oxidized HCHO via mediated mechanisms (as explained above).In order to be able to make a direct comparison of the produced Ni thin films for HCHO oxidation catalytic ability, the output currents were normalized to the electrochemically active surface areas (Supplementary data). From Table 1 it is clear that the increased current densities of the aHCHO and the decreased HCHO oxidation onset potentials (or aHCHO peak potentials) were achieved in the following sequence: as-deposited Ni thin film (pH = 5.5) = 0 < as-deposited Ni thin film (pH = 2.5) < KOH-modified Ni thin film (pH = 2.5) < KOH-modified Ni thin film (pH = 5.5).The Tafel slopes (Fig. 10 ) for the as-deposited Ni thin film (pH = 2.5), KOH-modified Ni thin films (pH = 2.5 and 5.5) were examined to describe the influence of the onset potential on the steady-state current density. The as-deposited Ni thin film (pH = 5.5) consisting of NiO2 surface species was excluded since it shows complete catalytic inactivity towards HCHO in alkaline media. The Tafel plots, which reflect the charge-transfer kinetics, were determined by fitting the CV data (Fig. 9a and b) to the equation η = a + b × log j , where η is the over-potential, b is the slope of the Tafel curve and j is the current density [20]. The observed slope of the KOH modified Ni thin film: pH = 5.5 (blue) is 69 mV dec−1 smaller than the KOH modified Ni thin film: pH = 2.5 (green), and 141 mV dec−1 smaller than the as-deposited Ni thin film: pH = 2.5 (wine). These results indicate that the KOH-modified Ni thin film (pH = 5.5) shows improved kinetics and thus exhibits the highest catalytic activity towards HCHO oxidation with the highest current density (0.24 mA cm−2) and the lowest onset potential (0.44 V), which is lower than the reported values of 0.55 V for the nanoporous-NiPh-modified electrode [49] or 0.5 V for the Ni/IL/CPE [8], Ni(OH)2/POT (TX-100)/MCNTPE [7] and Ni/P-nanozeolite-modified electrode [10]. Assuming the Bode’s diagram [34] is valid for α-Ni(OH)2/γ-NiOOH transformation, our results suggest that the poorer electro-catalytic activity of KOH-modified Ni film (pH = 2.5) is due to the presence of the smaller amount of highly active disordered β-Ni(OH)2/β-NiOOH surface species and higher amount of α-Ni(OH)2/γ-NiOOH surface species as the α–γ transformation requires a higher input energy for the transfer of 1.6–1.67 electrons in HCHO electrocatalytic oxidation. Upon that an enhanced electro-catalytic activity (low onset overpotential of 0.44 V) of the KOH-modified Ni thin film (pH = 5.5) is attributed to a larger amount of highly active structurally disordered β-Ni(OH)2/β-NiOOH surface species which enable a transfer of only 1 electron during the charging and discharging of β-Ni(OH)2/β-NiOOH in HCHO alkaline electrolyte. Hence, this β-Ni(OH)2/β-NiOOH transformation consumes less input energy needed for electrocatalytic HCHO oxidation.This study proposes a mechanism for the formation of the highly active, structurally disordered β-Ni(OH)2/β-NiOOH redox pair on the surface of two different Ni thin films, produced at pH = 2.5 and pH = 5.5. A highly active structurally disordered β-Ni(OH)2/β-NiOOH species were produced from the available surface NiO2 via KOH modification process that encompasses potential cycling through the NiOOH/Ni(OH)2 redox peaks in KOH up to 60 cycles. Based on the quantitative analysis of β-Ni(OH)2/β-NiOOH (i.e., calculations based on electric charge) it was shown that the amount of β-Ni(OH)2/β-NiOOH surface species plays a crucial role in the electrocatalytic oxidation of HCHO in alkaline media. The experimental data showed the enhanced electrocatalytic activity with the lowest onset overpotential of 0.44 V vs Ag/AgCl for the KOH-modified Ni film (pH = 5.5). The improved electrocatalytic activity in later case is attributed to the higher amount of structurally disordered β-Ni(OH)2/β-NiOOH species which enable a transfer of only 1 electron during the charging and discharging, which in comparison to α-Ni(OH)2/γ-NiOOH thus consumes less input energy. We have shown that via finetuning of the character of the Ni-surface species, we can tune their electrochemical and catalytic performances. Špela Trafela: Investigation, Methodology, Formal analysis, Conceptualization, Writing - original draft. Sašo Šturm: Funding acquisition, Writing - review & editing. Kristina Žužek Rožman: Supervision, Conceptualization, Funding acquisition, Writing - review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This study was supported by the Slovenian Research Agency through the J2-8182 and the PR-06805 research projects and the and P2-0084 program, of which this investigation forms a part. The corresponding author gratefully appreciates the financial support from the COST action MP1407 that provided a scholarship for four training-school events in the e-MINDS project. Spela Trafela would like to thank L’Oréal ADRIA and the Slovenian National Commission for UNESCO for a scholarship awarded by the National Programme for Women in Science in 2020.Supplementary data to this article can be found online at https://doi.org/10.1016/j.apsusc.2020.147822.The following are the Supplementary data to this article: Supplementary Data 1

The main challenge with electrocatalysis is finding low-cost electrocatalysts that can work efficiently to oxidize the HCHO. Here, we propose a mechanism for the voltammetric formation of a highly active, structurally disordered β-Ni(OH)2/β-NiOOH redox pair on the surface of electrodeposited Ni thin films to achieve an extraordinary catalytic performance with respect to HCHO oxidation in alkaline media. We report electrochemical, XRD and FT-IR measurements on as-deposited and voltammetrically treated (i.e., KOH-modified) Ni thin films, and calculations based on the electrical charge to investigate the changes in the surface composition, crystal structure and related HCHO oxidation activity. We found that the KOH-modification process plays a crucial role in the formation of surface highly active, disordered β-Ni(OH)2/β-NiOOH. The KOH-modified Ni film with the largest amount of the structurally disordered β-Ni(OH)2/β-NiOOH resulted in improved catalytic performance, i.e., an onset overpotential reduced by 400 mV and a catalytic rate increased by 69 mV dec−1. The presented technique has a wide range of applications and provides advances with a novel design idea and a new synthesis strategy for the preparation of highly active, structurally disordered Ni(OH)2/NiOOH redox systems on the surface of Ni thin films and other Ni-based nanostructured electrocatalysts for HCHO oxidation.

The deterioration of the global environment and depletion of fossil fuel resources have driven the global community to search for alternative, sustainable, and eco-friendly energy resources [1]. Among the available energy resources, H2, which has attracted significant attention as a clean resource, is extensively utilized in various applications such as in fertilizer, food, and petrochemical industries [2,3,39]. Currently, H2 is mainly produced by the steam reforming of fossil fuels such as coal, oil, and natural gas, because they are currently the most economical resources [3,4]. However, because fossil fuels are limited energy resources that produce greenhouse gases, there has been a strong demand for developing alternatives to the current H2 production methods [2–4].Biomass is one of the promising alternative resources owing to its benefit as a carbon-neutral source [2,4,5]. Terrestrial biomass (e.g., sugar- and starch-based crops, lignocellulose, and agricultural residues) and marine biomass (e.g., algae) are regarded as potential energy sources. However, the terrestrial biomass has the competition issue as the usage for foods and as the agricultural land, and the use of fertilizers and water for producing terrestrial biomass causes environmental issues. In comparison, macroalgal biomass has many advantages as an energy source, including less competition as a food source and land for growth, high productivity owing to a short growth cycle, and efficient carbon dioxide fixation (Scheme 1 ) [6]. Therefore, H2 production using macroalgae can be an attractive alternative to carbon neutralization.Hydrogen can be produced from the hydrocarbon using various methods, including the primary techniques of steam reforming, partial oxidation, and autothermal reforming [2–4]. Among these processes, steam reforming is regarded as an economical process owing to its low-temperature operation [2–4]. Water-soluble bio-oils, which are thermochemically converted by fast pyrolysis or hydrothermal liquefaction of biomass, are used as a feed for the steam reforming of biomass [2,4]. As most studies have been devoted to the steam reforming of terrestrial biomass [2,7–10], the steam reforming of marine algal biomass is still limited. In particular, because many mineral components derived from marine biomass would be harmful to the subsequent steam reforming reaction, the post-treatment of liquefied bio-oils should be carefully considered.The steam reforming of hydrocarbons has been suggested to take place owing via two reactions, steam reforming and water gas shift reactions, as shown in Eqs. (1) and (2) [2,10]. (1) C n H m O k + ( n − k ) H 2 O → n C O + ( n + m 2 − k ) H 2 (2) nCO + n H 2 O → nC O 2 + n H 2 The overall reaction can be summarized as, (3) C n H m O k + ( 2 n − k ) H 2 O → n C O 2 + ( 2 n + m 2 − k ) H 2 Ni-based catalysts have been utilized in steam reforming because of their high activity for C–C bond breakage (active for Eq. (1)) and low cost [11,12]. However, they are severely deactivated by coke deposition and sintering [10,13]. Many studies have been devoted to improving the stability of Ni-based catalysts using supports and additives. One of the studied methods is to use hydrotalcite-derived mixed oxide catalysts. Hydrotalcite, which belongs to a family of anionic clays, is a layered double hydroxide (LDH) of Al and Mg consisting of anions and water in the interlayers [14,15]. The structure of the hydrotalcite is closely related to that of brucite, Mg(OH)2, in which certain divalent cations in the layer structure are replaced by trivalent cations. The partial substitution of Mg2+ and/or Al3+ with other cations results in materials with isomorphous structures known as hydrotalcite-like compounds (HTLCs) [14]. The thermal decomposition of HTLC precursors leads to the formation of mixed oxides, resulting in good dispersion of metal cations, high thermal stability, and high surface area compared with those obtained from direct methods. Ni–Mg–Al mixed oxides derived from hydrotalcite exhibit superior activity and stability compared to Ni/Al2O3 and Ni/MgO in the reforming of hydrocarbons [16,17] and pyrolysis oil of biomass [18]. The second method is to add promoters to the Ni-based catalysts. For instance, Li et al. reported enhanced catalytic performance of Ni/MgAl hydrotalcite-like compounds by adding Fe or Cu in the steam reforming of biomass tar derived from cedar wood (terrestrial biomass) [9,19]. Especially, Cu addition improved the metal dispersion, oxidation ability, and coke resistance of the Ni/MgAl catalyst [9]. In particular, Cu has been reported to be active in the water gas shift reaction (Eq. (2)) and thus will improve H2 selectivity [20–22]. Although NiCu hydrotalcite-derived catalysts are shown to be the promising catalysts for steam reforming of terrestrial biomass-derived hydrocarbons, the investigation over the steam reforming of marine algae-derived hydrocarbons (bio-oil) is limited. Considering that hydrocarbon compositions difference between model compounds and real bio-oil critically affect the catalytic performance of the steam reforming [7], bio-oil derived from macro algae would be different over the NiCu hydrotalcite-derived catalysts.In this study, we studied hydrothermally liquefied bio-oil derived from S. japonica (macroalgae) as a clean resource for hydrogen by steam reforming reaction over NiCu hydrotalcite-derived mixed oxide catalysts (Scheme 1). A desalting process was performed to remove the minerals remaining in the raw bio-oil. Thereafter, we controlled the Ni/Cu composition of the NiCuMgAl catalysts and performed the steam reforming of the refined liquefied oil for H2 production.The hydrothermal liquefaction of brown algae (S. japonica) was performed under 300 °C, 2 h, autogenous pressure in an autoclave reactor including macro algae (Saccharina japonica) and distilled water at a 1:10 wt ratio [23]. The chemical components of liquefied oil were analyzed by gas chromatography-mass spectrometry (GC–MS, 7890 GC/5975C MSD, Agilent, USA). The elemental composition (C, H, and O) and water content were analyzed by elemental analysis (Unicube/rapid OXY cube, EA KOREA, KOR) and Karl Fischer titration.Desalted oil was prepared by adsorption process using the glass filter reactor (I.D.: 35.0 mm, O.D.: 38.0 mm, length: 890 mm, Vol.: 1.1 L) which was fed into a reactor by HPLC pump (Young Lin Instrument). We used H+ ion resin (SCR-BH, Samyang, KOR), OH− ion resin (SAR10MBOH, Samyang, KOR), and amberlyst-15 (Sigma-Aldrich, USA) ion-exchange resin. Desalting process was performed by combination H+ ion resin (500 mL) and OH− ion resin (500 mL) in two of glass reactor. The liquefied oil was fed in glass reactor (2 mL/min), after passed the resins, the oil mixed in the batch reactor for 2 h with ablerlyst-15. The mineral contents (Na, K, Ca, Mg, Si, Fe, and P) in liquefied oil were determined by the plasma atomic emission method (inductively coupled plasma atomic emission spectrophotometer, ICP-OES, Optima 7400DV, PERKIN ELMER).The chemicals used for obtaining the NixCu1.5–xMg1.5Al1.0 catalyst were Ni(NO3)2·6H2O (Junsei, 97.0%), Mg(NO3)2·6H2O (Katayama, 99.0%), Al(NO3)2·9H2O (Junsei, 97.0%), and Cu(NO3)2·3H2O (Junsei, 99.0%). The gases utilized in the steam reforming study were H2 and Ar (J.B. Korea Gases Co. Ltd., >99.999%).A series of NixCu1.5–xMg1.5Al1.0 catalysts were prepared by the co-precipitation of the nitrates present in the metal components [24]. An aqueous solution of Ni(NO3)2·6H2O, Cu(NO3)2·3H2O, Mg(NO3)2·6H2O, and Al(NO3)2·9H2O was slowly added into a beaker containing an aqueous solution of Na2CO3 (2 M) that was stirred at room temperature to achieve a constant pH of 10 ± 0.5. The pH of the solution was adjusted with an aqueous solution of NaOH (2 M). The resulting suspension was kept at 60 °C for 16 h. The precipitate was filtered, washed several times with deionized water, and dried at 110 °C for 12 h. Thereafter, the precipitate was ground to fine powders and calcined at 600 °C for 6 h in a static air atmosphere. Subsequently, the obtained material was pressed to a disk, crushed, and sieved to particles with 0.4–0.6 mm diameter (30–40 mesh size). The catalysts are denoted as NixCu1.5–xMg1.5Al1.0. The molar ratio of (Ni + Cu + Mg)/Al was fixed at 3 and the molar ratio of Ni/(Ni + Cu) varied from 0 to 1.0. The total moles of Ni and Cu were fixed at 1.5.After calcination, the catalysts were reduced under 10% H2/Ar (v/v) for 5 h at 550 °C and cooled down to room temperature under N2 atmosphere. The catalyst surface is passivated under 1% O2/He for 24 h and the catalysts are denoted as “reduced” catalysts.X-ray fluorescence (XRF) were obtained using X-ray fluorescence spectrometer (Shimadzu, XRF-1800, Japan) equipped with Rh-Ka radiation at 4 kW.X-ray diffraction (XRD) were obtained using an X-ray diffractometer (PHILIPS, X'Pert-MPD System, Netherlands) equipped with Cu–Kα radiation (λ = 0.15406 nm) at 45 kV, 40 mA.The H2-temperature programmed reduction (H2-TPR) was conducted in the AutochemⅡ 2920 (Micromeritics Instrument Corp., USA). The materials were first preheated at 100 °C for 2 h, and then reduced from 100 °C to 930 °C at a rate of 7.5 °C/min in 5% H2 in Ar flow.X-ray photoelectron spectrometer (XPS) experiment was performed in a THERMO VG SCIENTIFIC (MultiLab 2000, UK) operating Mg–Kα radiation at 14 kV and 20 mA. The binding energy was calibrated using C 1s at 284.6 eV.Brunauer-Emmett-Teller (BET) surface area and pore volume were measured using a surface area & pore size analyzer (Quantachrome autosorb-iQ) via adsorption of N2 at −195 °C. The samples were degassed in vacuum at 250 °C for 3 h.Thermogravimetric–differential thermal analysis (TG-DTA) was performed using a DTG-60H (Shimadzu, Japan) system below 1000 °C at a rate of 10 °C/min and 100 mL/min of air.Scanning electron microscopy (SEM) images were collected on a SU8020 (HITACHI, JPN) using 10–15 kV acceleration voltage and 9.8–10.0 mm working distance.Transmission electron microscopy (TEM) images, energy dispersive X-ray spectroscopy (EDS) analysis were obtained using TALOS F200X (Thermo Scientific™, USA) with a scanning transmission electron microscopy (STEM) unit and a high-angle annular dark-field (HAADF) detector at 200 kV. The samples were dispersed in ethanol and sonicated for 2 h, prior to placing a drop of liquid on a holey carbon coated copper grid and followed by evaporation for 5 min in a vacuum oven at 25 °C for the sample preparation.H2 chemisorption was performed on a chemisorption analyzer (ASPA 2020, Micromertics) 0.5 g of sample was reduced at 400 °C for 2 h under 5% H2. After vacuum at 400 °C, the sample was cooled to 25 °C and H2 chemisorption was performed till 67 kPa. H2 desorption profiles were also obtained with decreasing the pressure. Metal dispersion was calculated based on the stoichiometry of H/Ni = 1 [22,25].N2O chemisorption was performed on a pulse chemisorption analyzer (AutoChem 2920, Micromertics) 0.1 g of sample was reduced at 400 °C for 1 h under 5% H2/He. After purging under He, the catalysts were cooled to 90 °C and N2O was introduced using repeated 250 μl pulses (95% N2O/He). Metal dispersion for Ni and Cu were calculated based on the following reaction (Eqs (4) and (5)) [22]. (4) Ni + N2O →NiO + N2 (5) 2Cu + N2O →Cu2O + N2 The catalytic activity test for liquefied oil steam reforming was conducted in a fixed-bed reactor made of Inconel 625 material (outside diameter (OD) of 63.5 mm and length (L) of 600 mm), which was heated by a furnace. The catalyst (0.4–0.6 mm) was placed in a reactor and reduced at 550 °C for 5 h with 10% H2/Ar (v/v) flow. The liquefied oil feed was then fed into a reactor using a high-performance liquid chromatography pump (Young Lin Instrument). The steam reforming was conducted at 440–860 °C under atmospheric pressure with a S/C ratio of 10 and a liquid hourly space velocity (LHSV) of 0.2–2.3 h−1. The outlet gas was cooled using a trap and analyzed online using gas chromatography (HP-5890 Model). The concentrations of H2, CO, CH4, and CO2 were analyzed by gas chromatography (HP-5890, Agilent) with a thermal conductivity detector (TCD) equipped with a HayeSep DB column. After reaction, the catalysts were cooled to room temperature under Ar and passivated under 1% O2/He for 24 h. The obtained catalysts are denoted as “spent” catalysts.The carbon conversion of the liquefied oil, H2 yield, and the selectivity of H2, CO, CH4, and CO2 were calculated using Eqs. (6)–(8).Carbon conversion of a liquefied oil: (6) X C ( % ) = C m o l e s l i q u e f i e d o i l , f e e d − C m o l e s l i q u e f i e d o i l , o u t C m o l e s l i q u e f i e d o i l , f e e d H2 yield: (7) Y H 2 ( % ) = p r o d u c t i o n o f e x p e r i m e n t a l H 2 p r o d u c t i o n o f t h e o r i t i c a l l y H 2 × 100 Selectivity of product gases: (8) S i ( % ) = m o l e s o f g a s i o b t a i n e d m o l e s o f p r o d u c t g a s e s o b t a i n e d × 100 We performed hydrothermal liquefaction of brown algae (S. japonica) under the following conditions: 300 °C, 2 h, S. japonica/H2O ratio of 10 (w/w), and autogenous pressure [23,26]. Table 1 and S1 show the chemical composition of the liquefied oil, as determined through elemental analysis (C, H, and O), moisture analysis, and ICP analysis. The liquefied oil was composed of 5.2 wt% C, 11.7 wt% H, 80.9 wt% O, and 82.2 wt% H2O. Furthermore, it contained a high content of mineral components (mainly ∼2,900 ppm of Na and ∼11,000 ppm of K, and others (Ca, Mg, Si, Fe, and P) derived from S. japonica, as shown in Table S1. To eliminate the minerals, we performed a desalting process using a combination of H+ and OH− ion-exchange resin and Amberlyst-15, as shown in Fig. 1 a. Fig. 1b and Table S1 showed that most of the mineral components of the liquefied oil were removed to <10 ppm after the desalting process. Fig. 2 shows the main chemical components of the hydrothermally liquefied oil derived from S. japonica (detailed chemical components are shown in Table S2), as determined by GC-MS. As the main components of the liquefied oil were ketones (74.64%) and nitrogenous compounds (18.34%), the subsequent steam reforming of this liquefied oil should be mainly based on a ketone steam reforming mechanism.We prepared a series of NixCu1.5−xMg1.5Al1.0 catalysts by co-precipitation method with different atomic ratios of Ni/(Ni + Cu) by varying x in the range of 0–1. The physicochemical properties of the NixCu1.5−xMg1.5Al1.0 catalysts are summarized in Table 2 . All catalysts exhibited similar BET surface areas in the range of 100–134 m2/g. Figure S1 shows the Barret-Joyner-Halenda (BJH) pore size distribution of reduced NixCu1.5−xMg1.5Al1.0 catalysts. It indicates that pore radius of reduced catalysts ranges from 2 to 16 nm. The N2 adsorption/desorption hysteresis shows H3 type related to the slit-shaped pores from aggregates of plate-like particles (Sing. et al. [27]). SEM imaging (Figure S2) consistently showed the agglomerated particles with platelet-like morphology without apparent difference among catalysts. The Ni and Cu metal contents determined by XRF were essentially the same as the nominal metal contents used in the synthesis.The crystalline phases were analyzed by XRD (see Figure S3 and Fig. 3 ). XRD patterns of synthesized NixCu1.5-xMg1 . 5Al1.0 catalysts (Figure S3) shows diffraction peaks at 2θ values of 11.4°, 22.8°, 34.6°, and 38.9° indicating that the as-synthesized catalysts represent hydrotalcite phases (JCPDS 22–0700). In addition, the Ni0.5Cu1.0Mg1.5Al1.0 and Cu1.5Mg1.5Al1.0 catalysts contained the CuO phase (JCPDS #80–1916) and yielded diffraction peaks at 2θ values of 35.5° and 38.7°. After calcination at 600 °C (Fig. 3a and b), the hydrotalcite layer structures collapsed, and all the catalysts transformed into mixed metal oxides consisting mainly of the MgO periclase phase (JCPDS 75–1525) [9,28]. No distinctive NiO peaks were observed, indicating that the Ni2+ cations were well-incorporated in the MgO phase or undetectable highly dispersed particles. Ni0.5Cu1.0Mg1.5Al1.0 exhibited additional signals of the CuO phase at 2θ values of 35.5° and 38.7° [21,28]. As shown in Fig. 3b, the Cu1.5Mg1.5Al1.0 catalyst without Ni provided strong diffraction peaks of CuO and minor peaks of spinel phases (MgAl2O4 or CuAl2O4). The minor peaks are more likely due to CuAl2O4 because the Cu catalyst only showed a diffraction peak at 2θ = 32.8°. After reduction at 550 °C (Fig. 3c and d), the MgO periclase structure was still maintained. The reduced Ni1.5Mg1.5Al1.0 catalyst (without Cu) showed Ni(111) and Ni(200) diffraction peaks at 2θ = 45.5° and 51.9°. On the other hand, the reduced Cu1.5Mg1.5Al1.0 catalyst (without Ni) showed Cu diffraction peaks at 2θ = 43.2° and 49.5°. Fig. 3d shows that as the Cu/(Ni + Cu) atomic ratio increased from 0 to 1.5, the Ni(200) diffraction peaks gradually shifted to lower angles and finally to Cu(200) peaks at 49.5°. The gradual diffraction peak shift indicates that NiCu alloys were formed in the NixCu1.5−xMg1.5Al1.0 catalysts [9].The reducibility of the NixCu1.5−xMg1.5Al1.0 catalysts was studied by H2-TPR (Fig. 4 ). The Ni1.5Mg1.5Al1.0 catalyst exhibited a broad reduction feature at ∼750 °C, which is related to the reduction of Ni2+ species (NiO) in the mixed oxide [18,29]. For the Cu1.5Mg1.5Al1.0 catalyst, a sharp peak was observed at 170 °C, with another small peak at 645 °C. The former is attributed to the reduction of Cu2+ species in the mixed oxide [28,30,31] and the latter might be due to the reduction of the CuAl2O4 spinel phase [9].Upon increasing the Ni content in the NixCu1.5−xMg1.5Al1.0 catalyst, the reduction temperature of CuO increased from 170 °C (Cu1.5Mg1.5Al1.0) to ∼200 °C (Ni0.5Cu1.0Mg1.5Al1.0), suggesting that Ni addition leads to a stronger interaction between the Cu2+ species and parent hydrotalcite [28,30,31]. Upon increasing the Cu content instead of Ni, the high-temperature reduction peak of Ni2+ shifted to lower temperatures as compared with that of the Ni1.5Mg1.5Al1.0 catalyst. This originates from the hydrogen spillover from the reduced Cu metal during the H2-TPR process [28,31–33]. In summary, H2-TPR indicated synergetic interaction between Cu and Ni in the NixCu1.5−xMg1.5Al1.0 catalysts, suggesting that the reducibility of the catalyst changes with the Ni/Cu atomic ratio.The chemical states of the surface elements in the reduced catalysts were characterized by XPS. The XPS profiles for Cu 2p3/2 and Ni 2p3/2 are shown in Fig. 5 . Fig. 5a shows two Cu 2p3/2 peaks at 932.6 and 934.5 eV, indicating metallic Cu0 and Cu2+ oxidation states, respectively [22,34]. Fig. 5b shows three Ni 2p3/2 peaks. The peaks centered at 855.6 and 852.6 eV are assigned to Ni2+ and metallic Ni0, respectively [22,34], while the peak appearing at ∼ 6 eV at the higher binding energy is considered as the satellite peak. After the deconvolution process, the relative Ni2+/Ni and Cu2+/Cu ratios were quantified, and the results are shown in Table 2. The Cu1.5Mg1.5Al1.0 catalyst without Ni contained ∼51% of Cu0. As the Ni content was increased, the relative proportion of metallic Cu0 increased up to ∼82% in the Ni1.0Cu0.5Mg1.5Al1.0 catalyst. On the other hand, the Ni1.5Mg1.5Al1.0 catalyst without Cu contained ∼6.6% metallic Ni, and as the relative Cu content was increased, the relative proportion of metallic Ni0 increased to ∼17.9% in the Ni0.5Cu1.0Mg1.5Al1.0 catalyst. These results indicate that the addition of Ni or Cu induces the reduction of Cu or Ni species, respectively, on the surface, suggesting close interaction between the Ni and Cu atoms. This result is consistent with the H2-TPR and XRD results.Further, the metal dispersion was characterized by TEM and chemisorption. Figure S4 showed that Ni1.5Mg1.5Al1.0 catalyst had metal particles around <20 nm. Figure S5 shows TEM, STEM images and EDS mapping for Ni0.75Cu0.75Mg1.5Al1.0 catalyst. EDS mapping showed Ni and Cu are overlapped over the catalysts and metal particles (<20 nm) are observed. Figure S6 exhibited that Cu1.5Mg1.5Al1.0 catalysts had also 10–20 nm particles. However, because of local information from TEM images and particle overlap with support (mixed oxide), the accurate particle sizes are difficult to obtain. So, we additionally performed the chemisorption. The H2 volumetric chemisorption results in Table 2 reveal that the addition of Cu resulted in a decrease in the amount of H2 chemisorbed on the metallic particles from 1.8% (Ni1.5Mg1.5Al1.0) to 0.8–0.9% (Ni1.0Cu0.5Mg1.5Al1.0 or Ni0.75Cu0.75Mg1.5Al1.0) and 1.6% (Ni0.5Cu1.0Mg1.5Al1.0). Previous studies on NiCu catalysts have indicated that alloying Ni with Cu leads to weaker H2 adsorption, which hinders accurate characterization of NiCu alloy particles [35]. The decreased chemisorbed H2 amounts prove that the NixCu1.5−xMg1.5Al1.0 catalysts formed NiCu alloys. Instead, N2O probe molecules can characterize both the Ni and Cu surfaces by oxidizing the Ni and Cu species (Eqs. (4) and (5)) [22]. The metal dispersion data obtained from N2O chemisorption are summarized in Table 2. The average metal crystal sizes ranged from 19 nm to 33 nm. Interestingly, the metal crystal sizes are the smallest as 19 nm at 1 to 1 Ni:Cu atomic ratio, which were smaller than those of the Ni1.5Mg1.5Al1.0 (without Cu) or Cu1.5Mg1.5Al1.0 (without Ni) catalyst, suggesting that alloying Ni with Cu at a 1:1 atomic ratio results in a higher dispersion of metallic particles.The catalytic activities of the series of NixCu1.5–xMg1.5Al1.0 catalysts in the steam reforming of liquefied oil were evaluated comparatively under optimized conditions. First, the reaction conditions (LHSV and reaction temperature) were optimized to maximize the carbon conversion and H2 yield from the reaction of the liquefied oil on the Ni1.5Mg1.5Al1.0 catalyst based on a response surface methodology (RSM) [23,26,36]. Fig. 6 shows the contour plot of the carbon conversion and H2 yield with a change in the reaction temperature (440–860 °C) and LHSV values (0.2–2.3 h−1). The RSM analysis indicated a maximum H2 yield of 83% and carbon conversion of 98% at 750 °C and an LHSV of 1.0 h−1. Therefore, all catalysts were evaluated at the same condition.The catalytic activity and product distribution after the steam reforming of the liquefied oil on the Ni1.5Mg1.5Al1.0 catalyst with time on stream are presented in Fig. 7 . This catalyst was stable during 5 h of the reaction, and the carbon conversion and H2 selectivity were 92.8 and 80%, respectively. Further, the CO2 selectivity was ∼13%, and the CO, C2+, and CH4 selectivities were lower than 6, 2, and 2%, respectively. The catalytic activities and product selectivities obtained with the NixCu1.5−xMg1.5Al1.0 catalysts are shown in Fig. 8 and Figure S7. Fig. 8a showed that the Cu1.5Mg1.5Al1.0 catalyst initially underwent fast deactivation during 1 h and then stabilized. In contrast, the other Ni- and Ni/Cu-containing catalysts exhibited stable C conversion (>90%) and H2 selectivities (76–78%). Fig. 9 summarizes the catalytic performances of the NixCu1.5−xMg1.5Al1.0 catalysts after 5 h of the steam reforming of the liquefied oil. The catalyst containing only Cu showed the lowest C conversion of ∼67%, while the others showed >89% C conversion. Most of the gaseous product was composed of H2, with the highest H2 selectivity of 76–78%, followed by CO2 selectivity of 10–17%. Interestingly, the Ni0.75Cu0.75Mg1.5Al1.0 catalyst showed the lowest CO selectivity (∼2.8%), leading to the highest H2/CO ratio of ∼28. This result indicates that this catalyst is the most selective one for H2 production among the NixCu1.5−xMg1.5Al1.0 catalysts.We further compared the H2 production rates per exposed metallic sites (determined by N2O chemisorption) of all catalysts. Fig. 10 shows the variation of the H2-production rate per exposed metallic sites and metal dispersion with the Ni/(Ni + Cu) atomic ratio determined by XPS. Interestingly, a volcano-type plot of the H2-production rate vs. Ni/(Ni + Cu) atomic ratio was obtained, which was a similar trend between the metal dispersion and Ni/(Ni + Cu) atomic ratio. In particular, the highest H2 production rate was achieved at a Ni:Cu atomic ratio of 1:1, at which the metal dispersion in the catalyst was the maximum. It should be noted that this catalyst also showed the highest selectivity for H2 (i.e., the highest H2/CO ratio). This result suggests that the synergetic alloying of Ni and Cu at an optimum 1:1 atomic ratio leads to smaller metallic particles and more exposed metallic sites, resulting in a higher and selective H2 production rates compared to those of the catalyst containing Ni or Cu only [9]. In general, Ni exhibits good activity for steam reforming reaction due to its high C–C bond scission ability [11,12]. Furthermore, Cu prevented the sintering of Ni by forming NiCu alloy, and promoted H2 production through its high activity in the water gas shift reaction (Eq. (2)) [20–22]. Especially, designing smaller NiCu alloy particles based on the optimized Ni/Cu atomic ratio is critical for H2 production by the steam reforming of liquefied oil derived from S. japonica [9].After the reaction, the spent catalysts were analyzed by XRD (Fig. 11 ). These results suggested that the Ni/Cu ratio had considerable effect on the crystallite aggregation in hydrotalcite precursors catalysts [24]. Notably, the Ni0.75Cu0.75Mg1.5Al1.0 catalyst, which showed the best catalytic performance, had the smallest metal particle size compared to the other catalysts, suggesting that small metallic particles are important for H2 production. Also, TEM/STEM and EDS analysis were performed on selected spent catalysts, as shown in Figure S8–S10. Ni1.5Mg1.5Al1.0 catalyst (Figure S8) showed the agglomerated Ni particles (>100 nm). Figure S9 exhibited that Ni0.75Cu0.75Mg1.5Al1.0 catalyst had 40–100 nm particles, smaller than those of Ni1.5Mg1.5Al1.0 catalyst although some region also showed sintered particles (>100 nm). Detected metal particles are confirmed as NiCu alloy from EDS analysis (Figure S10). Considering both XRD and TEM results, we conclude that Ni0.75Cu0.75Mg1.5Al1.0 catalyst had smaller metal particles than Ni1.5Mg1.5Al1.0 catalyst. Cu1.5Mg1.5Al1.0 catalyst (Figure S11) also showed very large particles (>100 nm) and also agglomerated regions with 20–30 nm particles. Table 2 showed that BET surface area of NixCu1.5-xMg1.5Al1.0 (x = 0.5–1.5) and Cu1.5Mg1.5Al1.0 catalysts decreased to 37–53 m2/g and 5 m2/g after the reaction, respectively. Figure S12 shows the BJH pore size distribution of spent NixCu1.5−xMg1.5Al1.0 catalysts, indicating the pore radius of studied catalysts ranges from 6 to 16 nm, which increased compared to reduced catalysts (2–16 nm). Particle agglomeration and carbon species accumulation during the reaction decreased the BET surface area and also affect the pore size distribution of spent catalysts. SEM analysis (Figure S13) showed no notable morphology change. Coke analysis of the spent catalysts (Figure S14) revealed that only the Cu1.5Mg1.5Al1.0 catalyst showed the weight loss at 300–400 °C, indicating that amorphous coke was deposited on the catalyst. This might be the cause of the initial deactivation of the Cu1.5Mg1.5Al1.0 catalyst (Fig. 8) and the drastic decrease of BET surface area on Cu1.5Mg1.5Al1.0 catalyst. Other NixCu1.5−xMg1.5Al1.0 catalysts (x = 0.5–1.5) exhibited weight increases owing to the oxidation of metallic particles [37,38]; therefore, notable coke deposition (weight loss) could not be detected. These results suggest that the NiCu hydrotalcite-derived catalysts are promising catalysts for the steam reforming of liquefied bio-oil derived from S. japonica (macro algae).Liquefied oil from marine biomass can be a good resource for hydrogen by steam reforming it on NiCu hydrotalcite-derived mixed oxide catalysts (NixCu1.5−xMg1.5Al1.0). S. japonica is liquefied into bio-oil by hydrothermal liquefaction at 300 °C. GC-MS analysis showed that Bio-oil from Saccharina japonica mainly contain ketone and N-containing compounds. After the desalting process combined with H+/OH− ion-exchange resin and Amberlyst-15, the remaining salts are mostly removed to <10 ppm. XRD, H2-TPR, TEM, XPS, chemisorption exhibited the synergetic interaction between Ni and Cu. XRD showed the Ni diffraction peak shifts with varying Ni:Cu atomic ratio, indicating the NiCu alloy formation. With increasing Cu content, the reducibility of Ni was improved, evidenced by H2-TPR and XPS. EDS analysis also showed that Ni and Cu mapping are overlapped. N2O chemisorption showed that particle sizes also vary where 1:1 atomic ratio of Ni and Cu had the smallest particle sizes (∼19 nm).The steam reforming of bio-oil from S.japonica was performed over NixCu1.5-xMg1.5Al1.0 catalysts. NixCu1.5-xMg1.5Al1.0 catalysts, except Cu1.5Mg1.5Al1.0 catalyst, were stable with >89% of carbon conversion and H2 selectivity of 76–78% during 5 h. In particular, at an optimized Ni:Cu atomic ratio of 1:1, the synergetic interaction of Ni and Cu led to the smallest NiCu alloy particles and maximized H2 production rates along with the highest selectivity to H2. Thus, designing smaller NiCu alloy particles over NiCu hydrotalcite-derived catalysts is critical for H2 production via the steam reforming of liquefied oil from S. japonica. Seong Chan Lee: Investigation, Data curation, Writing – original draft, Writing – review & editing. Jae Hyung Choi: Validation, Formal analysis, Writing – original draft, Writing – review & editing. Chul Woo Lee: Resources, Methodology, Investigation. Seung Han Woo: Resources, Methodology, Investigation. Jaekyoung Lee: Project administration, Supervision, Writing – original draft, Writing – review & editing. Hee Chul Woo: Conceptualization, Supervision, Project administration, Funding acquisition, Writing – review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No.2017R1E1A1A01074445 and 2021R1A2C2094256).The following is/are the supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.renene.2022.03.161.

H2 is highlighted as a sustainable energy resource, and mainly produced by steam reforming of fossil fuels, which emit greenhouse gases. Marine biomass can be an alternative because of high productivity and carbon neutrality compared to terrestrial biomass. In this work, we studied bio-oil from Saccharina japonica (macroalgae) as a renewable H2 resource by steam reforming on NiCu hydrotalcite-derived catalysts (NixCu1.5-xMg1.5Al1.0). After the hydrothermal liquefaction of S. japonica, minerals were removed by the desalting process. GC-MS showed bio-oil mainly consists of ketone and N-containing compounds. Increasing Cu content improved the reducibility of Ni, evidenced by H2-TPR and XPS, suggesting the synergetic interaction between Ni and Cu. Chemisorption showed the catalyst had the smallest particle sizes (∼19 nm) at 1 to 1 Ni:Cu atomic ratio. As for steam reforming of bio-oil, NixCu1.5-xMg1.5Al1.0 catalysts, except Cu1.5Mg1.5Al1.0, were stable with >89% of carbon conversion and H2 selectivity of 76–78% during 5 h. Especially, at 1:1 Ni:Cu atomic ratio, the catalyst maximized H2 production rates with the highest H2/CO ratio of 28. It suggests that designing small NiCu particles is critical for H2 production. In summary, NixCu1.5-xMg1.5Al1.0 catalysts are promising for H2 production by the steam reforming of the bio-oil from macro algae.

In 2018, the global carbon dioxide emissions increased by 2.7% [1]. The challenges of meeting the growing need for electricity and transport fuel of the world must be dealt with a substantial decrease in CO2 emissions [2]. Hydrogen, which can replace fossil fuels such as coal, diesel, gasoline, and natural gas, may uniquely decarbonize the electricity and transportation sectors when produced from greenhouse gases [3]. Hydrogen has the potential to be a major source of low-carbon energy, with the possibility of reducing CO2 emissions to nearly zero. Hydrogen demand was ∼8 EJ in 2015, and is anticipated to increase tenfold by 2050 [4]. This fact motivates the search for novel technologies for inexpensive, CO2-free H2 production on an industrial scale. Developing large-scale, economically competitive H2 production processes is essential for producing low-carbon fuels and fertilizers [5,6]. The use of methane would take advantage of the existing natural gas infrastructure, reducing the conversion costs to a hydrogen-based energy system. Dry reforming of methane (DRM), as shown in equation (1) below, is one of the viable routes to produce hydrogen [7–9]. Catalytic DRM reaction utilizes greenhouse gases, comprising methane and carbon dioxide, to form syngas (CO and H2) [10,11]. DRM has also been receiving extensive consideration in recent years [12–14] due to the formation of syngas (H2/CO) in a mole ratio close to unity, which is required for Fischer-Tropsch synthesis (FTS). Reverse water gas shift (RWGS), a major side reaction (Equation (2)) in CO2 reforming of CH4, gives rise to H2/CO mole ratio less than unity [15–17]. (1) CH4 + CO2 → 2CO + 2H2 (2) CO2 + H2 → CO + H2O Nickel-based catalysts are less expensive than noble metal-based catalysts for DRM but are prone to sintering at high reaction temperatures. The importance of the catalyst support is evident because it provides a surface area for the dispersion of Ni catalyst, and its surface chemistry greatly facilitates CO2 activation [18]. Alumina (Al2O3) is cheap, abundant catalyst support for DRM due to its high surface area and thermal stability [19]. γ-Al2O3 support was found to be more suitable than β-Al2O3 support for Ni-based catalyst in DRM reaction because γ-Al2O3 can hold large numbers of Cα species (completely dehydrogenated carbides carbon) over its surface, which resulted in a higher CO2 conversion [20]. Kim et al. [21] used the solvothermal method to prepare the Ni/Al2O3 nanosheet catalyst, which exhibited significantly more stable activity than Ni/Al2O3 with an arbitrary configuration. In addition, considerable efforts have been made to boost the catalytic activity and stability by tuning the textural property of the support using promoters. Fig. 1 indicates the type of promoters inducing Al2O3-supported Ni catalyst towards DRM.Among s-block elements, Alipour et al. [22] utilized MgO-, CaO- and BaO-modified, Al2O3-supported Ni-based catalysts. They found that the moderate addition of MgO as a modifier improved the catalytic activity over 5.0 wt% Ni/Al2O3 catalyst due to the formation of MgAl2O4 phase. In a different study, the catalyst of 2.5 wt% Ni supported on Mg-Al mixed oxide showed more than 65% CH4 conversion and 75% CO2 conversion for 7 h on stream [23]. Khoja et al. prepared 10 wt% Ni/La2O3-MgAl2O4 (1:4) catalyst and was tested in catalytic dielectric barrier discharge reactor for DRM and found more than 79% CH4 conversion and 84% CO2 conversion with H2/CO mole ratio of ∼1.0 [24]. On the other hand, the 0.75 wt% Sr promoter on Ni/Al2O3 catalyst facilitated strong metal-support interaction and boosted the basicity, which induced the dissociation of CO2 over the catalyst and, in turn, decreased the coke deposition. It demonstrated minimum deactivation and more than 75% CH4 and CO2 conversion with H2/CO = 0. 95 [25]. Among p-block elements, the role of 5.6 wt% boron as a modifier was found to suppress 86% of carbon deposition without affecting the catalytic activity due to a more uniform distribution of Ni catalyst over the support [26]. Moreover, the incorporation of boron nitride into a nickel-based catalyst not only avoided metal particle sintering but also enhanced coke resistance [27,28]. Wei et al. coated monolithic SiC foam with “Ni embedded in mesoporous Al2O3 layer”, which caused a high Ni dispersion over a larger specific surface area. It showed more than 30% CH4 conversion and 40% CO2 conversion [29]. In contrast, 3.0 wt% Si modifier exhibited excellent coke resistance, and it also induced strong metal-support interaction or the formation of NiAl2O4 mixed oxide. It demonstrated more than 64% CH4 conversion, 70% CO2 conversion, and an H2/CO mole ratio of ∼0.90 over 7 h of reaction [30]. The lack of negative peak in the TPO spectra of used 1.0 wt% Ga-promoted, loaded on Ni/Al2O3 catalyst (concerning other transition metal promotors of Cu, Gd, and Zn) indicated the retention of Ni in its metallic state. It showed more than 74% CH4 conversion and 84% CO2 conversion over 7 h on stream [31].In f-block elements, due to the development of strong metal-support interaction, 1.0 wt% of Gd promoter resulted in >83% CH4 conversion, 88% CO2 conversion, and an H2/CO mole ratio of ∼1.0 over 7 h on stream [31]. Lanthana added basicity to the catalyst system, inducing more CO2 adsorption, increased metal-support interaction, and neutralized the acid sites of alumina support (responsible for excessive coke deposit). The formation of La2O2CO3 can prevent Ni from sintering and facilitate NiAl2O4 formation. La2O2CO3-modified, Ni/Al2O3 exhibited more than 60% CH4 conversion and CO2 conversion with H2/CO > 0.8 [32]. The presence of Ceria and praseodymium oxide presence offered additional redox properties. Mobile lattice oxygen is readily available during the redox cycle for carbon deposit oxidation. Thus, 5.0 wt% ceria addition resulted in >75% CH4 conversion and CO2 conversion with H2/CO ∼0.8 at 800 °C [33]. Praseodymium oxide promotional addition induces the transport of electrons through the oxygen vacancies of the redox pair of Py4+/Py3+. The catalyst with 3.0 wt% Py-10 wt% Ni/delaminated clay, showed no carbon formation, 38% CH4 conversion, 43% CO2 conversion, and H2/CO = 0.7 [34]. The addition of 2.0 wt% Yb controlled the Ni particle size, resulting in the narrowest particle size distribution and the highest reducibility over Al2O3 support. It demonstrated ∼80% CH4 conversion, 87% CO2 conversion, and H2/CO > 0.9 [35].In d-block elements, the role of Cu and Ti were examined, but each induced the formation of free NiO particles over the surface (or weak metal-support interaction). Therefore, the addition of Cu and Ti was not beneficial for DRM [30,36]. The promotion of Ni/Al2O3 catalyst with Mn resulted in coke resistance, but it suppressed the catalytic activity due to the partial blocking of the catalytic active sites “metallic Ni” by manganese oxide [37]. However, adding 2 wt ratio (%) potassium further stabilized its catalytic activity. The promotion with Mo for the Ni/Al2O3 catalytic system was found to be inferior due to the formation of the MoNi4 phase and the weak interaction of Ni with support [38], while Mo promoted/modified ZSM-5 facilitated efficient coke removal for nickel-based catalysts [39]. The addition of 10 wt% ZrO2 promoter was found to enhance the dissociation of CO2 through the formation of oxygen intermediates (near the ZrO2 and nickel interface), which oxidized coke. Thus, 10 wt% ZrO2-promoted, 15 wt%Ni/Al2O3 catalyst showed more than 70% CH4 conversion and 60% CO2 conversion [40]. The 3.0 wt% Co-promotional addition into 10 wt%Ni/Al2O3 catalyst was remarkable. Co-controlled Ni particle size, causing coke depression, high CO2 conversion (>90%), high CH4 conversion (>90%), and high H2/CO ratio (∼0.9) at 850 °C [35].Both the 3.0 wt% W and 3.0 wt% Si as promoters into Ni/Al2O3 catalyst showed similar catalytic performance in DRM as both had attained >64% CH4 conversion, >70% CO2 conversion, and >0.90 mol H2/CO ratio over 7 h of reaction [30]. W-modified Ni/Al2O3 catalysts were examined in DRM reaction [41], where 10 wt% Ni and various tungsten loadings were tested. It was found that 11.9 wt% W reduced the carbon deposition by 76% compared to W-free catalyst. Researchers have enhanced the performance of nickel-based catalysts by adding other promoters of d-block metal oxide. Nevertheless, WO3 was found to be more reactive than other metal oxides like Fe3O4, ZnO, SnO2, and V2O5 because of its enhanced reducibility [42]. WO3 had superior properties over B2O3, TiO2, ZrO2, and MoO3 as promoters because it enhanced the interaction between NiO and γ-Al2O3 support, and hence, improved the dispersion and stability of nickel particles in the catalytic system during DRM [6]. The CH4 decomposition over WO3 (forming tungsten carbide, WC), the gasification of carbon deposits by CO2, and the redox property of tungsten oxide (WO3 → WC → WO3) have sparked interest in the use of tungsten as a promoter [42]. The presence of WC during the reaction also increased the catalyst system's thermal stability, which was a must condition for high-temperature DRM reaction [43]. Al-Fatesh and his co-workers [6] employed ZrO2-supported Fe catalysts for the catalytic decomposition of methane and investigated the influence of La2O3 and WO3 dopants on the catalytic activity and stability. WO3 strongly influenced the methane conversion, hydrogen yield, and stability at 800 °C, 4000 mL/(h.gcat.) space velocity because of the higher dispersion, stabilization, and stronger interaction of iron nanoparticles on the surface of ZrO2 modified with WO3. Therefore, tungsten trioxide is worth studying owing to its high stability, coke resistance, and reactivity [44]. Overall, low-cost tungsten oxide may have redox properties, and its presence over a supported Ni system may induce Ni dispersion, enhance stronger metal-support interaction and promote coke oxidation which must favour the dry reforming of methane. In this context, this study aimed to optimize the amount of tungsten trioxide promoter for the best catalytic performance of nickel nanoparticles supported on mesoporous gamma-alumina (Ni/γ-Al2O3). The effect of 1.0–9.0 wt% of tungsten trioxide loadings on the textural, morphological, and catalytic properties was investigated.Ammonium tungstate [(NH4)10H2(W2O7)6; 3060.45 g/mol; 99.99% trace metals basis; Aldrich], mesoporous γ-alumina (meso-Al2O3, 1/8" pellets, Alfa Aesar), and nickel nitrate hexahydrate [Ni (NO3)2.6H2O, 98%, Alfa Aesar] were purchased and were used as received. Ultrapure water was obtained via a Milli-Q water purification system (Millipore).A two-step synthesis procedure based on dry impregnation was followed to prepare all catalysts. The first step was the synthesis of mesoporous γ-alumina support promoted with tungsten trioxide (x%WO3/γ-Al2O3), where x = 1.0–9.0 wt% in 2.0 wt% increments. The second step was to load the active catalyst as nickel oxide with 5.0 wt% on the various supports. These two steps are described in detail below.The required amount of ammonium tungstate to produce the required weight percent loading of tungsten trioxide and the required amount of mesoporous γ-alumina were mixed and ground together to give a solid white mixture. Ultrapure water was added dropwise to get a thick paste, which was mechanically stirred until dry. The addition of water and mechanical stirring was repeated three times to ensure homogeneous distribution of ammonium tungstate within the matrix of mesoporous γ-alumina. The solid mixture was finally calcined under static air, at 600 °C, over 3 h to give white supports promoted with various loadings of tungsten trioxide (1.0, 3.0, 5.0, 7.0, or 9.0 wt%).The required amount of nickel nitrate hexahydrate was mixed and was ground with the required amount of the desired support of xWO3-doped γ-Al2O3 (x = 1.0, 3.0, 5.0, 7.0, or 9.0 wt%) to produce a solid green mixture. This mixture was then converted to a paste by adding drops of ultrapure water. The paste was mechanically stirred to ensure complete drying and the formation of a green solid. The addition of water and stirring was repeated three times, and then the solid mixtures were calcined for 3 h at 600 °C to produce brown solids.Powder X-ray diffraction (XRD) patterns of the prepared catalysts were recorded on a Miniflex Rigaku diffractometer, worked at 40 kV and 40 mA, and fitted with Cu Kα X-ray radiation.The isotherms of N2 physisorption were measured by using a Micromeritics Tristar II 3020 surface area and porosity analyzer at −196 °C after outgassing the samples at 200 °C for 3 h to desorb accumulated gases or vapors on the surface and into the pores. The Barrett–Joyner–Halenda (BJH) model was used to investigate the distributions of pore size of the samples.The H2-TPR, CO2-TPD, and NH3-TPD analyses of the synthesized and the spent catalysts were performed on a Micromeritics Auto Chem II 2920. The tests were done over a temperature range of 50–800 °C and 2.40 L/h flow of 10% H2/Ar mixture for the H2-TPR analysis, 10% CO2/He mixture for CO2-TPD basicity measurement, and 10% NH3/He mixture for NH3-TPD acidity assessment. During the H2-TPR analysis of the catalyst, 0.070 g of the catalyst precursors were first heated to 150 °C and held at that temperature for 60 min in the presence of Ar at the rate of 1.8 L/h and then cooled to room temperature. Next, the sample temperature was raised to 900 °C at 10 K/min under 10% H2/Ar mixture in an automatic furnace at 1 atm. The amount of consumed H2 was determined by a thermal conductivity detector (TCD). For the NH3-TPD acidity assessment, ∼150 mg of each tested catalyst was placed in a reaction tube. After pretreating at 600 °C for 2 h under a He flow of 50 mL/min, the sample was cooled to 170 °C and dosed for 30 min with a 10% NH3 in He (balance). After dosing, the samples were cooled to 50 °C, followed by heating to 600 °C at 5.0 °C/min, under He flow of 50 mL/min. The NH3 concentration in the output was recorded via a TCD.The mass of carbon deposits on the surface of spent catalysts was determined using a thermo-gravimetric unit (Shimadzu-TGA). The deposited carbon was removed by heating the samples under air up to 1000 °C at a 10 °C/min heating rate and recording the weight loss.Laser Raman spectra of the spent catalysts were measured, in the spectral range of 1400–1600 cm−1, on a JASCO NMR-4500 spectrometer. The wavelength of the excitation beam was fixed to 532 nm with the use of the objective lens of 100X magnification. The laser power was tuned to 1.6 mW for 10 s of exposure time at three accumulations to prevent sample damage by laser irradiation. Spectra Manager Ver.2 software (JASCO, Japan) was used to process the spectra.DRM experiments were performed at 700 °C and ambient pressure. Stainless-steel reactor (i.d. = 0.0091 m; length = 0.3 m) was used. An amount of 0.1 g of the catalyst was used for catalytic testing. The temperature was monitored using a sheathed, stainless-steel K-type thermocouple (placed at the centre of the catalyst bed). Before the catalytic examination, the catalyst was activated at 800 °C, under H2 flow, for 1 h. Methane, carbon dioxide, and nitrogen gases were mixed in a 3:3:1 vol ratio during the experiments. This gas mixture was used as a reactant feed with a space velocity of 42 L/h/gcat. The effluent gas was analysed by online GC-2014 SHIMADZU, equipped with a thermal conductivity detector, molecular sieve 5A column, and porapak Q column. Methane and CO2 conversions and deactivation factor were calculated as shown below: (3) C H 4 Conversion ( % ) = moles of ( C H 4 ) i n – moles of ( C H 4 ) o u t moles of ( C H 4 ) i n × 100 × 100 (4) C O 2 Conversion ( % ) = moles of ( C O 2 ) i n – moles of ( C O 2 ) o u t moles of ( C O 2 ) i n × 100 (5) Deactivation Factor = ( Initial C H 4 Conversion − FinaI C H 4 Conversion ) Initial C H 4 Conversion Nitrogen physisorption was used to determine the catalysts' surface areas. Table S1 presents the BET-specific surface area (SBET), pore volume (Pv), and average pore diameter (Pd) for each promoted and un-promoted catalyst. The highest surface area was found for the un-promoted catalyst (5%Ni/γ-Al2O3) with a surface area of 159.4 m2/g. Upon incorporating 1.0 wt% tungsten trioxide, SBET, Pv, and Pd of the catalyst were decreased due to the deposition of WO3 in the pore with values of 149.7 m2/g, 0.46 cm3/g and 11.4 nm, respectively. On further addition of tungsten trioxide, there were no discernible trends in surface area, pore volume and pore diameter results; however, the surface areas of the promoted systems were always lower than those of the un-promoted catalytic system [45]. Fig. 2 demonstrates the N2 adsorption-desorption isotherms, which, for all catalysts, fell within the mesoporous range of type IV isotherms, according to the IUPAC classification [46]. H1 hysteresis loop on type IV isotherm was specified by the sharp inflection in the 0.6–0.75 relative pressure region. It indicated the presence of uniform cylindrical mesopores.X-ray diffraction (XRD) analysis was carried out to investigate the prepared catalysts' crystallinity, as shown in Fig. 3 . The diffraction peaks with Miller indices of (311), (400), and (440) were identified at Bragg angles (2θ): 37.10, 45.72, and 66.77°, respectively. These diffraction peaks were attributed to the cubic γ-aluminium oxide phase (JCPDS card Nos.: 01-029-0063). No diffraction peaks were observed for nickel oxide, tungsten trioxide, and nickel aluminate (NiAl2O4), implying either their high dispersion on γ-Al2O3 support or their diffraction peaks were overlapped with those of the support. The absence of tungsten trioxide diffraction peak even at 9 wt% loadings indicated that this amount was still below monolayer coverage, interacting with the surface, and thus was not detectable by XRD [46]. Variation of γ-Al2O3 crystallite size upon the addition of tungsten trioxide was quite informative. The un-promoted catalyst had a crystallite size of 26.8 nm (at 2θ = 66.7°) for the γ-Al2O3, whereas just after the addition of 1.0 wt% WO3, the size of γ-Al2O3 crystallite drastically decreased to 10.1 nm (Table S2).The reduction temperature profiles (TPR) of the fresh catalysts are shown in Fig. 4 . H2-TPR showed no reduction peaks in the WO3-γ-Al2O3 catalyst system because no reducible metal oxides were present. Interestingly in the literature, WO3-loaded on silicate system had shown H2-TPR reduction peaks exclusively in the high-temperature range of 600–1000 °C due to reduction of W+6 state [47]. However, WO3-γ-Al2O3 catalyst system had no reduction peaks, indicating that WO3-γ-Al2O3 matrix was stable in H2 stream up to a wide range of temperatures. While upon loading 5.0 wt% NiO, reduction peaks appeared in various temperature regions, and all of them were detected below 1000 °C. Due to the strength of the interaction between the active NiO and mesoporous-γ-Al2O3 support, different reduction temperature regions were observed. The reduction peaks in the temperature range of 500–700 °C were attributed to the moderate interaction of NiO with the support. At high temperatures, nickel ions were diffused into the γ-Al2O3 support, coordinated tetrahedrally and octahedrally and formed NiAl2O4 [48]. Thus, NiO was in strong interaction with γ-Al2O3 support, and so the reduction peak in the temperature range of 700–900 °C was due to the reduction of NiAl2O4 species [48]. At temperatures above 900 °C, some aluminum ions in NiAl2O4 may be replaced and form new NiO-WO3 species. Incorporating W with a higher oxidation state than Al in the matrix produced strong bonding over oxygen. Thus, NiO-WO3 species would be harder to reduce than NiAl2O4 species. Thus, the peak in the temperature range of 900–1000 °C could be attributed to the reduction peak of NiWOAl species [48]. The sudden decrease in crystallite size of Al2O3 after adding WO3 was also noticed in XRD, confirming the substitution of Al ions with W ions. The negative peaks in the temperature range of 100–300 °C may be due to the hydrogen spillover in the mesopores of γ-Al2O3 support [30]. Table 1 displays the H2 consumption quantities of the various promoted and un-promoted tungsten catalysts. The unpromoted catalyst had a reducible peak of NiAl2O4 species majorly. Upon tungsten trioxide loading, reducible species of moderately interacting NiO and strongly interacting NiWOAl were grown. 5Ni + xWO3/γ-Al2O3 (x = 1–5) catalyst exhibited all types of NiO-interacting species, as discussed above, with support. With increasing tungsten trioxide loading, the total amount of reducible NiO-interacting species was increased over the catalyst surface (Table 1). Above 5 wt% WO3 loading, 5Ni+7WO3/γ-Al2O3 and 5Ni+9WO3/γ-Al2O3 catalysts had no reducible NiWOAl species.The CO2-temperature-programmed desorption (CO2 -TPD) of the fresh catalysts is shown in Fig. 5 A. It depicts the CO2 desorption peaks within the temperature range of 50–650 °C. The desorption peaks below 100 °C were attributed to weak basic sites/surface hydroxyl, while the peaks below 200 °C were attributed to medium-strength basic sites/carbonates, and the peaks from 200 to 400 °C were for strong basic sites/surface O2− species [49,50]. The 5.0 wt% WO3 catalyst sample exhibited a broad range of CO2 adsorption over a wide distribution of basic sites over the catalyst surface. However, it had a moderate amount of basic sites/surface anion, whereas 3.0 wt% WO3 had the maximum number of strong basic sites/surface anions. The NH3-TPD profiles (Fig. 5B) showed that 5.0 wt% WO3 loading had the maximum number of acidic sites, implying that the surface of this catalyst was the most enriched with acidity. The acidic sites were claimed for carbon accumulation in DRM [51]. Overall, 5.0 wt% WO3 catalyst had a moderate number of basic sites and the maximum number of acidic sites, whereas 3.0 wt% WO3 had a moderate number of acidic sites and the maximum number of basic sites. Previously, it was reported that the surface WOx species on Al2O3 was more acidic due to the high electronegativity of the Al and the formation of more acidic bridging W–O–Al [52]. Therefore, upon the addition of WO3 over alumina support, a rise in acidity can be expected. However, after an optimum loading of WO3 (>5 wt%), the drastic fall in the acidity of the catalyst was noticeable in 5Ni+7%WO3/γ-Al2O3 and 5Ni+9%WO3/γ-Al2O3 catalysts. The drastic fall in acidity may be claimed to the coverage of the acid site of alumina by surmounting WO3 [53]. However, the same observation was found over the zirconia-supported WO3 catalyst. Zirconia support is not acidic, but upon more than 5 wt% WO3 loading, the acidity of the catalyst was fallen [54]. It indicates that above than 5 wt% WO3 loading, whether support is alumina or zirconia, the acidity of catalyst was decreased. It pointed out that bulk WO3 was formed after optimum loading, leading to a decrease in ammonia adsorption and surface acidity [47]. Fig. 6 shows the SEM images for two fresh catalysts of 5Ni + xWO3/γ-Al2O3, where x is either 3.0 wt% (Fig. 6A) or 9.0 wt% (Fig. 6B). Chunky with irregular-shaped particles were observed for both samples. The morphology was not affected by varying the weight percent loading of WO3. The EDX surface elemental analysis of 5Ni+3WO3/γ-Al2O3 is shown in Figure S1. Surface oxygen could help in the gasification of carbon deposition and played a role in surface basicity [55]. Aluminium is the second element on the surface in terms of abundance (∼42 wt %). The loaded nickel appeared on the surface with ∼4 wt%, which could be linked to the observed catalytic performance because nickel is the active catalyst and interacts with the support.Experimental CO2 and CH4 conversion results in Fig. 7 A and Fig. 7B, respectively, showed that the catalytic performance of all catalysts was fairly stable within a reaction time of 7.5 h. The operation of the RWGS reaction during the DRM was manifested by the higher values of fractional conversion of CO2 than those of CH4 [56]. The conversions of CH4 and CO2 were 72–73% and 78–79%, respectively, over 5Ni/γ-Al2O3 within 7.5 h at 700 °C. Incorporation of tungsten trioxide and increasing its loading to 5.0 wt% WO3, the CH4 conversion, and CO2 conversion increased progressively. Tungsten trioxide was also known for its ability to decompose CH4. Thus, the increase in CH4 conversion could be primarily correlated with the increase of tungsten trioxide in the catalyst [42]. The conversions of CH4 and CO2 were 79% and 83%, respectively, over 5Ni+5WO3/γ-Al2O3 within 7.5 h at 700 °C. The incorporation of tungsten trioxide into the catalyst had no effect on CO2 conversion when compared to CH4 conversion. Table S3 displays that the H2/CO mole ratio was ∼1.0 for all catalysts during the 7.5 h of time-on-stream. This molar ratio is the proper one for higher conversions, better stability, and reduction in CO2 participation in RWGS reaction [57]. Tungsten-promoted catalysts demonstrated substantial stability in syngas production, with the minor carbonaceous compound formation on the surface of catalysts likely attributable to the presence of tungsten [58]. It was reported that tungsten promoter covered particularly the deformed sites at the catalyst surface, which improved the rate of carbon gasification and thereby induced stability. Furthermore, the existence of Ni-stabilized reducible species (i.e., NiAl2O4 and NiWOAl) in the presence of WO3 may substantially affect deactivation. At 5.0 wt% WO3 loading, the deactivation factor (0.88) was the minimum. The 5.0 wt % WO3 loading was the optimum loading in terms of CH4 conversion and minimum deactivation. Further loading above 5.0 wt% loading resulted in a reduction in both CH4 conversion and deactivation factor.Thermogravimetric (TGA) analysis for the spent catalysts (Fig. 8 ) was performed in the temperature range of 100–1000 °C after running DRM for 7.5 h at 700 °C. The catalyst with 9.0 wt% WO3 had the highest carbon deposition (88 mg/g), whereas the amount of carbon deposit formed on 5.0 wt% tungsten trioxide was the least (31 mg/g). With increasing tungsten trioxide loading from 1 to 5 wt%, the amount of carbon deposition decreased over the spent catalysts because of the reducible characteristic of tungsten [19]. In addition, the improved stability and higher resistance to carbon deposition of the tungsten-promoted catalysts could be ascribed to the cover of active sites with the tungsten promoter, particularly deformed sites, which lowered the amount of formed carbon. Sayed and his co-workers have claimed the coke resistance property of the tungsten-promoted catalyst [42]. The alternative explanation can be derived from the acidic-basic profile of the catalyst. Carbon decomposition depends on CH4 dissociation and, is then delayed in the oxidation of CHx species by CO2. Basicity enhances CO2 adsorption, whereas acidity encounters CH4 dissociation [58]. Herein, up to 5.0 wt. W%, there was fine-tuning between acidic and basic sites for proper carbon removal, but after this optimal loading, fine-tuning was lost, resulting in an increase in carbon deposition at 7.0 wt% Ni and the highest in 9.0 wt% Ni.The laser Raman spectra (Figure S2) for all spent catalysts in the spectral region of 1400−1600 cm−1 showed two intense peaks at Raman shift of 1475 ± 5 cm−1 and 1530 ± 10 cm−1, which corresponded to the D and G bands, respectively [59,60]. The D band is related to amorphous/disordered carbon deposits, while the G band is characterized by graphitic/ordered carbon deposits [61,62]. The ratio of the intensity of disordered carbon and ordered carbon (ID/IG) was 1.15 over the unpromoted catalyst and 1.08–1.01 over tungsten-promoted catalysts, implying that the unpromoted catalyst had relatively more amount of disordered carbon than ordered carbon. Among tungsten-promoted catalysts, ID/IG value decreased nominally from 1.08 to 1.01 on increasing loading of tungsten trioxide promoter from 1.0 wt% to 9.0 wt%. It indicates that the degree of graphitization was nominally increased with increasing tungsten trioxide loading.Previous literature on WO3-promoted Ni/Al2O3 catalyst showed better coke-resistant but inferior catalytic activity than the unpromoted catalyst due to partial coverage of catalytic active sites by WO3 (6.3–26.3 wt%), less exposure of Ni species, and the formation of Ni17W3 alloy, which was less active for methane decomposition [41,42]. Our two-step synthesis methodology used less amount of WO3 (5 wt%) and showed superior catalytic activity (∼79% CH4 conversion and ∼83% CO2 conversion during up to 7.5 h time-on-stream) than the earlier reported co-impregnated catalyst. The synthetic methodology may be one of the reasons for the good catalytic performance. We first synthesized mesoporous alumina support, which was then impregnated with tungsten trioxide and calcined. The second step was the dry impregnation for loading nickel precursor onto “mesoporous alumina support promoted with tungsten trioxide” followed by calcination. WO3-Al2O3 support in our catalyst system was not reducible under H2 steam. Consequently, WO3 interacted well with Al2O3 in this synthetic approach and was stable up to 1000 °C under H2 steam. The Ni/Al2O3 catalyst system was quite active for CH4 conversion (∼72%). Certainly, the potential catalytic active sites are metallic Ni, where CH4 was decomposed. On the addition of 1.0 wt% WO3 interacted strongly with γ-Al2O3, and non-reducible “WO3-Al2O3” support was formed. Further substitution of Al ions with W ions in 5Ni+1WO3/γ-Al2O3 catalyst formed NiWOAl species, which was reducible in the range of 900–1000 °C under the H2 stream. The strong interaction between WO3 and Al2O3 was also verified by the sudden decrease of crystallite size of Al2O3 after the 1.0 wt% addition of WO3. In total, the reducible NiO-interacting species was increased to about 38% just after the addition of 1.0 wt% WO3. Therefore, the 5Ni+1WO3/γ-Al2O3 catalyst showed marginal progress in CH4 conversion (74%) and CO2 conversion (80%).When the tungsten trioxide loading was 5.0 wt%, the total reducible NiO-species concentration over the catalyst surface was increased more than double that of the nonpromoted catalyst (Table 1). The 5.0 wt% tungsten-promoted catalyst contained all types of reducible NiO-interacting species with support. The acidic-basic profile of the catalysts were quite informative. The 5.0 wt% tungsten-promoted catalyst had a moderate amount of wide range of basic sites and the maximum number of acidic sites in comparison to the other catalysts. Moderate basic sites were related to moderate CO2 adsorption. Methane decomposition increases as surface acidity increases, as reported in [63]. In earlier literature, WO3 was also recognized for its ability to decompose CH4, which was decomposed over Ni and WO3 sites to form NiC3 and WC species. WC itself was active during the methane reforming reaction [64]. WC was oxidized by carbon dioxide into WO3 and CO [42]. However, we have not found carbide phases in XRD (WC, NiC3) and reducible W+6 species in H2-TPR. Overall, it could be said that the presence of Ni metal, derived from thermally stable NiAl2O4, and NiWOAl species and decomposition of CH4 over Ni metal pivoted the path of the optimum performance in DRM reaction over the 5Ni+5WO3/γ-Al2O3 catalyst, which resulted in >79% CH4 conversion and >83% CO2 conversion over 7.5 h of the time-on-stream test. Adjusting acidic and basic sites was crucial for the high catalytic performance of tungsten-promoted, Al2O3-supported Ni-based catalysts. As 5Ni+5WO3/γ-Al2O3 had a moderate number of basic sites and the maximum number of acidic sites, whereas 5Ni+3WO3/γ-Al2O3 had a moderate number of acid sites as well as the maximum number of basic sites. 5Ni+3WO3/γ-Al2O3 catalyst performed comparably less than 5Ni+5WO3/γ-Al2O3 in the mean of CH4 conversion (∼75%) and CO2 conversion (81%). After the optimum tungsten loading (5.0 wt%), catalytic activity declined even below the un-promoted catalyst.The systematic reaction scheme of dry reforming of methane over nonpromoted and tungsten-promoted γ-Al2O3 supported Ni catalyst is shown in Fig. 9 . The dry reforming reaction can be specified by “rate of CH4 dissociation” and “rate of carbon deposit oxidation by CO2”. The delay in carbon deposit oxidation results in carbon deposit over the catalyst surface. Nagaoka et al. found that CH4 was decomposed over metal active sites and acid sites of Al2O3 support [65] (Fig. 9 A-9a). However, dehydrogenated methane (CHx) over acid support was not reactive towards CO2-species, causing severe coke decomposition and deactivation of catalyst (Fig. 9α). Upon the addition of WO3 over alumina support, a rise in acidity is expected due to the formation of more acidic bridging W–O–Al [66]. In our case, 5Ni+5%WO3/γ-Al2O3 catalyst system had maximum acid sites, optimum CH4 conversion and least coke decomposition. It indicated that the new support, formed by “interaction of WO3 and Al2O3” as well as “substitution of Al ion by W ion”, was unique and CHx species over such support remained active with CO2-species for oxidation (Fig. 9B-9b-9β). Thus, 5Ni+5%WO3/γ-Al2O3 catalyst had maximum CH4 conversion (∼79%), CO2 conversion (∼83%), and the least coke decomposition (31 mg/g). Above 5.0 wt% WO3 incorporation (5Ni+7%WO3/γ-Al2O3 and 5Ni+9%WO3/γ-Al2O3 catalyst), the interaction between WO3 and Al2O3 species was weakened, and bulk WO3 grew, resulted into a drop in acidity and less CH4 conversion (Fig. 9C-9c-9γ). However, catalytic activity declined below the unpromoted catalyst [67]. 5Ni+7%WO3/γ-Al2O3 and 5Ni+9%WO3/γ-Al2O3 catalysts showed ∼71.4% and 70.3% CH4 conversion, respectively. It indicates that only a drop in acidity did not cause a such severe decline in catalytic activity. It might be due to the covering of potential catalytic active sites (metallic Ni) by WO3 overloading. Upon higher tungsten trioxide loading, unique WO3-Al2O3 support was defoliated and became inactive for the interaction of dehydrogenated methane (CHx) and CO2-species. It caused more coke deposition over 5Ni+7%WO3/γ-Al2O3 and 5Ni+9%WO3/γ-Al2O3 catalysts. The degree of crystallinity of deposited carbon over non-promoted and tungsten-promoted catalysts can be answered by Raman results. The degree of graphitization increased markedly from unpromoted to a tungsten-promoted catalyst, and it continued to increase nominally on increasing loading of tungsten trioxide.Alumina-supported Ni-based catalyst system with noble and non-noble metals-based promoters, including Mg [22], Ca, Ba, Sr [25], B [26], Si [30], Ti, Mo, Zr, W, Cu [31], Zn, Gd, La [32,34], Yb [35], Co [36], Ce [68] and Fe [69] have been reported for the dry reforming reaction. Table 2 compares catalytic activity in terms of CH4 conversion, CO2 conversion at different reaction/catalytic conditions such as catalyst weight, catalyst surface area, internal reactor diameter, feed ratio, activation temperature, reaction temperature, and time on stream (TOS). Our catalyst 5Ni+5WO3/γ-Al2O3 was superior and comparable to 5Ni+1Cu/Al [31] and Ni2Yb/γ-Al2O3 [35] catalysts over a minimum catalyst amount of 100 mg and moderate reaction temperature of 700 °C.ID= Internal Diameter of reactor, SA = surface area, CA = Catalyst amount, Cg = Carrier gas, AT/RT = Activation temperature/Reaction temperature, TOS = Time on stream, C (CH4) = Conversion of CH4, C (CO2) = Conversion of CO2, Ref. = Reference.The coke resistance and high catalytic activity for dry reforming of methane increased with increasing the loading of the WO3 promoter up to 5.0 wt%. We obtained 79% CH4 conversion and 83% CO2 conversion over 7.5 h of time-on-stream when using the 5.0 wt% WO3-promoted catalyst because it possessed stable, reducible NiAl2O4 and NiWOAl species, the moderate density of basic sites with the highest density of acidic sites. Increasing the weight load percent of tungsten trioxide above 5.0 wt% decreased the catalyst stability, increased the carbon deposition on the catalyst, and decreased the CH4 and CO2 conversions due to WO3 overloading covering catalytically active Ni sites. The degree of graphitization over spent catalyst was increased markedly from unpromoted to the tungsten-promoted catalyst and nominally on increasing loading of tungsten trioxide.The views and opinions expressed in this paper do not necessarily reflect those of the European Commission or the Special EU Programmes Body (SEUPB).The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to appreciate sincerely to Researchers Supporting Project number (RSP-2021/368), King Saud University, Riyadh, Saudi Arabia. Dr Ahmed I. Osman wishes to acknowledge the support of The Bryden Centre project (Project ID VA5048). The Bryden Centre project is supported by the European Union's INTERREG VA Programme, managed by the Special EU Programmes Body (SEUPB). The authors would like to thank Charlie Farrell for proofreading the manuscript. RK acknowledges the Indus University, India for supporting research.The following is the Supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijhydene.2022.09.313.

Syngas production via dry reforming of methane was conducted over 5 wt%Ni + xWO3/γ-Al2O3 (x = 1, 3, 5, 7, or 9 wt%) catalysts at 700 °C and ambient pressure for 7.5 h in a tubular fixed-bed reactor. Textural, morphological, and catalytic properties were investigated in relation to the weight percent of tungsten trioxide loading. The physicochemical properties of the catalysts were evaluated using XRD, N2-physisorption, TGA, H2-TPR, CO2-TPD, NH3-TPD, SEM, EDX, and Raman techniques. N2-physisorption analysis showed that tungsten trioxide promoter had a minor impact on the textural properties upon varying its weight percentage loading. With increasing tungsten trioxide loading, the total amount of reducible NiO-interacting species was increased over the catalyst surface. 5Ni+5WO3/γ-Al2O3 catalyst showed stable 79% CH4 conversions and 83% CO2 conversion with the lowest carbon deposition due to the presence of stable metallic Ni species (derived from reducible NiAl2O4 and NiWOAl), the highly acidic sites, and moderate basic sites.

Data will be made available on request.An attractive route towards CO2-free production of hydrogen is the thermal catalytic decomposition of methane. In addition to H2 formation, carbon nanomaterials, such as fibers, single or multiwalled carbon nanotubes, or other carbon nanostructures can be grown as a potentially valuable by-product [1–5 ]. (1) CH 4 g → 2 H 2 g + C s ∆ H 298 0 = 74.5 kJ mol These carbon nanostructures have interesting properties, e.g. high electron conductivity and/or mechanical strength [6]. They can be used in a wide range of applications like in semiconductors, energy storage, catalyst supports or building blocks for strong and light-weight materials (tennis rackets or planes) [7–10].The most commonly used catalysts for this reaction are first row transition metals (Ni, Fe and Co) supported on different inert materials [4,11–15]. These metals have a high carbon solubility which is essential for carbon transport [16]. Among these transition metals, supported Ni catalysts have been most widely investigated. Ni is highly active for methane decomposition, and can usually be implemented at reaction temperatures between 500 and 700 °C. However, Ni-catalysts suffer from fast deactivation due to surface encapsulation by carbon. Addition of copper to the Ni-catalysts improves the catalytic activity and lifetime [4,11,12,17–22]. It must be noted that the catalytic performance strongly depends on the support that is used [15,21,23]. For example, Pinilla et al. found that the carbon yield was significantly higher when using NiCu catalysts supported on Al2O3 or MgO rather than on TiO2 or SiO2 [21].To restrict the influence of support effects, we use a graphitic carbon support that is weakly interacting (chemically inert) and thermally stable [24–26]. With carbon as a support, tip growth for the formation of carbon nanofibers is expected (the metal nanoparticle is lifted from the support by the growing carbon nanostructure) [27–29]. Another practical advantage of using carbon as a support is that one only has to remove the metal catalyst, but not the support material from the final product, possibly avoiding harsh conditions for example to remove silica from the products. After the growth of carbon nanofibers, the graphitic support is only a small fraction of the final reaction product. For fundamental studies, it is also relevant that carbon is such a light element, that the contrast with the metal in electron microscopy makes more detailed studies possible.In this paper, we describe how the carbon yield and catalytic lifetime of carbon-supported NiCu catalysts depend on the reaction conditions. A thermal gravimetric analyzer was used to follow the carbon growth in-situ, by monitoring the sample weight. Interestingly, we found that two different operation regimes can be identified, with distinctly different reaction orders and activation energies, and different types of carbon nanostructures being formed. The highest carbon yield was obtained at the interface between these two operating regimes.Nickel(II) nitrate hexahydrate (Ni(NO3)2·6 H2O, ≥ 97%) and Copper(II) nitrate trihydrate (Cu(NO3)2·3H2O, 99%) were purchased from Sigma Aldrich. Graphene nanoplatelets (xGnPC-500, later referred to as C or GNP-500) were used as support material for the catalyst synthesis. This support material consists of stacked graphitic sheets ( ∼ 500 m2/g surface area, ∼ 0.9 mL/g pore volume) and was obtained from XG Sciences.The catalysts were prepared via incipient wetness (co-)impregnation. The carbon support, GNP-500, was impregnated with an aqueous solution containing one or both metal precursors (Ni(NO3)2·6 H2O and/or Cu(NO3)2·3 H2O, pH ∼1). The Ni catalyst was prepared with a 4 M Ni-solution. The bimetallic catalyst was prepared using a 3 M Ni, 1 M Cu precursor solution. After impregnation, the samples were dried overnight under dynamic vacuum. Thereafter, the samples were treated at 330 °C (mono- and bimetallic) for 2 h under nitrogen flow (200 mL min−1 gcat −1) and reduced at 280 °C (Ni-Cu) or 300 °C (Ni) for 2 h in 5% H2/N2 (200 mL min−1 gcat −1).Methane decomposition was performed in a thermogravimetric analyzer (TGA, PerkinElmer TGA 8000) coupled to a Mass Spectrometer (Hiden Analytical HPR-20). The sample weight was constantly monitored during the reaction. 1–2 mg catalyst powder was loaded in a ceramic crucible sample holder (d = 6 mm, h = 2 mm, set-up shown in Fig. S1.1). First, the catalyst was dried at 70 °C for 15 min under argon. To make sure that the catalysts were in fully reduced state before catalysis, the sample was reduced in-situ at 280 °C (Ni-Cu) or 330 (Ni) °C for 3 h in 5% H2/Ar (total flow = 100 mL min−1). During this step, the sample weight was closely monitored. As the sample weight stabilized within the 180 min in-situ reduction, we could safely assume that reduction was completed before the catalysis started (for details see Fig. S1.2, which shows the sample weight during the in-situ reduction step). Then, the sample was heated to reaction temperature (5 °C min−1) under Ar. As soon as the reaction temperature was reached, methane gas was introduced to the system by an external gas mixing device. The experiments were carried out with a total flow rate of 127 mL min−1 and at atmospheric pressure. The reaction temperature was varied in the range of 450–600 °C, while keeping the partial pressure of CH4 and total flow constant. The partial pressure of methane was varied between 0.20 and 0.40 bar and argon was used as balance gas. Blank experiments were performed to exclude reactivity of the sample holder or the support (for details, see supporting information Section S1.3.) All experiments were performed in duplo using the same catalyst batch (see section S1.4). The conversions in the experiments did not exceed 2%, which is an order of magnitude less than the equilibrium conversions at any conditions used. This comparison in shown in supporting information section S2.Nitrogen physisorption at 77 K was performed on a Tristar II Plus apparatus (Micromeritics) to characterize the surface area and porosity of the support material and catalysts. The BET surface area was determined in the relative pressure range of p/p 0 = 0.03–0.14. The total pore volume was determined from the adsorbed quantity at p/p 0 = 0.995. Prior to analysis, the samples were dried at 170 °C under vacuum for 24 h.The presence of crystalline phases and the crystallite sizes were analyzed using powder X-ray diffraction (XRD) in a Bruker D2 Phaser 2nd Generation diffractometer with a Co radiation source (λ = 1.7889 Å). A Bruker D8 Advance (Co irradiation) was used for X-ray diffraction under inert atmosphere.H2-Temperature programmed reduction (TPR, Micromeritics, AutoChem II 2920) was used to check the presence of bimetallic phases. 50 mg sieved fraction (75–150 µm) of catalyst (after heat treatment) was loaded in a quartz U-tube in between quartz wool. Prior to reduction, the sample was dried at 120 °C. Then, the sample was cooled down and heated up from 30 to 500 °C (5 °C min−1) in 5% H2/Ar (flow = 40 mL min−1). A thermal conductivity detector (TCD) was used to obtain the reduction profiles.Transmission Electron Microscopy (TEM) was used to determine the particle size of the fresh catalysts and to study the structure of the products formed during the decomposition experiments. A ThermoFischer Talos F200X was operated at 200 kV in TEM mode to capture bright-field images of the catalyst. Scanning transmission electron microscopy Energy dispersive X-ray spectroscopy (STEM-EDX) mapping was used to map the distribution of Ni and Cu in the catalyst before catalysis.Weight loadings were determined via ICP analysis, which was carried out by an external institute (the Mikroanalytisches Laboratorium Kolbe). The elements were measured on an Arcos Model Acros Spectro ICP-OES after microwave digestion on a CEM MARS 6.For the TGA measurements, the increase in sample weight during the reaction – when exposed to methane - is directly related to the amount of carbon that is formed; the carbon yield. The carbon deposition was normalized for the amount of Ni present in the sample. This also holds for the bimetallic catalysts, as copper itself is inactive for the reaction.When the reaction temperature is reached, methane is added to the gas feed. As a consequence, the total flow rate becomes higher and the gas density changes. Both events induce a small drift in observed sample weight due to the buoyancy effect [30,31]. In addition, after triggering the gas switch, it takes a moment before the new feed composition reaches the sample. For example, at 500 and 600 °C it takes about a minute before the sample weight starts to change significantly due to carbon nanostructure formation. Therefore, we have looked at when the increase in sample weight was the largest after the introduction of methane and took that as our t = 0 for each of the individual experiments. The results of each pair of duplo measurements were averaged. On average, the deviation between the two measurements was approximately 8% (for more details see supporting information section S1.4.).The carbon growth rate (r) was calculated by taking the derivative of the carbon yield with respect to time: (2) r t = ∆ C yield ∆ t where C yield is the carbon yield (gC/gNi), and t is the reaction time (min). The initial growth rate, r 0 , is the growth rate derived from the first two points of the reaction.In this work, the lifetime of the catalyst is defined as the time when r t = 0.01 ∙ r 0 . For the experiments with varying the partial pressure of methane, the lifetime of the catalysts was determined manually with a 10% error. For the temperature experiments, the catalysts did not deactivate to 1% of the initial growth rate within the measuring time. Therefore, the growth rate was fitted using the following equation of Borghei et al [32]: (3) r ( t ) = 1 1 + d − 1 r d t 1 ( d − 1 ) ∙ r 0 in which d is the deactivation factor, r d the deactivation rate, r 0 is the initial/maximal growth rate. This equation was derived by a kinetic study of carbon formation over NiCu/MgO catalysts. Important to note is that r d might be dependent on temperature and the partial pressures of methane and hydrogen. A further explanation of the lifetime determination can be found in section S3 of the supplementary information.Powder X-ray diffraction was used to characterize the crystallite phases present in the catalysts. Fig. 1a shows XRD patterns of the support material (black line) and pre-reduced catalysts. Diffraction peaks at 2θ = 30, 52 and 64° are characteristic for graphitic materials and are attributed to the graphite support [33]. Peaks at 2θ = 51 and 59° are ascribed to the presence of crystalline Ni and Cu, 2θ = 88° can be ascribed to Cu. As the weight loading of Cu in the bimetallic catalyst is low, it was not possible to resolve this peak in the NiCu/C diffractogram. Peaks at 2θ = 34, 43 and 74° indicate the presence of metal oxides. This indicates that the pre-reduced catalysts had partially been re-oxidized before or during the XRD measurement due to air exposure.H2-Temperature programmed reduction (TPR) was used to study the reduction behavior of the catalysts. Prior to the measurements, the catalysts only had been heat treated at 330 °C under nitrogen flow and had not yet been reduced. Fig. 1b shows the reduction profiles of the carbon-supported 15% Ni (orange), 13% Cu (blue) and 11% Ni 4% Cu (green) catalysts. For the Cu/C sample, a peak in hydrogen consumption is observed at Tmax= 145 °C which corresponds to the reduction of copper oxides to metallic copper. The Ni/C sample shows multiple reduction peaks which can be ascribed to reduction of NiO to Ni, weakly or more strongly interacting with the support [34,35]. It was calculated that theoretically approximately 60 mL H2/ gcat was needed to reduce all NiO to Ni. However, when integrating the whole peak area of the reduction profile (120–500 °C) a total of 118 mL H2/gcat was consumed ( Table 1). This can be explained by the fact that above 300 °C Ni catalyzes the reaction of hydrogen with the graphite support to methane, resulting in a hydrogen uptake [25].In the profile of the NiCu/C catalyst multiple peaks are present. The first peak is observed at Tmax = 175 °C, which is shifted 30 °C to a higher temperature with respect to the copper sample. The total hydrogen consumption of these peaks are listed in Table 1. The area underneath the first peak in the bimetallic sample is larger than the area of the peak in the Cu/C sample. Yet, the Cu content in the bimetallic sample is much lower (4 wt% compared to 13 wt%). Therefore, it can be concluded that not only the reduction of copper oxides but also the reduction of (part of) the nickel oxide contribute to the observed hydrogen uptake. From the fact that Cu and Ni influence each other’s reduction behavior, it can be concluded that they are in close proximity to each other. The other peaks are observed at similar temperatures as the particles of the Ni/C catalyst. Also here, support methanation becomes significant above 300 °C (see Fig. S4.1). Yet, it could also mean that there is still some isolated Ni present.Transmission electron microscopy was used to determine the average particle size and to study the metal distribution in the catalysts. We evaluated the particle sizes at the start of the reaction. In order to do this, we loaded catalyst powder in the TGA. Part of the regular method was used (as described in Section 2.3.) with a drying step, in-situ reduction and subsequent heating to 500 °C in argon (reaction temperature). After that, the sample was cooled down, and transferred to the TEM (under ambient conditions). Fig. 2 shows representative TEM images and the corresponding size distributions of the fresh Ni/C (top; a,b) and NiCu/C (bottom; c,d) catalysts. The metal particles are well visible in the bright-field TEM images due to the high contrast with the graphitic support. The metal particles (dark) are well distributed over the graphitic sheets (light gray). The average particle sizes are 8.2 ± 2.0 (Ni/C) and 10.8 ± 3.4 nm (NiCu/C). Fig. S4.2 provides a comparison between the particle sizes after ex-situ reduction and the fresh catalyst before the start of the reaction. From this, it can be seen that the bimetallic particles are a little bit larger than the monometallic particles. This is possibly due to more particle growth during heating. Scanning transmission electron microscopy energy-dispersive X-ray spectroscopy (STEM-EDX) was performed to map the distribution Ni and Cu in the particles (Fig. S4.3 in supporting information). It was observed that both Ni and Cu are present in the same particles. Table 2 summarizes the most important properties of the support, and the catalysts. The metal weight loadings were determined via ICP.We explored how the lifetime of the carbon supported Ni-based catalysts was influenced by the presence of Cu, and operation conditions (the reaction temperature and the partial pressure of methane). In addition, we looked at the carbon products formed under different reaction conditions. First, we compare the performance of the two catalysts. Fig. 3a shows the evolution of carbon production over time over 15% Ni/C (orange) and 11% Ni 4% Cu/C (green) catalysts, at 500 °C with a partial pressure of methane (p CH4 = 0.34 bar). First of all, it is observed that the initial growth rate over the NiCu/C catalyst is significantly higher than for the monometallic catalyst. In addition, the lifetime of the NiCu/C catalyst is much longer than for Ni/C. Together this leads to a 6–7 times higher final carbon yield. Thus, the bimetallic catalyst clearly outperforms the monometallic catalyst. Experiments were reproducible within an average relative error of 8% (see for details section S1.4. of the supporting information). Gas analysis during catalysis with a mass spectrometer did not show any formation of CO2 (see section S5). Fig. 3b shows an electron micrograph of the carbon nanofibers formed using the Ni/C catalyst. Thin and thick fishbone-type fibers are observed, originating from smaller and larger particles, respectively. The Ni particle at the top (magnification in Fig. 3c) shows the presence of carbon layers around the metal particle which could have stopped further carbon growth.It was proposed in literature that Cu helps to prevent carbon layer formation on the active surface of metal Ni(Cu) particles, thereby enhancing the lifetime and performance of the catalysts [36]. Yet, work of Pinilla et al. showed that the effect of Cu depends on the type of support [21]. Here, we show that for a carbon support the presence of Cu in the catalyst leads to a significant enhancement of the catalytic performance. One possible explanation is the influence of Cu on the particle size during catalysis. It was already postulated in Section 3.1. and shown in Fig. S4.2. that NiCu particles grow more at the reaction temperature in inert atmosphere than pure Ni metal particles, possibly due to a reduced melting point of the nanoparticles by Cu addition. In literature it has been reported that the catalytic performance can be highly dependent on the size of catalyst particles [37]. Optimal particle sizes were reported in the range of 10–40 nm depending on the reaction conditions and the composition of the catalyst [21,38]. Fig. 4a and c show TEM micrographs of the catalysts after 5 min reaction time at 500 °C with p CH4 = 0.34 bar. A lot of fibers were observed in both samples. It was clear that in the Ni/C sample, the fibers were smaller in diameter than most of the fibers in the NiCu/C sample. Also, a few very large particles were observed (50–60 nm) in the bimetallic sample. We have evaluated the average particle size in both cases. The size distributions are shown in Fig. 4b and d. The bimetallic particles have grown from 10.8 ± 3.4 nm to an average size of 23.4 ± 9.2 nm. On the other hand, the Ni particles have only grown from 8.2 ± 2.0 nm to 10.6 ± 2.1 nm. Clearly, Cu containing particles are more prone to particle growth as they have become more than twice as large. Under our reaction conditions, the NiCu particles probably obtain a more optimal size than the monometallic catalyst particles. Hence, we will further discuss the influence of reaction conditions on the catalytic performance specifically of NiCu catalysts.We investigated the influence of the reaction temperature on the performance of NiCu/C catalysts. In Fig. 5a the formation of carbon over time is shown for experiments at temperatures ranging from 450 to 600 °C using p CH4 = 0.34 bar. Increasing the temperature resulted in higher initial growth rates (r 0). Fig. 5b shows the relation between r 0 and the temperature in an Arrhenius plot. Surprisingly, the data cannot be fitted with a single straight line, but with two. We clearly distinguish two regimes: temperatures up to 500 °C in regime 1 (trendline in red) and starting from 550 °C in regime 2 (trendline in blue). This results in two apparent activation energies: 86 ± 8 kJ/mol and 45.4 ± 0.4 kJ/mol for regime 1 and regime 2, respectively. Assessment of the Weisz-Prater criterion showed that diffusion limitations are not present in these experiments (see supporting information Section S6 for more details) [39,40]. The activation energy is lower in the high temperature regime than in the lower temperature regime. In literature, values between the two are reported depending on the reaction conditions used. For example, Borghei and co-workers reported an Ea of 50.4 kJ/mol for a NiCu/MgO catalyst (6–10% CH4, 550–650 °C) [32]. Reshetenko et al. reported values in the range of 65–77 kJ/mol for NiCu/Al2O3 catalysts with varying Ni:Cu ratios [22]. However, for the first time we report that depending on the reaction conditions, two regimes can be distinguished, with different activation energies, indicating that different steps in the reaction sequence can be rate determining. Fig. 5c shows the lifetime of the NiCu/C catalysts at the different reaction temperatures. It is clear that increasing the temperature resulted in a strong decrease in catalytic lifetime. The lifetime was determined by fitting Eq. 3 to the growth rates obtained (a detailed explanation can be found in supporting information Section 3). In addition to the lifetime, also two other valuable parameters can be obtained from this fit: the deactivation rate and the kinetic deactivation order [32]. As expected from the decreased catalytic lifetime, the deactivation rate increases rapidly with increasing temperature (Fig. S3.4). Using the Arrhenius law the deactivation energy was found to be Ed = 70 ± 3 kJ/mol. Interestingly we could not distinguish two different regimes here (Fig. S3.4). The deactivation order was found to be 1.5–1.8 at 450–550 °C, yet becomes ∼1 at 575 and 600 °C (Fig. S3.4). This indicates a change in the deactivation mechanism at higher temperatures.The final carbon yield obtained for all experiments (after reaction time) is shown in Fig. 5d. An optimal yield was found around T = 500 °C. This corresponds to a maximal yield of approximately 18 gC/gNi. For monometallic Ni catalysts an optimum was reported in literature at 500–580 °C [5,41]. We identify the optimal temperature for our Ni-Cu/C system, and find that it is exactly in between the two temperature regimes.To further explore the characteristics of the two temperature regimes, we looked at the influence of the partial pressure of methane on carbon formation. For this, we varied the partial pressure of methane from 0.2 to 0.4 bar at a fixed temperature and at atmospheric pressure. Fig. 6a shows the growth curves for the experiments with varying methane concentrations in the feed conducted at 500 °C (regime 1). An increase of the initial growth rate was observed upon increasing the amount of methane in the feed (Fig. 6b). From the slope (0.91), we derived a close to first order dependence of the reaction rate on the methane partial pressure. This means that the reaction rate scales linearly with the amount of the reactant gas in the feed.The formation of carbon nanostructures from dissociated methane comprises many steps, but starts with the (dissociative) adsorption of methane, followed by recombination of hydrogen atoms yielding H2 and dissolving carbon atoms into the metal nanoparticle. A common explanation for first order reaction kinetics is that the overall reaction rate is limited by the amount of methane available at the surface of the metal catalysts. This is the case if the adsorption of methane is relatively weak (in the Henry regime of the adsorption isotherm) and hence linearly dependent on the methane concentration. In literature, the dissociative adsorption of methane is indeed several times reported to be the rate limiting step of this reaction, rather than the growth of carbon nanostructures [32,42–44].Increasing the CH4 partial pressure led to an enhanced carbon yield at this temperature which is illustrated in Fig. 6c. Fig. 6d shows the catalytic lifetime as function of the partial pressure. The lifetime increases when the amount of methane in the feed is increased. As the carbon supply by the metal nanoparticle seems rate limiting, we do not have an a priori expectation of the influence of the amount of methane on the lifetime, which depends on the fraction of carbon deposited as a carbon shell (leading to deactivation) rather than used for the formation of well-defined nanostructures. An explanation for the increased lifetime could come from the hydrogen produced during the reaction. If the reaction rate is higher, also the local H2 concentration increases. This H2 could react with the defective carbon covering the surface of the metal particle, rather than with the more stable carbon nanofibers that were grown. This helps to keep the surface clean and available, thereby avoiding deactivation [5].Interestingly, in literature there has not been consensus yet about the influence of the methane concentration on the catalyst lifetime. It is reported that the lifetime is negatively influenced by increasing the amount of methane in the feed [3,45–47]. On the contrary, recently Hadian et al. (2022) reported a slight increase in lifetime when increasing the methane partial pressure from 0.40 to 0.90 bar (total pressure 1 bar) [5]. They postulated that the difference between their work and previous work arose from working on a different scale and with a higher concentrations of methane in the feed. However, our reactor scale and methane concentrations are quite similar to the conditions reported in literature. Nevertheless, the reaction temperature reported in most papers is high (650–900 °C) [3,45–47], while we are conducting experiments also at lower temperatures (500 °C). We hence postulate that the lifetime of the catalysts correlates with the methane concentration depends on which of the two regimes we operate in.To clarify this point, our experiments were repeated at a higher temperature (575 °C). Based on the results of Section 3.2.2, we now operate in regime 2. Fig. 7a shows the dependence of the initial growth rate on the partial pressure at 575 °C. Now, the slope of the plot (and thus the reaction order) has increased to 1.47. This suggests that the dissociative adsorption of methane is not (exclusively) the rate limiting factor. It is more likely that other elementary reaction steps, such as the dissolution and migration of carbon, and/or the formation of the final carbon nanostructures, influence the overall reaction rate. For the carbon nanostructure growth probably more than one carbon atom is involved at a time, as the carbon nanofibers form layer by layer. This is in agreement with a reaction order in methane higher than 1 [48]. Fig. 7b gives the lifetime for the catalysts as a function of the methane partial pressure. In contrast to what had been found for regime 1 (at lower temperatures), at this higher temperature the lifetime is negatively influenced by an increase in partial pressure of methane, in line with the findings in other papers [3,45–47]. We have also evaluated the deactivation parameters (deactivation rate and order) for these experiments (Fig. S3.5). It was found that these parameters are not or barely influenced by the partial pressure of methane, but only with the temperature. The deactivation order was found to be 1.4–1.8 at 500 °C and 1.2–1.5 at 575 °C. This stresses the importance of analyzing in which of the two regimes one is operating to understand the influence of reaction conditions on the lifetime of the catalyst.Carbon structures formed during the experiments at 450, 500 and 600 °C using the 11% Ni- 4% Cu /C catalysts are shown in Fig. 8. Two different carbon nanofiber structures are formed: solid (full) fishbone fibers and fibers with a hollow core, referred to as hollow fibers. Relevant details to identify the two structures are highlighted with arrows in the TEM micrographs. For example, in Fig. 8a,b (operating at 450 °C) the formation of full fishbone fibers is observed. Occasionally, thin hollow fibers are also found (Fig. 8b), but at 450 °C, 90% of the product consisted of full fibers and only about 10% were hollow fibers. At 500 °C, different structures are formed as visualized in Fig. 8c,d. The formation of hollow fibers is now more pronounced. During this experiment, 68% of the products were full fibers and the rest were hollow fibers (Fig. 8f). During reaction at 600 °C, only hollow fibers had been formed which can be seen in Fig. 8e. The average fiber diameter was determined from the TEM images, the histograms are shown in Fig. S7.1. Typically, the fiber diameter is very close to the diameter of the particles they have grown from. At 450 °C the average fiber diameter is 15 ± 5 nm (n = 121) and at 600 °C it is 17 ± 5 nm (n = 91). Also, X-ray diffraction was used to characterize the used catalysts and products (Fig. S7.2). Peaks of metallic phases NiCu were observed in the diffractograms, indicating that the crystalline part of the catalyst nanoparticles is reduced, even after cooling down and exposure to air. The crystallite sizes range from 8.6 nm (450 °C) to 13.3 nm (600 °C). Unfortunately, the sample could not be uniformly distributed over the sample holder, due to its limited amount (∼5 mg), and hence additional peak broadening is caused by the uneven distribution, influencing the apparent crystallite size. Next to the peaks for the turbostratic support, additional peaks indicating an interlayer spacing of 0.334 nm were observed, typical for graphite. More details can be found in the supporting information (page 22–23). Fig. S7.3 shows the analysis at two different partial pressures. Here, no difference in average diameter was observed with different partial pressures. Fig. 8f gives a semi-quantitative overview of the carbon structures formed at the different temperatures. From this it is clear that the dominant carbon structure changes from mainly full fibers in regime 1 to mainly hollow fibers in regime 2. To the best of our knowledge, this is the first time that the dependence of the structure of the carbon formed is related to the growth conditions.The balance between the rates for methane adsorption and dissociation (carbon supply), carbon dissolution and transport and nanostructure formation is of key importance [49]. The rate-limiting factors during the carbon formation from methane depend on the nature of the catalyst and the reaction conditions and, as we have shown in this paper, two different regimes can be distinguished. A schematic overview of the different steps in the process is visualized in Fig. 9, based on the findings in this paper and literature.We consider three main steps: carbon supply (step 1), carbon transport (step 2), and carbon nanostructure formation (step 3). Step 1 comprises the adsorption and dissociation of methane on the catalyst surface, and the release of hydrogen gas and incorporation of carbon in the metal nanoparticle. Step 2 involves the transport of carbon atoms through the metal nanoparticle, via surface or bulk diffusion. Thereafter, carbon atoms can recombine to form solid carbon in different structures depending on, among others, the type of catalyst and the reaction conditions (step 3) [1,4,6,11,12,21,37].First, we consider the first step: methane adsorption and dissociation, yielding hydrogen and carbon atoms that are adsorbed or dissolved in the metal particle. Methane decomposition is an endothermic reaction, therefore, performing the reaction at lower temperatures will give rise to lower equilibrium conversions (see Fig. S2.1). Also kinetically, the carbon supply is slower at lower temperatures. In the low temperature regime (regime 1), a reaction order close to 1 in methane is found, with an apparent activation energy of 86 kJ/mol. This is in line with step 1, the carbon supply, being the rate determining factor, with a low coverage of the metal nanoparticles surface with methane (Henry regime). Physisorption is not an activated process, but the activation energy represents the dissociation barrier of the methane molecule. In literature, values for the activation energy of methane dissociative adsorption of nickel in the range of 71–127 kJ/mol have been reported [50–53], fully in line with our experimental values in regime 1.In this low-temperature regime, a higher methane concentration increases the overall carbon yield. It also slightly increases the lifetime of the catalysts, which is determined by the balance in step 3 between desired processes (3a and 3b, forming well-defined crystalline carbon nanostructures) and step 3c (forming less ordered/amorphous carbon which covers the metal nanoparticle surface and hence causes catalyst deactivation). We postulate that the slight improvement of the lifetime of the catalyst with methane concentration is caused by the fact that a higher reaction rate also causes a somewhat higher local hydrogen concentration, which shifts the balance slightly from the undesired formation of less stable and less crystalline carbon (3c) to the formation of desired nanostructures. This also means that the deliberate addition of hydrogen to the reaction mixture probably could much further improve the carbon yield [5].Now we consider the high temperature regime 2. It is clear that the rate limiting step in this regime is different from that in regime 1, as evidenced by the different activation energy (Fig. 5b) as well as a different reaction order (Fig. 7a). If step 2 (diffusion) would be rate limiting under any conditions, this would be expected to lead to a reaction order close to 0.5 in methane concentration. This is never observed. In the high temperature regime the reaction order in methane is about 1.5, even higher than the reaction order of 1 at the lower temperatures. It is likely that at these temperatures step 3 of the process, the formation of solid carbon, becomes (partially) rate limiting.The driving force for the formation of fibers from dissolved or adsorbed carbon atoms is its higher stability. The growth of the fibers is known to occur in stages, layer by layer, which involves several carbon atoms at a time and hence would be in line with a reaction order in methane higher than 1.0. Although no direct correlation can be made, because we do not know the carbon atom concentration in the metal nanoparticle as a function of the experimental parameters.It is expected that the formation of well-defined, crystalline, carbon nanofibers (3a and 3b) is thermodynamically more favorable, but kinetically less favorable, than the formation of low-crystallinity or disordered solid carbon. The combination of this kinetic enhancement and solid carbon formation as rate determining step, induces a shift from well-controlled carbon formation to the formation of disordered or amorphous shells. This is reflected in the observed decrease in catalytic lifetime with increasing reaction temperature as the formation of carbon shells around the catalyst particle cause deactivation [1,5,54].With increasing temperature (going from regime 1 to regime 2), also the structure of the carbon nanofibers changes from solid to hollow (Fig. 8). Different explanations are possible. A first one is that, as at these high temperatures the carbon structure formation is rate limiting, hollow nanofibers are kinetically easier to form than full solid carbon nanofibers [49]. Another possible explanation might be related to the mode of diffusion of the carbon in the metal nanoparticle. At higher temperatures the solubility of carbon in the nickel nanoparticle, which is already lowered by the addition of Cu, is further lowered. Possibly carbon surface diffusion becomes more dominant.Overall, it is clear that it is important to distinguish different operating regimes, where the transition between the regimes is dependent on the catalyst and process conditions. The sweet spot for the maximum carbon yield seems to be exactly between the two. At this transition, neither the methane decomposition nor the carbon nanostructure formation are particularly rate limiting, but the rates of these two processes are balanced.In this study, we evaluated how the operating conditions influence the carbon yield during the decomposition of methane over carbon supported Ni-based catalysts. Having Cu present next to Ni in the catalyst greatly enhanced the activity, catalyst lifetime, and, hence, carbon yield. Varying the reaction temperature showed two carbon growth regimes for the NiCu/C catalysts. Different activation energies and reaction orders were obtained in the two regimes. In addition, we found that the dominant carbon nanostructure formed changes from full, in lower temperature regime, to hollow fibers in the higher temperature regime. From this, we postulate that the rate determining step changes from methane dissociation to carbon transport and nanostructure formation. It also allows us to define the conditions at which the maximal yield is obtained, at the interface between the two regimes. S.E. Schoemaker: Conceptualization, Investigation, Methodology, Validation, Visualization, Writing – original draft. Tom A.J. Welling: Software, Visualization, Writing – review & editing. Dennie F.L. Wezendonk: Resources. Bennie H. Reesink: Writing – review & editing. Alexander P. van Bavel: Writing – review & editing. Petra E. de Jongh: Conceptualization, Writing – review & editing, Supervision, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work is part of the Advanced Research Center for Chemical Building Blocks, ARC CBBC, which is co-founded and co-financed by the Dutch Research Council (NWO) and the Netherlands Ministry of Economic Affairs and Climate Policy (Project number: 2018.017.C). The authors acknowledge Joy Bodde for her work on the synthesis and preliminary characterization of the catalysts used in this paper.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.cattod.2023.114110. Supplementary material .

An alternative route towards COx-free production of hydrogen is the thermal catalytic decomposition of methane. In addition to hydrogen, valuable carbon nanomaterials can be formed. Carbon nanofiber formation from methane decomposition over carbon supported NiCu catalysts was studied in-situ using a Thermogravimetric analyzer. We especially investigated how the carbon yield is influenced by reaction parameters. Based on experiments with varying temperature (450–600 °C), two distinct temperature regimes were identified. Different kinetic parameters were derived for the two regimes. Activation energies of 86 and 45 kJ/mol, and reaction orders in methane of close to 1 and 1.5, were found in the low and high temperature regimes, respectively. We postulate that at lower temperature the methane dissociation is rate limiting, while at higher temperature the carbon formation plays a more critical role. At low temperatures mostly full fishbone fibers are formed, whereas at higher temperatures mainly hollow fibers are formed. The maximum carbon yield is obtained at the transition between the two regimes, when the carbon supply and carbon nanostructure formation are balanced.

Nitrophenols (NPs) are widely presented as precursors, intermediates, or by-products in various industries, such as pharmaceutical, agrochemical, fine-chemical, dye, etc. [1, 2]. Due to the electron withdrawing effect of nitro group (-NO2) and benzene ring, as well as the conjugate effect, NPs are always stable molecules, which makes them hardly to be biodegraded through traditional wastewater treatment process. Fortunately, researchers have found that if the -NO2 group of NP is reduced to amino group (-NH2), the resulted aminophenol (AP) shows improved biodegradability and reduced toxicity [2].Generally, NP conversion to AP could be achieved by constructing NaBH4-based advanced reduction system [3–5]. Metal nanocatalysts exhibited the characteristics of large specific surface area and high surface activity, which thus attract much attention. Since the early report using metal nanoparticles (Au, Ag, Cu, Zn, etc.) for 4-NP reduction applying NaBH4 as reductant [6–10], a number of nanocatalysts have been developed for 4-NP reduction, including noble metal, transition metal and metal oxide/phosphide, etc. [11–15]. Considering their earth abundance and relatively low cost, transition metal, especially Ni-based catalysts have drawn much attention. Metallic Ni, nickel hydroxide, nickel phosphide in sole or loaded on support are reported powerful catalysts towards NP reduction, for example 4-NP to 4-AP [16–25]. The presence of support can not only provide abundant adsorption and reaction sites for reactants, but also may facilitate charge transfer, especially in the case of advantageous carbon material as catalyst support. For example, as catalytic center support, reduced graphene oxide, graphene, polymeric (polycaprolactone(PCL)/chitosan) nanofiber, N-doped carbon, metal organic framework-derived porous carbon, show promotional effects for 4-NP reduction [17, 19, 24, 26, 27].Attempts of NixPy for 4-NP reduction reaction prove its capability for 4-NP conversion to 4-AP [21–24, 28]. However, controllable synthesis of NixPy using sustainable and environmental-benign phosphorus precursor is still a challenging work. The commonly adopted synthesis strategies, such as organic precursor decomposition, solid phase conversion, hydro/solvothermal methods, require costly or high-hazardous risk phosphorus precursors (trioctylphosphine, (NH4)2HPO4, NaH2PO2, red phosphorus, white phosphorus etc.) [21, 22, 24, 28, 29]. Recently, we have demonstrated that phosphorus-containing biomass, such as yeast cell, rice bran, can serve as sustainable, cost-effective, and environmental-benign precursor for transition metal phosphide (Co2P, FexP, Fe2P) fabrication through an anoxic pyrolysis process [25, 30–32]. Meanwhile, carbon and nitrogen elements in biomass can be converted to advantageous carbon material as support for metal phosphide, bringing additional benefits for catalytic reactions.Besides of phosphorus-containing biomass, typical biomolecule of nucleic acid extracted from yeast cell is also demonstrated to be an appropriate precursor for CoP/N-doped carbon composite formation [33]. Interestingly, owing to the complexing ability of nucleic acid with Co2+, the molecular level mixing of phosphorus component with Co2+ leads to a dramatically different morphology of CoP/N-doped carbon and related high efficiency for organic pollutant removal through advanced oxidation process [33]. Adenosine triphosphate (ATP) is also one important biomolecule existing in cell, which contains abundant phosphorus, carbon and nitrogen element. Similar to nucleic acid, ATP also possesses the potential to form a molecular level mixing with metal ions [34], which may bring unexpected material property after pyrolysis treatment. In this study, for the first time, we adopted ATP as phosphorus precursor for nickel phosphide/carbon composite fabrication. By simply mixing ATP with Ni2+ in aqueous solution and lyophilization, single precursor of ATP-Ni complex was obtained for the following sample synthesis. An anoxic pyrolysis process ensured the formation of nickel phosphide/carbon composite, which was proved by X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS) characterizations. The catalytic activity of as-obtained sample for 4-NP conversion to 4-AP was evaluated with NaBH4 as reductant. The present work brings new insight into metal phosphide fabrication and develops an efficient catalyst for advanced reduction process.Adenosine 5′-triphosphate disodium salt (ATP-2Na, >98%) was purchased from Aladdin Co., Ltd. 4-nitrophenol (4-NP, 99%) was acquired from Adamas-beta. NiCl2∙6H2O (98%) and other commonly used chemicals and reagents were of analytical grade and provided by Chengdu Kelong Reagent Company. All regents were used without further purification. De-ionized water was used for all experiments.A single precursor of ATP-Ni complex was firstly prepared, which was subsequently transformed to nickel phosphide/biocarbon composite by anoxic pyrolysis. Detailly, 2 g of ATP was dissolved in 40 mL de-ionized water. After adding 0.405 g of NiCl2∙6H2O and stirring for 30 min, the clear solution was lyophilized in a freeze-drier to obtain the single precursor of ATP-Ni. Then, 1 g of ATP-Ni powder was placed in a porcelain crucible and treated by an anoxic pyrolysis process (900 °C, 2 h) under continuous Ar flow (30 mL min−1) in a tube furnace. The obtained black powder was thoroughly washed by de-ionized water and dried in a 60 °C oven. The finally obtained sample was denoted as Ni2P/BC. Control sample without Ni incorporation was also prepared following the same procedure and denoted as BC.X-ray diffraction (XRD) pattern of the as-synthesized sample was obtained on a Rigaku D/max-TTR III X-ray diffractometer with a Cu Kα (λ = 1.54056 Å). X-ray photoelectron spectra (XPS) were acquired on an X-ray photoelectron spectroscopy (Thermo Fischer, ESCALAB Xi+). The morphology of sample was observed on a field emission scanning electron microscope (FESEM, JSM-7610F, JEOL Ltd., Japan) and a transmission electron microscope (TEM, JEOL JEM-2100 F). N2 adsorption/desorption measurement was conducted on a Micromeritics ASAP 2020 equipment. Sample was degassed at 150 °C for 8 h before measurement. The specific surface area was calculated by Brunauer-Emmett-Teller (BET) method. The total pore volume data was obtained at P/P0=0.986.The 4-NP reduction reaction was conducted in a 3 mL quartz cuvette adopting the as-synthesized Ni2P/BC sample as catalyst and NaBH4 as reductant. A 4-NP stock solution of 1 g L −1 was prepared. Catalyst of Ni2P/BC was dispersed in de-ionized water by ultrasonication to form a stock suspension of 1 g L −1. The NaBH4 solution of 10 g L −1 was freshly prepared before each catalytic reaction. In a typical reaction run, 50 μL of 4-NP stock solution, 1 mL of de-ionized water, 1 mL of freshly prepared NaBH4 solution were added into a 3 mL quartz cuvette in sequence. After adding 75 μL of catalyst suspension, the cuvette was put into a UV–visible spectrophotometer (UV-1800 PC, MAPADA Instruments) immediately. At pre-determined time interval, a full wavelength spectrum scan (250–500 nm) was conducted to monitor the concentration change of 4-NP. Correspondingly, the initial 4-NP concentration, catalyst dosage and NaBH4 dosage are 23.53 mg L −1, 35.29 mg L −1 and of 4.706 g L −1, respectively. The weight ratio and molar ratio of NaBH4:4-NP was 200 and 735, respectively.Control catalytic experiment without catalyst was conducted following the same procedure as described above.In order to test the reusability of catalyst, cycle runs were conducted by supplementing 4-NP stock solution (50 μL) and NaBH4 (50 mg) into the cuvette after every 8 min catalytic reaction.Phosphorus-containing biomass (yeast cell, nucleic acid, rice bran, etc.) has been proved to be effective and environmental-benign precursor for transition metal phosphide synthesis [25, 30–33, 35]. During anoxic pyrolysis treatment, high-valance phosphorus of phosphate (P5+) in biomass could be reduced to low-valance phosphorus in phosphide (P y −) with the assistance of in-situ generated reducing gasses (CO, H2, etc.) [30, 33]. ATP is one typical biomolecule containing three phosphate groups in one molecule, which also possesses abundant carbon, oxygen and hydrogen elements. Meanwhile, the phosphate group in ATP could complex with metal ions (M x +) [34], leading to a single precursor of ATP-M. The molecular mixing of different precursors is believed to be beneficial for the formation of uniform sample [33, 36]. Therefore, it is considered that ATP may be one appropriate precursor for transition metal phosphide (TMP) synthesis.By simply mixing ATP and Ni2+ in aqueous solution and lyophilization, the single precursor of ATP-Ni was obtained. After pyrolyzing ATP-Ni complex at 900 °C for 2 h under Ar flow, nickel phosphide crystal with Ni2P nature was obtained as evidenced by the XRD pattern shown in Fig. 1 . Based on the Scherrer equation, the fitting results for several peaks ((111), (201), (210), (300), (211) plane) were employed for estimation of the particle size, which was calculated to be 46.3 ± 2.6 nm. XPS characterization results shown in Fig. 2 give another evidence of nickel phosphide formation. Besides of the typical oxidized Ni species (856.6 eV, 874.6 eV) and satellite (862.2 eV, 880.6 eV) peaks, another XPS Ni 2p peak with low binding energy of 853.4 eV is found (Fig. 2a), which can be assigned to Niδ+ (0 < δ < 2) in Ni-P bonding [21, 37]. The deconvolution of XPS P 2p spectrum gives three peaks (Fig. 2b), which can be assigned to P-O (133.7 eV), P-C (132.0 eV) and P-Ni (129.9 eV) bonding, respectively [21, 37]. Therefore, Ni2P is successfully synthesized by the designed experimental procedure. Considering the appearance of P-C bond in Ni2P/BC sample (Fig. 2b), it is suggested that P doping into biocarbon framework exists [37]. It is also worth to mention that N doping into biocarbon is achieved as evidenced by the formation of CN bond and N-related species (Fig. 2c and 2d) [38]. The abundant P and N elements in ATP molecule (16.86 wt% P, 12.71 wt% N) enables the possibility of P and N co-doping into biocarbon. Foreign element doping into carbon framework may alter the original carbon structure and provide abundant free flow of electrons, thus beneficial for catalytic reactions [24, 37, 38]. Fig. 3 shows the morphology of as-obtained Ni2P/BC sample. Thin layer sheet-like structure with wrinkles is found for biocarbon (Fig. 3a). The formation of such carbon structure is believed to be related to the in-situ generated abundant gasses during pyrolysis process, which are also reported when applying biomass, guanine, hexamine as precursors for advanced carbon materials formation [32, 39, 40]. Ni2P with large particle size (hundreds of nanometers) is found on the surface of biocarbon (Fig. 3b). Besides, TEM of as-synthesized sample clearly shows that Ni2P with an average particle size of 2.4 nm are inlaid on layer sheets (Fig. 3c). Both large particles of hundreds of nanometers and small particles of ∼2.4 nm exist for Ni2P. The measured lattice distance of 0.227 nm for Ni2P particle (Fig. 3d) agrees well with the d spacing of Ni2P (111) plane (0.221 nm), which further implies the successful formation of Ni2P. Normally, thin layer sheet-like structure will bring high specific surface area, which is an advantageous feature for heterogeneous catalyst. Indeed, N2 adsorption-desorption measurement for Ni2P/BC sample gives a type IV isotherm with H3 hysteresis loop (Fig. 4 a), implying the formation of assembled layer structure. The calculated specific surface area (S BET) and pore volume (V total) are as high as 261.06 m2 g −1 and 0.3457 cm3 g −1, respectively. The average pore diameter is measured to be 5.30 nm.Overall, we adopted the typical biomolecule of ATP as an unusual P precursor. Through the formation of ATP-Ni complex and followed anoxic pyrolysis, Ni2P nanoparticles were formed and loaded on sheet-like P/N-doped biocarbon. The advantageous features of Ni2P/BC sample, such as good crystallinity of Ni2P nanoparticle, P and N doping into biocarbon, high specific surface area, may bring unexpected catalytic performance.Supported or unsupported noble metals (Au, Pt, Pd, Ag, etc.) and transition metals (eg., nickel) are typically active catalysts for 4-NP reduction to 4-AP in the presence of NaBH4 [2, 11, 41, 42]. Since TMPs always exhibit metallic properties, it is anticipated that the developed Ni2P/BC may bring high activity towards 4-NP reduction.The capability of Ni2P/BC for 4-NP reduction to 4-AP in the presence of NaBH4 was firstly monitored by a UV-visible spectrophotometer. As illustrated in Fig. 5 , 4-NP solution presents with light yellow color, corresponding to a maximum absorbance peak at 318 nm. After adding NaBH4 into the solution, a bright yellow color is observed with the maximum absorbance peak shifting to 400 nm, indicating the generation of 4-nitrophenolate [21, 41]. Once Ni2P/BC is added into the solution, a distinctly color change from bright yellow to pale yellow and finally to colorless is observed, which is the typical characteristic of 4-NP reduction reaction [26, 41]. Correspondingly, the maximum absorbance peak shifts from 400 nm to 300 nm, which belongs to 4-AP. Control experiment was conducted to exclude the contribution of 4-NP adsorption by solid catalyst. As shown in Fig. 6 a, in the absence of NaBH4, the absorbance peak belonging to 4-nitrophenolate (λ = 400 nm) was slightly increased. The possible reason is that the surface functional groups on biocarbon support and the possible existence of Na+ in Ni2P/BC from ATP precursor may induce the hydrolyzation or ion exchange of 4-NP, leading to the formation of 4-nitrophenolate (λmax= 400 nm). However, both the reaction and adsorption of 4-NP by Ni2P/BC are negligible. Around 9.6% of 4-NP is removed. Similarly, in the absence of Ni2P/BC, sole NaBH4 is also not efficient for 4-NP reduction with only 1.7% of 4-NP conversion after 30 min (Fig. 6b). Therefore, both catalyst of Ni2P/BC and reductant of NaBH4 are essential for 4-NP reduction.We further monitor the concentration change of 4-NP versus time by UV-visible spectra as presented in Fig. 7 a. Along with the reaction proceeding, the maximum absorbance peak belonging to 4-nitrophenolate (λ = 400 nm) gradually decreases and finally vanishes after 180 s, indicating the fully reduction of 4-NP. Meanwhile, absorbance peak belonging to 4-AP (λ = 300 nm) gradually appears and increases. Since the concentration of NaBH4 is much higher than that of catalyst dosage, it is reasonable to use pseudo-first-order reaction model to investigate the reaction kinetic. Relevant kinetic equation can be written as ln (C t/C 0) = −kt, where C t and C 0 are the concentration of 4-nitrophenolate ion at time t and 0, respectively, and k represents the reaction rate constant. A good linear fitting is found between ln (C t/C 0) and t as displayed in Fig. 7b, which gives a k value of 0.019 s −1. Such value is comparable with other transition metal (Ni)-based catalysts for 4-NP reduction in the presence of NaBH4 (Table 1 ). If comparing the normalized rate constant by introducing catalyst dosage, the developed Ni2P/BC catalyst in this study possesses a k normal value of 253 s −1 g −1, which outperforms most of reported Ni-based catalysts (Table 1), and even better than a few noble metal-based catalysts [11, 43].As a composite material, both Ni2P nanoparticles and thin layer sheet-like carbon structure exist in Ni2P/BC catalyst. Carbon materials are also reported active for 4-NP reduction [44–46]. Therefore, it is necessary to understand the contribution of biocarbon for 4-NP reduction. As shown in Fig. 7c, biocarbon sample of BC obtained from ATP pyrolysis does show catalytic activity for 4-NP reduction to 4-AP. However, it requires up to 420 s to achieve satisfactory conversion efficiency, which is much longer than that of Ni2P/BC catalyst (180 s). The calculated pseudo-first-order reaction rate constant of 0.0154 s −1 is also lower than that of Ni2P/BC catalyst (k = 0.019 s −1). Therefore, biocarbon structure not only supports Ni2P nanoparticles, but also participates in 4-NP reduction reaction. And the presence of Ni2P nanoparticles in the composite material obviously accelerates reaction.The reusability of Ni2P/BC was tested in consecutive reaction runs as illustrated in Fig. 8 . After each 8 min-reaction run, fresh 4-NP and NaBH4 were supplemented into the reaction solution for the next run. Therefore, along with the vanishing of 4-nitrophenolate peak after each run, 4-AP is accumulated in reaction cell with continuously increased absorption peak intensity (λ = 300 nm, Fig. 8a). It is noticed that even after 10 consecutive runs, the conversion efficiency of 4-NP to 4-AP still remains as high as 95.8% (Fig. 8b). Therefore, the good reusability of Ni2P/BC for 4-NP reduction reaction is verified.As illustrated in Fig. 9 , 4-NP reduction reaction over Ni2P/BC in the presence of NaBH4 can be described by Langmuir-Hinshelwood (L-H) heterogeneous catalytic reaction model [2, 41, 49]. Firstly, BH4 − is adsorbed on the surface of catalyst and hydrolyzed to BO2 − and active hydrogen species. Meanwhile, 4-nitrophenolate ions resulted from 4-NP are also adsorbed on the surface of catalyst. The contact of active hydrogen species and 4-nitrophenolate ions results in the reduction of nitro group to amino group, leading to the formation of 4-AP finally.Morphology and structure analysis indicates that the developed Ni2P/BC composite material possesses the typical features of dispersed Ni2P nanoparticles with good crystallinity, thin layer sheet-like biocarbon support and high specific surface area, N and P-doping into carbon framework, etc. Such features are beneficial for 4-NP reduction reaction. Firstly, the point of zero charge value (pHpzc) for Ni2P/BC was measured to be 7.93. The pH of 4-NP solution was determined to be 6.85. Therefore, the surface of Ni2P/BC is positively charged under investigated reaction condition, indicating that the adsorption of 4-nitrophenolate anions and BH4 − ions on Ni2P/BC surface was favorable. Meanwhile, Ni2P could efficiently adsorb active hydrogen species owing to the electronegativity of Py− in Ni2P [37]. Secondly, N and P doping into biocarbon could change the electronic structure of carbon framework and bring strong positive polarization charges, which is favorable to attract negatively charged 4-nitrophenolate anions [24, 37, 38]. Thirdly, the high specific surface area could provide abundant adsorption sites as well as reaction sites, which is favorable for the contact between reactants. Fourthly, metal phosphide of Ni2P could exhibit metal-like property, which ensures the good electron conductivity. Meanwhile, thin layer biocarbon obtained from high temperature treatment also possesses good electron conductivity. Therefore, fast electron transfer is suggested for Ni2P/BC material, which is critical for the fast and efficient reduction of 4-NP. Overall, the high affinity of Ni2P/BC towards reactants, as well as the good electron conductivity and transfer efficiency enables its superior activity for 4-NP reduction reaction.In summary, Ni2P/biocarbon composite was successfully assembled by adopting ATP as an unusual P precursor through an anoxic pyrolysis process. Benefiting from the molecular mixing of Ni2+ and P precursor, as well as abundant N and P contents in ATP, the as-obtained biocarbon showed thin layer sheet-like structure with wrinkles and both hundreds of nanometers and ∼2.4 nm small particles of Ni2P were dispersed on the surface of biocarbon. When applied for advanced reduction process, the developed Ni2P/BC sample exhibited the features of superior catalyst for 4-NP reduction to 4-AP, such as fast reaction kinetic, good reusability, etc. The superior catalytic activity of Ni2P/BC was suggested to be related to its high affinity for reaction substrates, high specific surface area of thin layer biocarbon and Ni2P nanoparticles with good crystallinity. This study not only enriches the synthesis method for transition metal phosphide, but also provides a highly efficient catalyst for advanced reduction process.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors gratefully acknowledge funding support from the National Natural Science Foundation of China (grant number 21808147), Sichuan University “Chemical Star” Excellent Young Talents Training Program (2021) and Institute of Engineering Technology of Petro China Southwest Oil & Gas Field Company. We would like to thank Yanping Huang from Center of Engineering Experimental Teaching, School of Chemical Engineering, Sichuan University for the instrument support (Field Emission Scanning Electron Microscope, JEOL JSM 7610F).

Conversion of 4-nitrophenol (4-NP) to 4-aminophenol (4-AP) through advanced reduction process is important to lower the hazardous risk of 4-NP, which requires the development of highly efficient catalyst. Herein, we proposed a novel strategy to fabricate nickel phosphide/carbon composite for 4-NP reduction in the presence of NaBH4. An unusual precursor of adenosine 5′-triphosphate (ATP) was applied to provide the essential phosphorus and carbon elements for nickel phosphide and biocarbon, respectively. ATP-Ni complex was firstly prepared and then transformed to nickel phosphide/biocarbon composite by an anoxic pyrolysis treatment. Material characterizations confirmed both large particles of hundreds of nanometers and small particles of ∼2.4 nm existing for Ni2P, which were dispersed on N/P-doped biocarbon with thin layer sheet-like structure. When applied for 4-NP reduction, Ni2P/BC exhibited fast reaction kinetic with a pseudo-first-order reaction rate constant of 0.019 s −1 (normalized rate constant 253 s −1 g −1) and good reusability in 10 consecutive reaction runs (95.8% conversion efficiency). It is suggested that such superior activity may originate from the high affinity of Ni2P/BC towards reactants and the good electron transfer ability of Ni2P/BC. This study not only enriches the synthesis strategy of transition metal phosphide, but also provides a superior catalyst for advanced reduction process.

The oxygen evolution reaction (OER) is a crucial half reaction to realize efficient hydrogen production via water electrolysis [1–20]. To date, tremendous efforts have been devoted to designing advanced OER catalysts in low price and high earth-abundance, and current studies indicate that a pre-oxidation process with surface reconstruction in phase, composition and morphology is essential for such transition metal-based catalysts, or so-called “pre-catalysts” [21–24]. Therefore, designing pre-catalysts with high reaction tendency and rich reactive surface sites is of great promise in enhancing the OER behavior.During the past few years, the Prussian blue analogues (PBAs) have attracted substantial research interests as efficient OER pre-catalysts owing to their multiple advantages including tunable chemical compositions in cationic and anionic sites, various nanostructures and high reaction tendency towards the pre-oxidation [25–28]. For instance, our group proposed a binary CuFe PBA pre-catalyst for highly active and ultrastable OER catalysis, which undergoes an obvious activation owing to the efficient pre-oxidation process with in-situ formed catalytically active species [29]. Zou et al. demonstrated that the introduction of Zn ions in CoFe PBA could simultaneously enrich the catalytically active Co3+ ions and increase the surface area, thereby leading to enhanced OER performance than the binary counterparts [30]. Tour group proposed that the cationic modulation in CoNi–Fe PBAs could effectively regulate the OER behavior owing to multi-metal synergy, and indicated that the catalytically active Ni and Co sites could lead to lower onset potential and easier activation kinetics, respectively [31]. Recently, our group developed a high-entropy amorphous oxycyanide pre-catalyst derived from the thermal oxidation of CoFeNiCuMn PBA, which undergoes a facile pre-oxidation process and achieves higher OER performance than the counterparts with fewer metallic elements [32]. Therefore, realizing rational elemental modulation in PBAs would be of great significance in promoting the pre-oxidation process and OER behavior.To date, the elemental modulation strategies are basically focused on the cationic sites, while systematic elemental modulation in both cationic and anionic sites of the PBA structure and the related elucidation of structure-performance relationship still remain blank [33,34]. Conducting synergistic dual elemental modulation in cationic and anionic sites of the PBA pre-catalysts is expected to be effective on further optimizing the OER behavior, and figuring out the role of components during the pre-oxidation process and OER catalysis is of high significance for designing advanced PBA-based catalysts for related applications. Aiming on the above considerations, herein we proposed a controllable dual elemental modulation in both cationic and anionic sites of the multi-metal PBA pre-catalysts, realizing highly active and ultrastable OER performance. Detailed investigations indicate that the Co ions in cationic sites is essential for the high intrinsic activity, and the multi-metal synergy could further enhance the intrinsic activity of the PBA pre-catalyst. In addition, mixed FeIIICoIII cyanide anions could benefit the OER process owing to the enriched Co3+ active sites and intermetallic synergy. As a result, the optimal NiCuCoII–FeIIICoIII PBA (denoted as NiCuCoII–FeIIICoIII) pre-catalyst with high intrinsic activity, enriched local Co3+ sites and optimal multi-metal synergy displays a low η10 of 288 ​mV, and a remarkable 1.4–61.2 times enhancement in OER activity can be identified compared with the counterparts with variable cations. Not only that, benefitted from the facile pre-oxidation process of NiCuCoII–FeIIICoIII, abundant high-valence Ni, Cu and Fe active species can be in-situ formed and accumulated, resulting in substantially enhanced OER activity with ultrastable durability. The activated NiCuCoII–FeIIICoIII shows an ultralow η10 of 251 ​mV and a 1.81 times enhancement in OER activity, outperforming most PBA-based/-derived catalysts and making the PBA pre-catalyst with dual elemental modulation a promising candidate for water electrolysis.The chemicals were purchased from SinoPharm Chemical Reagent Co., Ltd..Taking the synthesis of NiCuCoⅡ-FeⅢCoⅢ as an example, 2 ​mmol NiCl2·6H2O, 2 ​mmol CuCl2·2H2O, 2 ​mmol CoCl2·6H2O and 9 ​mmol of trisodium citrate dehydrate were dissolved in 20 ​mL of distilled water by vigorous stirring. Meanwhile, 2 ​mmol K3 [Co(CN)6] and 2 ​mmol K3 [Fe(CN)6] were dissolved in 20 ​mL of distilled water under stirring. Then, the above solution was mixed by stirring at 600 ​rpm for 20 ​min. The powdery products were collected by centrifugation, rinsed with water and ethanol, and dried under vacuum. The other PBAs were fabricated by the same operation, and the elemental modulation can be realized by simply adjusting the kinds and amounts of the salt precursors. The metal salts used in the synthesis involve metal cations (Ni2+, Cu2+, Co2+ and Fe2+) and metal anions ([Co(CN)6]3- and [Fe(CN)6]3-). The specific amounts of the salt precursors are shown in Table 1 .All the electrochemical measurements were performed in a three-electrode system linked with an electrochemical workstation (Ivium Vertex. C. EIS). All potentials were calibrated to a reversible hydrogen electrode (RHE) according to the Nernst equation and the data were presented without iR correction. Typically, 4 ​mg catalyst and 50 ​μL Nafion solution (Sigma Aldrich, 5 ​wt%) were dispersed in 1 ​mL water-isopropanol mixed solution (volume ratio of 3:1) by sonicating for at least 30 ​min to form a homogeneous ink. Then 5 ​μL of the dispersion (containing 20 ​μg catalyst) was loaded onto a glassy carbon electrode with 3 ​mm diameter, resulting in a catalyst loading of 0.285 ​mg ​cm−2. The as-prepared catalyst film was allowed to be dried at room temperature. The cyclic voltammetry (CV) and linear sweep voltammetry (LSV) with a scan rate of 2 ​mV ​s−1 were conducted in O2-purged 1 ​M KOH solution. A Hg/HgO electrode was used as the reference electrode, a platinum gauze electrode (2 ​cm ​× ​2 ​cm, 60 mesh) was used as the counter electrode, and the glassy carbon electrodes loaded with various catalysts were served as the working electrodes. The electrochemical impedance spectroscopy (EIS) measurements were operated in the same configuration at 1.55 ​V vs. RHE from 10−2–105 Hz.The PBA pre-catalysts with the cationic and anionic elemental modulation were fabricated via a rapid room-temperature approach by simply adjusting the metal salt precursors (Experimental section). In brief, the total amounts of metal salts were fixed, while the kinds and amounts of specific cations and anions were modified (Table 1; cations: Ni2+, Cu2+, Co2+ and Fe2+; anions: [Co(CN)6]3- and [Fe(CN)6]3-). The as-obtained products were named according to the type of cations/anions, and the valence of Co and Fe was labeled. For instance, the product fabricated by Ni2+, Cu2+, Co2+, [Co(CN)6]3- and [Fe(CN)6]3- was named as NiCuCoII–FeIIICoIII, in which trimetallic cations and mixed FeIIICoIII cyanide anions were involved.The X-ray diffraction (XRD) patterns in Fig. 1 A indicate that all the products are of high phase purity with high consistency to the standard pattern of PBA (JCPDS card No. 01–0239). Besides, the characteristic M ​− ​C and CN bonding can be detected from the Fourier transform infrared spectra (FT-IR) and Raman spectra (Figs. S1–2), further confirming the formation of PBA structure. The transmission electron microscopy (TEM) images in Fig. 1B and S3-8 reveal that the PBA products are in nanocube morphology with size of 50 ​nm. Of note, as the number of elements increases, curved surface of nanocubes can be observed, which may originate from the lattice relax owing to the high surface energy caused by the multi-metal nature. In addition, the selected area electron diffraction (SAED) pattern of NiCuCoII–FeIIICoIII in the inset of Fig. 1B shows weak 4-fold symmetry, which can be indexed to the (100) and (200) facets of the PBA lattice. Of note, bright halo can be observed from the SAED pattern, demonstrating the existence of amorphous component in NiCuCoII–FeIIICoIII. The high-resolution TEM (HRTEM) image further reveals the poor crystallinity, and only short-range ordering of the PBA lattice can be identified (Fig. 1C). The low crystallinity can be attributed to the multi-metal feature which brings in considerable lattice mismatch due to the different ion radius of metallic elements. The low crystallinity and disordered lattice could lead to the exposure of more undercoordinated metal sites, which is beneficial to the pre-oxidation process and OER catalysis [35].The high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) image and element maps of NiCuCoII–FeIIICoIII reveal the homogeneous distribution of Ni, Cu, Co and Fe in the nanocubes (Fig. 1D–E), further confirming the multi-metal feature of the product. Besides, X-ray photoelectron spectroscopy (XPS) was used to survey the compositional and valence information of the products. As shown in Fig. 2 and S9-14, the mixed valence feature of metallic elements and the characteristic signals of CN can be confirmed for all the products, further demonstrating the successful elemental modulation of the multi-metal PBA pre-catalysts. Taking the NiCuCoⅡ-FeⅢCoⅢ as an example, the Ni 2p spectrum can be indexed to Ni3+ (875.0 and 857.3 ​eV) and Ni2+ (873.6 and 856.2 ​eV), indicating the mixed valence of Ni (Fig. 2 A) [34,36]. In addition, it can be observed two dominant peaks for the Cu2+ at the binding energies 935.7 and 956 ​eV (Fig. 2B) [37,38]. Besides, Co 2p spectrum indicates the co-existence of Co2+ and Co3+ in NiCuCoⅡ-FeⅢCoⅢ, where the main peaks at 782.4 and 797.7 ​eV can be labeled as Co2+ ions, and the peaks at 781.9 and 796.8 ​eV can be assigned to Co3+ species (Fig. 2C) [39]. The binding energy of the Fe 2p3/2 region can be assigned to two peaks at 708.3 and 710.0 ​eV (Fig. 2D), which match well with the oxidation states of Fe2+ and Fe3+, respectively [39]. Furthermore, the binding energy of the Fe 2p1/2 region can be deconvoluted into two peaks, which are located at 721.2 and 723.8 ​eV corresponding to Fe2+ and Fe3+. In addition to the metallic elements, non-metallic elements such as C, N, and O were also investigated by XPS analyses. Apart from the C–C reference, CN, C–O, and CO bonds can be identified at the binding energies of 285.2, 286.1 and 288.6 ​eV, respectively (Fig. 2E) [29]. The presence of CN can be further verified by the N 1s spectrum, where the intensive peaks centered at 398.0 ​eV can be detected (Fig. 2F) [40]. The O 1s spectrum can be deconvoluted into two independent signals, where the peak at 533.3 ​eV can be assigned to the C–O bonds, and the peak at 532.0 ​eV indicates the presence of crystal water in the PBA lattice (Fig. 2G) [41].The OER behavior of the PBA pre-catalysts with elemental modulation on cationic and anionic sites was investigated in 1 ​M KOH solution using a three-electrode system. As shown in Fig. 3 A, the linear sweep voltammetry (LSV) curves of the PBA pre-catalysts with mixed FeIIICoIII cyanide anions and variable cations reveal distinct activity towards OER catalysis, and the quaternary NiCuCoII–FeIIICoIII displays earlier OER onset and larger current density. For instance, NiCuCoII–FeIIICoIII only requires an overpotential (η) of 288 ​mV to achieve a 10 ​mA ​cm−2 current density, which outperforms the counterparts and previous PBA-based/-derived catalysts (Table 2 ). Besides, the geometric current density (jgeo) of NiCuCoII–FeIIICoIII reaches 195.8 ​mA ​cm−2 at η ​= ​500 ​mV, which is 1.7–61.2 times larger than that of the counterparts with variable cations. Interestingly, the OER activity is not directly correlated to the number of elements in cationic sites. The PBA pre-catalysts with Co cations display higher activity, and the OER behavior could be further enhanced by introducing more metallic elements in the cationic sites. That is, the Co ions in the cationic sites of PBAs are the dominating active sites OER catalysis, and the catalytic ensemble effect guaranteed by the multi-metal synergy in cationic sites could further improve the OER behavior. Not only that, the effect of elemental modulation in the anionic sites of PBAs was also surveyed. As shown in Fig. 3B, NiCuCoII–FeIIICoIII with mixed FeIIICoIII cyanide anions exhibits higher OER activity than the counterparts with unary anions (NiCuCoII–CoIII and NiCuCoII–FeIII), where a 1.4–2.2 times enhancement in jgeo can be identified. Hence, the PBA pre-catalyst with Co-containing multi-metal cations and mixed FeIIICoIII cyanide anions is favorable for OER catalysis, and the improved OER activity could be ascribed to the high intrinsic activity of the local Co3+ sites as well as the optimal multi-metal synergy.Tafel plots and electrochemical impedance spectra (EIS) were studied to survey the kinetic information during OER operation. As shown in Fig. 3C–D, NiCuCoII–FeIIICoIII displays a smaller Tafel slope of 97 ​mV dec−1 than the counterparts, suggesting the favorable electrochemical process combined by the pre-oxidation process and OER catalysis. The EIS data further prove the facile electrochemical behavior under OER operation (Fig. 4 ). As listed in Fig. 4C, the samples with Co-containing cations exhibit much smaller charge transfer resistance (Rct), and the introduction of other metallic elements could further reduce the Rct. For instance, NiCuCoII–FeIIICoIII exhibits the smallest Rct of 73.6 ​Ω among the tested catalysts, demonstrating its favorable reaction kinetics towards OER catalysis.The electrochemical surface area (ECSA) is another key parameter to determine the catalytic activity owing to the different contribution in active site density. The electrochemical double-layer capacitances (Cdl) were calculated to evaluate the ECSA and the influence on the intrinsic activity. As shown in Fig. 5A-B and S15, NiCuCoII–FeIIICoIII exhibits a large Cdl value of 1.49 ​mF ​cm−2, which is 1.02–4.26 times larger than the counterparts. However, the larger Cdl value (equivalently, the larger ECSA) of NiCuCoII–FeIIICoIII may not the only reason for the enhanced OER activity. To better evaluate the intrinsic activity of the PBAs with elemental modulation, LSV curves normalized by Cdl values were plotted [56]. As shown in Fig. 5 C–D, NiCuCoII–FeIIICoIII delivers the highest jCdl of 53.0 ​A ​F−1 at η ​= ​400 ​mV, which is 1.5–16.3 times larger than the cation-modulated PBAs and 1.4–2.0 times larger than the anion-modulated counterparts, thereby confirming its high intrinsic OER activity. The enhanced intrinsic activity may originate from the multi-metal synergy in both cationic and anionic sites, which not only boosts the pre-oxidation process for facile generation of the active species but also promotes the OER kinetics [41]. The enhanced intrinsic activity is responsible for the high OER activity of NiCuCoII–FeIIICoIII, further verifying the effectiveness of the elemental modulation in both cationic and anionic sites of the PBA pre-catalyst.Previous studies indicate that the transition metal-based pre-catalysts usually undergo obvious performance activation during the long-term OER operation owing to the in-situ formation and accumulation of the catalytically active high-valence species [57,58]. To elucidate the pre-oxidation process and evaluate the operational stability of the NiCuCoII–FeIIICoIII pre-catalyst, I-t stability test and related post-catalytic characterizations were conducted. As shown in Fig. 6 A, NiCuCoII–FeIIICoIII exhibits a remarkable activation with the mass activity increasing from 95.2 to 225.6 ​A ​g−1 during the first 2 ​h at η ​= ​400 ​mV, and a dramatic 140% enhancement in OER activity can be identified. After the activation, the OER activity of NiCuCoII–FeIIICoIII undergoes slight decrement, while a high mass activity of 172.4 ​A ​g−1 can still be achieved even after continuous catalysis for 72 ​h. The final activity shows 1.81 times enhancement than that of the fresh NiCuCoII–FeIIICoIII pre-catalyst, further demonstrating the crucial role of the pre-oxidation process in promoting the OER behavior. Of note, the excellent operational stability of NiCuCoII–FeIIICoIII outperforms most previously reported PBA-based/-derived catalysts (Table 2), making it a promising candidate for efficient and stable water electrolysis. The comparative study of the LSV curves for the fresh and activated NiCuCoⅡ-FeⅢCoⅢ further verifies the key role of the pre-oxidation process in OER catalysis. As shown in Fig. 6B and Table 2, the η10 of NiCuCoⅡ-FeⅢCoⅢ reduces from 288 ​mV to 251 ​mV after the activation, and the jgeo increases from 195.8 to 246.8 ​mA ​cm−2, showing a 26% enhancement in OER activity. Hence, the superior OER stability with substantially increased activity was confirmed for NiCuCoⅡ-FeⅢCoⅢ.The increased OER activity during the long-term catalysis could be attributed to the facile pre-oxidation process with substantial accumulation of the active species [21–23]. As indicated in the inset of Fig. 6B, the TEM image of NiCuCoⅡ-FeⅢCoⅢ after stability test (namely, NiCuCoⅡ-FeⅢCoⅢ-pc) reveals the obvious morphology change, where hollow nanocage built by ultrathin nanosheets can be observed, indicating the complete reconstruction of the PBA pre-catalyst. The evaluation of Cdl values before and after stability test also confirms the enlarged surface area. As shown in Fig. S16, the Cdl value of NiCuCoⅡ-FeⅢCoⅢ increases from 1.49 ​mF ​cm−2 to 3.44 ​mF ​cm−2 after the long-term OER operation, confirming the obviously enlarged surface area caused by the morphology changes and phase conversion owing to the pre-oxidation process. The large surface area could bring in more surface sites for OER catalysis and favor the mass transport during catalysis, which is advantageous to the improved OER performance [29]. Besides, the reaction kinetics is also improved after the activation. As shown in the EIS data in Fig. S17, the charge-transfer resistance reduces from 73.6 ​Ω for the fresh NiCuCoII–FeIIICoIII to a remarkably low value of 7.7 ​Ω for NiCuCoII–FeIIICoIII-pc, confirming the boosted reaction kinetic of the OER catalysis along with the proceeding of the pre-oxidation process.The pre-oxidation process causes the structural reconstruction of the PBA-based pre-catalyst. As shown in Fig. 6C, the post-catalytic FT-IR spectra reveal the vanished characteristic peaks of PBA at 460, 2091 and 2181 ​cm−1 and emerging signals of hydroxides at 629 and 981 ​cm−1 during the stability test, suggesting the phase conversion from PBA to metal hydroxides [40,59,60]. The emerging hydroxide phase can be further verified by means of XRD, HRTEM and SAED analyses (Fig. 6D and S18). As shown in Fig. 6D, (012) facets of metal hydroxide with interplanar spacing of 0.27 ​nm and typical six-fold symmetry can be identified for the ultrathin nanosheets in NiCuCoⅡ-FeⅢCoⅢ-pc. Of note, the nanosheet is of low crystallinity, suggesting the presence of amorphous component induced by the multi-metal feature that limits the crystal growth during the pre-oxidation process. The low crystallinity could bring in more undercoordinated metal ions as the active sites for OER [35]. Besides, ICP-OES and XPS analyses were conducted to survey the compositional change during the stability test. As shown in Fig. S19, the ICP-OES data reveal that only slight change in the atomic ratio of the metallic elements can be identified, suggesting the negligible elemental leaching during long-term catalysis. Besides, the post-catalytic XPS spectra prove the presence of all metallic elements, and the signal of CN bonding declines severely with complete conversion to oxidized N and C species (Fig. S20). As shown in Fig. 6E, the N/M ratio (M ​= ​Ni, Cu, Co and Fe) decreases from 3.64 to 0.77 after the activation, while the O/M ratio oppositely increases from 0.29 to 3.03, which are consistent with the phase conversion from PBA to metal hydroxides during the pre-oxidation [29]. Besides, as revealed from Fig. 6F, the contents of high-valence Ni, Cu and Fe exhibit significant increment after activation, while the Co3+/Co2+ ratio remains nearly unchanged. That is, the in-situ formed Ni3+, Cu2+ and Fe3+ are responsible for the increased OER activity during the long-term operation, while the abundant local Co3+ sites are the dominating factor for the high catalytic activity of the fresh pre-catalyst (Fig. 7 ). Hence, the facile pre-oxidation process of NiCuCoⅡ-FeⅢCoⅢ with morphological, structural and compositional reconstruction was confirmed, which endows the elemental-modulated PBA pre-catalyst with substantially promoted OER behavior during the long-term operation.In summary, synergistic elemental modulation on the cationic and anionic sites of the multi-metal PBA pre-catalysts was achieved for promoting the OER performance. Detailed investigations indicate that Co-containing multi-metallic cations and mixed FeIIICoIII cyanide anions are beneficial to OER catalysis. Thanks to the high intrinsic activity of the local Co3+ sites as well as the optimal multi-metal synergy, the optimal NiCuCoII–FeIIICoIII pre-catalyst displays a low η10 of 288 ​mV, and a remarkable 1.4–61.2 times enhancement in OER activity can be identified compared with the counterparts with variable cations. In addition, benefitted from the facile pre-oxidation process of NiCuCoII–FeIIICoIII, abundant high-valence Ni, Cu and Fe ions with high catalytic activity can be in-situ generated and accumulated, resulting in ultrastable OER performance with substantially enhanced activity. After the pre-oxidation-induced activation, an obviously reduced η10 of 251 ​mV and a 1.81 times enhancement in OER activity can be achieved, making the elemental-modulated PBA pre-catalyst a promising candidate for water electrolysis.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the National Natural Science Foundation of China (22171167 and 21927811).The following is the Supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.pnsc.2022.12.001.

Designing advanced electrocatalysts for the oxygen evolution reaction (OER) is of great significance owing to its crucial role in facilitating the production of clean hydrogen energy via water splitting. To date, it has been widely accepted that a pre-oxidation process with the in-situ generation of the catalytically active high-valence metal sites is essential for promoting the OER behavior of most transition-metal-based OER catalysts, or more felicitously speaking, pre-catalysts. Hence, exploring such pre-catalysts with high pre-oxidation reactivity is of high promise. Herein, we proposed the dual elemental modulation in the cationic and anionic sites of the multi-metal Prussian blue analogue (PBA) pre-catalysts, resulting in promoted OER behavior benefitted from the efficient pre-oxidation ability as well as the multi-metal synergy. Detailed investigations indicate that the Co-containing multi-metallic cations and mixed FeIIICoIII cyanide anions in NiCuCoII–FeIIICoIII PBA (denoted as NiCuCoII–FeIIICoIII) are beneficial to OER catalysis owing to the high intrinsic activity guaranteed by the local Co3+ active sites as well as the optimal multi-metal synergy. After the facile pre-oxidation process, additional high-valence Ni, Cu and Fe ions can be in-situ formed and serve as the active sites, thereby resulting in significantly improved OER behavior. For example, the OER current density of NiCuCoII–FeIIICoIII exhibits 1.81 times enhancement even after 72 ​h continuous OER catalysis, and the required overpotential for 10 ​mA ​cm−2 reduces from 288 ​mV for the fresh pre-catalyst to a remarkable record of 251 ​mV after the pre-oxidation-induced activation, making the optimal PBA-based catalyst a promising candidate for efficient and durable water electrolysis.

The discovery of carbon-based nanostructures, such as carbon fibers, carbon nanotubes (CNTs), carbon microtubes (CMTs), and carbon nanodots, appears to be beneficial for practical applications. In term of CNTs and CMTs, the structures have similarities particularly in hollow, tubular, and single- or multi- walled which only differ in the size, and since their discoveries (Ijima, 1991). Ever since the discovery, another attempt to produce similar carbon-based structures, such as CMTs is carried out. Both CNTs and CMTs are fabricated by applying certain temperature of treatments which most of them were in high temperatures with high pressure condition. Therefore, the synthesis of CMTs is influenced by three factors i.e., temperature, and to reduce the high temperature, the latter factors are catalysts and the carbon sources. The precursors for synthesizing have become a challenge, which commonly is organic compounds such as methane, methanol and acetone(Janas, 2020). Materials consisted of carbons were the most suitable precursors for synthesizing CNTs and CMTs, including hydrocarbon compounds such as methane(Kang et al., 2008) and ethanol(Kakehi et al., 2008). The drawbacks in using these two types of precursors require high temperatures of treatments even though its gaseous form reduces time reaction. Nonetheless, gaseous phases are not the only substances that could be used as precursors as a study had successfully synthesized CNTs from solid materials(Singh et al., 2002). On the other hand, the needs of carbon microtubes (CMT) for several applications are in needs such as for oil adsorption(Zhao et al., 2019), and batteries and energy storage(Huang et al., 2012);(Salahdin et al., 2022), which are synthesized from solid compounds. Hence, finding strategies via availability of precursors and selective catalyst is promising advantages for future use.In term of availability, cellulose provide advantages due to their availability of abundant material in carbon content. This organic material is commonly found in plants, for instance, angel’s trumpet plant contains high variation of acidic compounds which also can be precursors for synthesizing carbon-based materials (Mokbli, 2021). Furthermore, In Indonesia, cellulose is organic waste that is annually produced by the plantations, including empty bunch of palm oil, and even more a study utilized part of palm oil tree which is the kernel shell as precursor for bifunctional catalyst (Abdullah et al., 2020). Based on its structure, cellulose has been used as biofuel (Panneerselvam et al., 2016), suggesting its high carbon content. Other reports have also successfully carbon nano- and micro-structures with mesoporous morphological features from paddy rice(Hao et al., 2019), poplar-catkin(Huang et al., 2021), and plant tissues(Zhao et al., 2019). Another potential plants that can be a precursor is kapok randu, which is a native plant to Indonesia. Cellulose fibers from kapok randu were reported to be used in the synthesis of activated carbon(Chung et al., 2013); subsequently; cellulose-fiber from this plan can be used as the precursors in synthesizing CMTs.Attempts in synthesizing carbon-based microstructures seem to be high-cost due to the use of catalysts in order to overcome the non-steady phase of carbonization. Several reports have suggested the introduction of catalysts, such as silver nanoparticles (Gea et al., 2022) which is considered as high-cost. Whilst, the use of mono-(Co) combined with bimetallic (Fe-Co) catalysts during CVD synthesis is considered as complex strategies which also takes place in high energy(Balogh et al., 2008);(Kakehi et al., 2008). Although several reports have used transitional-metals (TiC, NiCl2, SnO2) and non-metal catalysts (Sulfur) with more controllable reaction, these show limited selection in finding affordable and available catalysts for synthesizing with low-cost aspects(Huang et al., 2021);(Ariyanto et al., Feb. 2019);(Anil Kumar et al., 2022). Hence, the purpose of this present work was to evaluate the possibility of cellulose fibers isolated from kapok randu to be used as the precursor with catalysts such as Fe, Ni, and Cu via heating treatments were.The Kapok Fibers (KFs) were collected from the fruit, in which the tree was located in the sub-district Tanjung Mulia, Deli Serdang Regency, Medan, Indonesia. The chemical reagents such as HCl, NaOH, H2O2, H2SO4, H3PO4, NaOCl, Na2S2O3, NaNO3 and acetone were purchased from Sigma Aldrich Inc. Meanwhile, the metal catalysts, such as FeSO4·7H2O, NiSO4·H2O, CuSO4·5H2O were supplied by Bratachem. The gasses utilized to control the condition including nitrogen, methane, hydrogen and helium, were purchased from supplied by PT Aneka Gas.The amount KF that was from the fruit was separated manually via man-labor. Then, these amounts of fibers were dried directly under daylight, and from these amounts, 75 g of KFs were cut into 2–3 cm of sizes. The fibers were immersed in 1 L of 3.5 % HNO3 and 10 mg NaNO3 mixture for 2 h at 90 °C. Then, the mixture was filtered and the fibers were washed by using distilled water until the neutral pH was achieved. Next, the fibers were soaked into 750 ml of 2 % w/v NaOH and 2 % Na2S2O3 solution for 1 h at 50 °C, and followed by washing and filtrating processes. Afterward, these KF samples were bleached with 250 ml of 1.75 % of NaOCl solution until its temperature boiled for 30 min. After being bleached, the samples were mixed with 500 ml of 17.5 % NaOH for 30 min at 50 °C to obtain cellulosic sample. Then, this cellulosic sample was washed by using distilled water, in which hereafter, 10 % of H2O2 were used to immerse the samples for 30 min at 60 °C. The cellulosic sample then was filtered and dried in an oven at 60 °C to remove the residual water content, which was followed by storing it in a desiccator.The first step in isolating nanofiber cellulose (NFC) from KFs began by acid hydrolysis treatment, which was previously described in several studies(Gea, Panindia, et al., 2018; Gea, Zulfahmi, et al., 2018; (Zulham Efendi Sinaga et al., 2018). The cellulosic fibers were soaked in 45 % H2SO4 with ratio of 1:25 w/v% for 45 min at 45 °C. Then, into the mixture, some bi-distillation water with ratio 1:25 v/v% were added and the mixture was allowed to stand for 12 h at room temperature to form suspension. The suspension was separated and washed to reach pH 7. Next, this mixture sample was placed in an ultrasonic bath for 3 h, and then followed by homogenization step for 3 h at 8000 rpm. Finally, these nano-cellulosic kapok fibers (NCKFs) were heated at 50 °C in an oven to remove water content, and the dried samples of NCKFs were obtained which hereafter is stored in desiccator.The characterization of TGA was performed to determine as basis of later heating treatments for the growth of CMT. The TGA was carried out via TGA DTG-60 in which the thermal rate was 10 °C.min−1 in nitrogen condition (flow rate 30 ml.min−1). The initial temperature was 27 °C, and final temperature was 600 °C, whereas the starting mass of the NCKFs was 3 mg.Thermal Gravimetric Analysis (TGA) was used to obtain the decomposition temperature. The decomposition process started at 320 °C based on the TGA result, so that the NCKFs were heated inside a furnace at 400 °C for two hours to produce carbon. Afterward, the carbon samples were obtained, and the chemical activation were performed by immersing these carbon samples into 1 M H3PO4 for 90 min with ratio 1:10 w/v%. Then, this mixture was filtered, and the filtered carbon samples were dried at 150 °C for 24 h. Afterwards, 5 N HCl was added to the dried carbon samples to perform reactivation. Thus, the removal of excessive chloride ions was done by washing these dried carbon samples with distilled water to reach pH 7, and followed by additional washing with cool distilled water as well as filtering to remove the residual of phosphate anion. After being washed and filtered, the wet carbon samples were dried at 150 °C for 24, and finally this sample was named after AC.The preparation of 0.09 M Cu(NO3)2, 0.09 M Ni(NO3)2, and 0.09 M FeCl3 catalyst solutions was carried out by respectively dissolving 1.08 g of Cu(NO3)2·3H2O, 1.3 g of Ni(NO3)2·6H2O, and 1.21 g of FeCl3·6H2O in acetone. Then, each solution was homogenized with constant stirring.The preparation of catalysts for CMTs was performed by mixing 0.09 M Cu(NO3)2, 0.09 M Ni(NO3)2, and 0.09 M FeCl3 with AC and ratio of 1:10 w/v%. Then, each of the mixture underwent ultrasonication for 2 h at 70 °C. The results of the impregnation samples were dried in oven for 12 h at 70 °C.An amount of impregnated AC samples was placed in a 25 ml porcelain dish into a gas furnace. On the surface of the dish, the end of furnace ceramic pipe was connected to gas source. During the heating process, the dish was covered to prevent small carbon particles from escaping. The first step was calcination process, which was done by streaming the impregnated AC with heat at 500 °C for two hours under inert conditions (nitrogen gas 100 ml.min−1). The second stage was a reduction process, where the temperature of 700 °C and hydrogen gas (with flow of 60 ml.min−1) were applied for two hours. In this process, metal oxides would be removed and the metals were converted into metal nanoparticles. In the third stage, the temperature in the reactor was increased to 950 °C, and followed by the increase of nitrogen gas rate to 100 ml.min−1. When the reactor reached the set temperature, the nitrogen gas rate was increased to 200 ml.min−1. The next stage was the flowing of mixture methane and nitrogen gas which respectively 1:2 ratio (rate of 100 ml.min−1) for two hours at 950 °C. Then, at the final process, Helium gas was flowed at 60 ml.min−1 into the furnace, thus; the temperature within it would drop to room temperature. Inert condition during the final process was important in order to prevent the destruction of CMTs.In this research, CMTs were analyzed by Transmission Electron Microscopy (TEM) instrument (JEM-1400) with acceleration voltage 120 kV. The photograph obtained from TEM was analyzed to observe the length and diameter of the tubes via Image-J application, as well as the structures of CMTs, which were different, depends on the catalyst used.The NCKF were analyzed by Fourier Transform Infrared (FTIR) instrument. Characterization by FTIR was done to confirm the functional groups in NCKF in Fig. 1 .The FTIR pattern shows that the functional groups of the NCKF were the same as cellulose fibers. The –OH functional group was shown at peak 3418 cm−1 with stretching vibration up to 2900 cm−1, whereas, the CH aliphatic group was at 2900 cm−1. It could also be clearly seen that the OH group related to carboxylate group was at 1635 cm−1. The bending vibrations of HCH, OCH, and CH, and rocking vibration of –CH2 were respectively at 1427 cm−1, 1373 cm−1, and 1334 cm−1, which these three vibrations were in C6 glucose chain. The Fig. 2 shows the morphological of NCKF. Fig. 2 shows morphological images by SEM of NCKF obtained with 250 and 100 times of magnification. Unlike those obtained by previous studies, the NCKFs had different shapes from palm oil bunches and corncobs (Zulham Efendi Sinaga et al., 2018). Although generally, kapok randu is one of tropical tress with fruits containing cellulosic fibers, its cellulosic material has been reported to be hydrophobic-oleophilic (Wang et al., 2018). Hence, it is assumed that the thermal properties of NCKF were different due to its utilization as a precursor in the synthesis of CMTs.As the NCKFs was used as the precursor to synthesize CMTs, TGA analysis was performed to determine the temperature breakdown. The Fig. 3 depicts the comparison of TGA analysis of NCKFs.According to Fig. 3, the initial temperature started from 27 °C to 600 °C, and the initial mass of NCKFs was 3 mg with 10 °C.min−1 of heating rate. The significant observation can be seen at the temperature above 300 °C as the NCKFs started to decompose and reached 60 % of mass change. Although the NCKFs differs to other biomass, this thermal analysis has confirmed similar data to what have been reported in several studies (Soykeabkaew et al., 2012). The region of initial and ending composition were above 300 °C and 600 °C respectively (Gea et al., 2020a).As it has been reported by several studies, around 32–47 % cellulose was isolated from various raw materials (Gea, Andita, et al., 2018; Gea et al., 2020a; (Marpongahtun et al., 2018), where its physical and chemical structures may have been different from one to another due to different alkaline treatments. The use of sodium hydroxide in cellulose isolation has been concluded to provide distinguishing impacts on the morphological structures, including the stiffness and orientation of the fibrils (Chakraborty et al., 2011).In this study, the AC was obtained from NCKFs, and in Figs. 2 and 3, the –OH, CH aliphatic, and –CH2 were confirmed. However, the spectrum of AC in Fig. 4 showed different results, particularly in the presence of new groups and the occurrence of reduction. The groups, such as CC, CO, and P = OOH was confirmed due to the treatments with H3PO4. This could occur as the surface of biomass was carbon-derived(Oginni et al., 2019), which were indicated the presence of CC stretching band around 1600 cm−1, –CH stretching band in interval of 2800–3000 cm−1, and -P = OO4 in 1100 cm-1(Xu et al., 2014).Subsequently, the confirmation of activated carbon was also performed to ensure the success synthesis. The following Fig. 5 demonstrates the XRD result of sample activated carbon. In overall, two broad peaks were noticeable observed in 40-50° and 60-70° which is related to (002) (Xu et al., 2014). It also can be observed that amorphous parts of activated carbons alongside, in which it could have attributed to the random stacking of layers, that were also noted from the SEM photographic images in Fig. 5 (inset). Fig. 6 shows significant differences of TEM photographic images in each sample synthesized from different catalysts. The commercial AC samples were with diameter of 45–50 nm (Fig. 6a). The image also showed metal nanoparticles attached to the cap of the CMTs with a mean tube length of 600 nm. Meanwhile, the sample with AC synthesized from NCKF with Cu catalyst for 11 h had tube diameter and length of 50 nm and 100 nm respectively (Fig. 6c). In Fig. 6c*, with samples made of Cu catalyst, produced mostly spherical amorphous carbon particles with sizes of under 10 nm. Meanwhile, the AC from NCKF with Ni catalyst had tube diameter and length about 40–50 nm (Fig. 6d).The first variation was sample of AC from NCKF with Fe catalyst treated for 6 h to produce CMTs with diameters of 200 nm with the average tube length of 2–3 µm (Fig. 6b). Then, the second variation was commercial AC with Ni catalyst treated for 11 h, produced CMTs with diameters of 50 nm and a tube length of 150 nm (Fig. 6e). This happened due to the solubility of carbon in metal particles, which would form solid filaments alongside with the width of the diameters(Duc Vu Quyen et al., 2019). Another report has shown that the decreasing of diameter of CMTs from kapok randu for almost a half (from 20 µm to ∼ 12 µm) after carbonization at the temperature of 500–1000 °C(Zhao et al., 2019). Fig. 6a and 6b display tubular structure even though rough shapes were found. Therefore, high temperature used could reduce catalyst activities to bind the carbons in CMTs arrangement during the reaction, such as direct calcination at 500 °C(Wang et al., 2018). It is assumed that due to the high temperatures that leads to high pressure condition, carbon atoms begin to degrade and create uncontrollable reaction, which allows the formation of carbon clusters. At the same time, the surface of AC could react to which form graphitization too, and due to the use of the metal catalysts which implied to the uncontrollable graphitization reaction(Ariyanto et al., Feb. 2019). The reaction caused the particles to agglomerate to each other, so that the reaction results tended to lead the formation of amorphous carbon particles with a size below 10 nm(Ahmad et al., 2018) as it is shown in Fig. 6.In general, the synthesis reaction of CNTs via CVD was carried out over a time span of 30–60 min for the growth of CNTs by using precursors(Costa et al., 2008);(Ramírez Rodríguez et al., 2018). Several studies also mentioned that the process could take longer time(Li et al., 2009);(Fathy, 2017). In this study, the results obtained based on TEM analysis showed a small amount of pile up (especially based on Fig. 6a and b), which was assumed to occur due to thermal treatments. However, the tubes produced in this study were seen to be consistent compared to previous reports(Duc Vu Quyen et al., 2019);(Ahmad et al., 2019). The optimum growth of CMTs was 30–60 min, more than 60 min would not produce more CMTs as the surfaces of catalyst would have been overgrown with CMTs. In conclusion, the use of a longer time with the same temperature would reduce the yield of CMTs as previously reported (Ahmad et al., 2019).The above Fig. 7 is the Raman shifts of pyrolysis sample of AC from NCKF with different catalysts, i.e., the impregnation of Ni, Fe, and Cu. The shifting could be clearly seen in the interval of 1500–1700 cm−1, indicating the first order of G band as well as implying the presence of CC structures. Among three of them, sample 1 and 2 had the higher intensity, and also, in Fig. 7B, the sample 2 which was treated with Cu as the catalyst, had higher intensity counts than that in sample 1 that was treated by Fe as the catalyst. The presence of this peak has confirmed the successful growth of single-wall carbon nanotube (SWCNT); as per what other studies have also reported the shifting 1500–1700 cm−1 both in red and green laser (respectively 785 and 514 nm)(Costa et al., 2008);(Li et al., 2009).Metal catalysts, mono-metals in particular, could affect the structural rearrangement structure of the carbonaceous materials(Ahmad et al., 2019). As this study only focused in the use of mono-metals, the selection of metals was based on the precursor, which is the cellulosic material(Duc Vu Quyen et al., 2019);(Ahmad et al., 2019). Both TEM and Raman results confirmed the differences in each growth samples in terms of the sample intensities as well as the length of the tubes. Copper (Cu) catalysts had amorphous structures, which were both proven by sample 3 in Raman shift results (Fig. 7a). These results were in accordance to the studies which utilized Fe and Ni catalysts, producing long-shape tube growths via CVD method(Arjmand et al., 2016). Meanwhile, another research reported Ni catalyst has successfully synthesized long-shape tubes with 10–40 nm diameters via microwave treatments(Burakova et al., 2018). Although this study investigated that sample Cu and Ni which are based on Fig. 6a, 6b, and 6d, the growth of the tubes via AC from NCKF may occur due the surface area. Thus, it is suggested to evaluate further about the relation of surface area with the growth of carbon microstructure, nevertheless; have been successfully obtained.The fibrous material from kapok randu that contains cellulose has the potential to be precursors for CMTs synthesis from activated carbon as the basis of growth. By using various catalysts, such as iron, nickel and copper, the growth of CMTs has been successfully obtained via thermal treatments with the introduction metal catalysts of Fe, Ni, and Cu at moderate temperature (700–950 °C) even though longer graphitization takes time. Based on the results, the size diameter of the CMTs were in between 50 and 200 nm with length diameter of less than 2 µm in average, as the growth is observed in the micrograph results. The highest presence of CMTs’ growth was observed in sample with nickel (Ni) catalyst, and the Raman shifts appears to show the growth with considerably peaks. Thus, further investigation related to the physical parameters of catalysts and graphitization reaction in moderate temperature is required.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to thank the rector of Universitas Sumatera Utara as its grants via TALENTA scheme program with given contract No. 2590/UN.5.1.R/PPM/2018.Supplementary data to this article can be found online at https://doi.org/10.1016/j.jksus.2022.102423.The following are the Supplementary data to this article: Supplementary data 1

Objective As a hydrocarbon material, cellulose could be used as precursor in synthesizing carbon micro-structure (CMTs), and this study aims to investigate the potential use of cellulosic material from kapok randu as a precursor in synthesizing carbon-based structures. Methods The isolation of cellulose was carried out via alkaline treatment, followed by mechanical disintegration. Meanwhile, the growth of CMTs was performed via heating treatment for 12 h with various catalysts (i.e., Fe, Ni, and Cu). Chemical characteristics were confirmed by FTIR, XRD spectra, while TEM, SEM and Raman spectra were performed to determine the growth of CMTs. Results The thermal characteristic suggested that the decomposition was initiated at 300 °C. The FTIR results confirmed the presence of functional groups in accordance to cellulose fiber, such as –OH (3418 cm−1), CH aliphatic group (2900 cm−1), OH (1635 cm−1), and CH (1334 cm−1). Whilst, the FTIR pattern also confirmed the presence of CC stretching at 1600 cm−1, –CH in between 2800 and 3000 cm−1, indicating the activated carbon. Raman shift indicated the growth on G-band in the interval of 1500–1700 cm−1, suggesting the presence of CC structures. Based on the morphological characteristics, the growth of CMTs were successfully obtained with different diameters and lengths due to the different catalysts used. The iron (Fe) catalysts produced CMTs with a diameter about 200 nm and average length of 2–3 µm, whereas the Ni catalysts formed tubes with around 150 nm of length and 50 nm of diameter. Meanwhile, the Cu catalysts formed amorphous particles with diameter below 10 nm. Conclusion From these results, the evaluation of cellulose isolated from kapok randu as a precursor in the growth of carbon micro-tube with distinguished characteristics was demonstrated.

Under the trend of dramatic population growth [1], rapid urbanization [2], as well as expansion of industrial and agricultural production scale [3], the issues of water pollution and scarcity are stern day after day and have become critical environmental issues facing the worldwide in the 21st century. Moreover, subject to the combined influence of multiple factors under the 2019 coronavirus disease (COVID-19) pandemic such as extreme weather [4], monetary policies [5], and supply-demand imbalance [6], the prices of fossil fuels (oil, coal, and natural gas) that underpin current world economic development and social progress have increased dramatically, which heralds global energy crisis is becoming even more intense and facing more complexity and uncertainty. Traditional water treatment (biological, physical, and chemical) technologies [7], have significant energy consumption problems in the treatment process and are unable to efficiently and cost-effectively treat wastewater containing large amounts of highly concentrated organic pollutants [8], which poses a serious burden to social and economic development. How to treat wastewater in an efficient and green way with both social benefits and economic value has become a hot topic for researchers, and on this basis has stimulated the development of a series of green and sustainable energy technologies such as microbial fuel cell (MFC).MFC is an environmentally friendly and efficient device for wastewater treatment and energy recovery [9], which can degrade organic pollutants while generating electricity, in line with the current concept of sustainable development. Specifically, MFC, as an important branch of fuel cells, has outstanding advantages in operation and function that are incomparable with other energy sources: wide sources of raw materials [10], all biodegradable organic matter can be used as MFC substrates in theory; no energy input is required [11], the air-cathode MFC is in direct contact with oxygen and does not require energy input for aeration; relatively higher energy conversion rates [12], MFC can convert the chemical energy contained in the organic matter directly into electrical energy; mild operating conditions [13], the microbial diversity in the MFC enables it to work under normal temperature and pressure; clean and environmentally friendly [14], the exhaust gas generated by MFC is mainly carbon dioxide (CO2), which has small emission and will not cause secondary pollution to the environment and does not require exhaust gas treatment.Compared with other fuel cells, MFC has broad application prospects due to the above characteristics and advantages, which contribute to the fields of environment, green renewable energy, and biomedicine. The expanded applications of MFC in emerging fields such as desalination [15], ecological restoration [16], alternative power sources [17], artificial organ power sources [18], biosensors [19], environmental monitoring [20] and coupling application [21] have emerged and achieved energy diversification. It well illustrates that the application potential of MFC as an emerging environment-friendly technology is huge. At present, the research on MFC is still mainly focused on wastewater resource treatment. Despite the dual advantage of wastewater treatment with simultaneous electrical energy production by MFC technology, the current relatively low output power has not reached the ideal state. Current stage research suggests that the wastewater treatment effectiveness and power production of MFC are influenced by various factors, mainly including MFC configurations [22], anode materials [23], cathode materials [24], inoculated microbiological [25], and proton exchange membranes (PEM) [26], etc. Among the many factors affecting MFC performance, the expensive cost [27], biofouling [28], and slow oxygen reduction reaction (ORR) kinetics [29] of the cathode have become the key factors preventing the effective operation and expected practical application of MFC. Therefore, the cathode can be considered as a pointcut to improve the overall performance of MFC.In the early stages of MFC research, the Pt/C catalyst with relatively high ORR electrocatalytic activity was generally used as a traditional cathode catalyst for MFC. The follow-up studies have shown that the shortcomings of the Pt in terms of high price [30], scarce resources [31], and easy deactivation by poisoning [32] have greatly limited the large-scale commercial application of MFC. The development of cathode catalysts with low cost, high efficiency, and good tolerance has become a crucial prerequisite for the industrialization of MFC. After extensive and in-depth research by many scholars, a series of optimized single-metal cathode catalysts with favorable ORR catalytic activity have been successfully prepared, such as transition metal macrocyclic compounds [33], transition metal oxides [34], metal sulfides [35], transition metal-nitrogen-carbon (M-N-C) catalysts [36] and metal-organic framework (MOF) catalysts [37], etc., which can be used as efficient and low-cost cathode catalysts to replace Pt/C catalysts. In recent years, bimetallic catalysts with two different metal elements or compounds as active ingredients have shown superior catalytic performance and durability than monometallic catalysts owing to the synergistic effects (geometric [38], electronic [39], and stabilizing effect [40]) between the bimetallic components. This is not limited to bimetallic catalysts, trimetallic or multimetallic catalysts also possess more outstanding catalytic properties, such as good active stability and selectivity, than monometallic catalysts, but are limited by the fact that their synthesis may be more complex, wordy, and relatively expensive than bimetallic catalysts [41], researchers have focused their attention more on bimetallic catalysts. Various types of bimetallic catalysts have been successfully applied in MFC as low-cost ORR catalysts with high activity and stability.In the last decade, there have been relatively few review articles on bimetallic ORR catalysts, especially lacking a comprehensive and systematic exposition of bimetallic ORR catalysts suitable for MFC. Based on previous studies, we reviewed from the perspectives of reaction mechanisms, advantages, and typical synthesis methods of bimetallic catalysts. The achievements of Pt-M alloys, transition metal alloys, transition composite metal oxides, transition metal macrocyclic compound-based bimetallic catalysts, MOF-based bimetallic catalysts, and bifunctional catalysts in MFC are also analyzed emphatically, to sort out a relatively clear vein for reference. Last but not least, a tentative suggestion of future research priorities for MFC cathodic bimetallic catalysts. Fig. 1 presents the thinking logic diagram of this review.Oxygen (O2) is considered to be an ideal electron acceptor superior to potassium ferricyanide [K3Fe(CN)6] and Potassium permanganate (KMnO4) due to its widespread presence in the environment and relatively high redox potential [42,43], which is mainly used in air-cathode MFC, dissolved-oxygen cathode MFC, and biocathode MFC. In air-cathode MFC (Fig. 2 a), the electrons and protons generated via the decomposition of organic pollutants in the wastewater by the electro-producing microorganisms of the anode reach the cathode through the external circuit and the PEM, respectively. Then ORR occurs with the incoming O2 under the action of the cathode catalyst to form water and electric currents, which ultimately achieve the degradation and energy conversion of organic pollutants. ORR is a multi-electron reaction with a rather complex reaction process at the electrode surface which involves multi-step elementary reactions (O–O breakage, transfer of protons and electrons, etc.) and numerous short-lived intermediate species (e.g. O, OH, O2 −, HO2 − and H2O2, etc.) [44]. If specific details are not considered, ORR can be simply divided into direct 4e− and indirect 2e− pathways, the possible reaction pathways are shown in Fig. 2b, where k1-k5 are the reaction rate constant. The ORR pathways are also affected by the acidity or alkalinity of the electrolyte [45], Fig. 2c shows the possible reduction pathways and reduction potentials for O2 in acidic and alkaline electrolytes.From the above, it is clear that the indirect 2e− ORR pathway is more complicated than the direct 4e− ORR pathway whether in acidic or basic electrolytes. H2O2 with a strong oxidizing effect may not only re-engage O2 in the reaction through reversible reactions, reducing the reaction efficiency, but may also damage the cathode properties [46] and corrode PEM [47]. Therefore, the ideal ORR pathway should be a 4-electron reaction, but the barrier of the 4e− ORR pathway is higher than that of the 2e− ORR pathway, the thermodynamic analysis revealed that the dissociation energy of the O=O bond dissociation energy in the unstable intermediate (H2O2) (149 kJ mol−1) generated by the 2e− ORR pathway is much lower than that in O2 (490 kJ mol−1) [48]. This makes it easier for actual ORR to follow the reaction path of 2e− or 2e− combined with 4e−. To address this conundrum, the ideal cathode catalyst should be designed to be highly selective for the 4e− ORR pathway to reduce O2 to water in one step at a higher potential, thus increasing the ORR efficiency to obtain high output voltage and energy conversion rates.There is a competitive relationship between the 4e− and 2e− pathways for ORR, and the ORR pathways of different catalysts strongly rely on the adsorption pattern of O2 [49]. According to Yeager et al. [50], the adsorption modes of O2 on metal surfaces are broadly classified into three types, as shown in Fig. 3 a [51]. The “Griftiths” mode: oxygen molecule interacts laterally with the catalytic active center, which is favorable to the O–O breakage to occur direct 4e− ORR pathway, which is usually observed on the surface of the Pt catalyst; the “Bridge” mode: oxygen molecule simultaneously interacts with two catalytically active centers, apparently favoring the 4e− ORR pathway, which may be observed on transition metal alloys (e.g. Pd–Co alloys) (Fig. 3b) [51]; the “Pauling” mode: only one side of the oxygen molecule interacts with the catalytically active center, which is not conducive to the breaking of the O–O bond and generally results in 2e− ORR, with most electrodes performing ORR in this mode. Recent studies have shown that Co8FeS8 microspheres containing Co, Fe, and S atoms may promote O2 dissociation and the efficient side-on (and/or bridge) adsorption pathways, N atoms with strong electronegativity may powerfully attract electrons from S and metals (Co and Fe), thereby enhancing the charge transfer rate and ORR kinetics on Co8FeS8/NSC-900 so that the main 4e− ORR pathway occurs (Fig. 3c) [52]. Overall, the design of an ideal bimetallic cathode catalyst should consider the effect of O2 adsorption so that the ORR is biased towards the 4e− pathway as much as possible.Elucidating ORR mechanisms and rate-determining steps (RDS) at the atomic level remains challenging, which is mainly attributed to the extremely complex process in ORR. Three currently commonly accepted ORR mechanisms were combed by Xia et al. [53], as shown in Fig. 4 a. The “Dissociative” mechanism: after O2 diffuses to the electrode surface to form O2∗ (∗ indicated the adsorbed state), the O–O bond is broken directly to form O∗ intermediates which are reduced successively to OH∗ and to H2O∗; the “Associative” mechanism: the O–O bond is cleaved to form O∗ and OH∗ intermediates after O2∗ forms OOH∗; the “Peroxo” mechanism: the O2∗ is reduced successively to OOH∗ and to HOOH∗ before the O–O bond cleavage. For many reasons, the actual ORR process may be carried out by one or more of these three mechanisms, alone or in combination.The free energy barriers gained via Density functional theory (DFT) calculations can be used to estimate the dominant mechanism in the ORR process. In general, the dissociation mechanism, which has the lowest energy barrier at low oxygen coverage, is the main ORR pathway, while the association mechanism, which presents the lowest potential barrier at high oxygen coverage, dominates the ORR process [54]. Through a new method of adsorption preference and electron affinity, Wu et al. [55] predicted that the dominant ORR mechanism on the Pd3Cu surface in alkaline media may be one association mechanism to achieve ultra-low overpotential. Liu et al. [56] demonstrated that the potential ORR RDS of the CoNi alloy nanoparticles (CoNi3–CoN4) was the protonation reaction of O2 (O2+H++e−→OOH∗) based on DFT calculations, moreover, the energy barrier of RDS at the CoNi3–CoN4 site (0.40 eV) was lower than that on the Co–N4 site (0.70 eV), indicating that CoNi3–CoN4 site exhibits better ORR catalytic activity (Fig. 4b). Recently, Chen et al. [57] proved by in situ Raman spectroscopy that the key intermediate species of disordered and ordered Au–Cu nanocatalysts was ∗OH during ORR, as well as the binding site of Au and Cu was the real active site of the catalyst (Fig. 4c and d), laterally reflecting the superiority of the bimetallic catalyst.Furthermore, renowned scientist Nørskov [54] used DFT to draw a “volcano” diagram of ORR activity trend for different pure metals to describe the relationship between ORR catalytic rate and metal-oxygen adsorption energy. As can be seen from Fig. 5 a, there are significant differences in the ORR activity of different metals, with Pt and Pd at the top having the best binding strength with oxygen, being the best monometallic catalysts in terms of ORR activity, and Pt/C thus became one common commercial catalyst, proving that the binding energy between the oxygen-containing intermediate species and the catalyst surface determines the catalytic rate of ORR. Subsequently, Greeley et al. [58] plotted another “volcano” diagram (Fig. 5b), which showed that Pt-based alloys not only have the same oxygen binding energy trend as pure metals but also Pt-based alloys at the top possess higher ORR activity. This conclusion is just in line with the Sabatier principle, which states that the adsorption strength of the catalyst on the reactants should be in the appropriate range to obtain the optimal catalytic activity. Structural characterization and DFT calculation [59] revealed that the interaction of Pt3Co(111) and Co–N4 active sites with low overall reaction-free energy in PtCo@NGNS reduced the adsorption energy of oxygen-containing intermediates and the activation energy of the reaction, so as to synergically improved the ORR performance through an associated 4-electron mechanism (Fig. 5c and d).Compared with monometallic catalysts, bimetallic catalysts exhibit synergistic effects due to metal-metal bonding interactions (Fig. 6 a), significantly improving the catalytic performance [60]. DFT calculations demonstrate that in catalytic reactions, the synergistic effects between bimetals can greatly decrease the reaction activation energy [61], achieving the remarkable effect of 1 + 1>2. A vivid diagram can be used to interpret the synergistic catalytic effect of bimetallic catalysts, i.e., the catalytic performance of A + B is greater than that of A or B (Fig. 6b). The synergistic effects between bimetallic components can be classified into three types as follows: (1) Geometric (or strain) effect: The addition of second metal makes mismatching of lattice constants (lattice distortion), leading to changes in the atomic spacing or average metal-metal bond lengths, which alters the geometrical configuration of the catalyst while improving the catalytic activity [62,63]. Wu et al. [64] confirmed that the Au–Pt alloy exhibited higher catalytic performance than the monometallic catalyst owing to changes in grain size and lattice structure (lattice shrinkage in Au and lattice expansion in Pt) during the formation of Au–Pt. (2) Electronic (or ligand) effect: The addition of second metal alters the electronic configuration of the active metal site via altering the electronic environment on the metal surface or by promoting electron transfer between metals [65,66]. The d-band center theory, which assumes that the d-orbital center (cannot be too high or too low) of the metal is linearly related to the adsorption strength of the reacting species on the metal surface, is one of the significant descriptors of ORR activity. Yin et al. [67] discovered that the NP-Ag4Cu catalyst made the d-band center of Ag closer to the Fermi energy level due to the alloying effect of Cu, thereby optimizing the electronic structure and significantly improving the ORR catalytic activity (7 times that of NP-Ag). (3) Stabilizing effect: The addition of second metal improves the stability of the catalytically active metal. The Pt6Ru1/C catalyst exhibited good SO2 resistance in ORR due to the addition of Ru that was not poisoned by SO2 [68], maintaining 60% mass activity even after SO2 poisoning, far better than commercial Pt/C catalyst (30%). Geometric (or strain) effect: The addition of second metal makes mismatching of lattice constants (lattice distortion), leading to changes in the atomic spacing or average metal-metal bond lengths, which alters the geometrical configuration of the catalyst while improving the catalytic activity [62,63]. Wu et al. [64] confirmed that the Au–Pt alloy exhibited higher catalytic performance than the monometallic catalyst owing to changes in grain size and lattice structure (lattice shrinkage in Au and lattice expansion in Pt) during the formation of Au–Pt.Electronic (or ligand) effect: The addition of second metal alters the electronic configuration of the active metal site via altering the electronic environment on the metal surface or by promoting electron transfer between metals [65,66]. The d-band center theory, which assumes that the d-orbital center (cannot be too high or too low) of the metal is linearly related to the adsorption strength of the reacting species on the metal surface, is one of the significant descriptors of ORR activity. Yin et al. [67] discovered that the NP-Ag4Cu catalyst made the d-band center of Ag closer to the Fermi energy level due to the alloying effect of Cu, thereby optimizing the electronic structure and significantly improving the ORR catalytic activity (7 times that of NP-Ag).Stabilizing effect: The addition of second metal improves the stability of the catalytically active metal. The Pt6Ru1/C catalyst exhibited good SO2 resistance in ORR due to the addition of Ru that was not poisoned by SO2 [68], maintaining 60% mass activity even after SO2 poisoning, far better than commercial Pt/C catalyst (30%).It is generally accepted that the catalytic activity and selectivity of bimetallic catalysts are mainly influenced by geometrical and electronic effects [69]. That is, the geometric and electronic properties are adjusted through geometric and electronic effects, breaking the activity-selectivity trade-off in the catalytic reaction and avoiding overshadowing efforts to optimize the performance of catalysts. Typically, the geometrical effect of bimetallic catalysts is always accompanied by an electronic effect. In situ X-ray absorption spectroscopy (XAS), including X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS), with strong correlation to DFT predictions, which can detect the electronic and geometric states of atoms in materials during catalytic reactions. Gibbons et al. [38] employed in situ XANES and found that the white line peak (dashed line) of both components was almost identical in intensity and position (Fig. 6c), indicating that the electronic structure of Ag was almost unchanged by the presence of Cu. Fourier transforms (FT) of EXAFS display no prominent movement of the Ag–Ag distance (dashed line) for Ag and CuAg measured at 0.75 V vs RHE (Fig. 6d), showing that the geometric state of Ag atoms in CuAg was not changing dramatically. DFT calculations predicted that the electronic structure of Ag changes only slightly in the presence of Cu (Fig. 6e), while the density of states of the Cu atom in a Ag lattice was dramatically altered compared to pure Cu (Fig. 6f). These results demonstrated that the highly active Cu-centered catalytic sites and the electronic effect rather than geometric effect was the main reason for the ORR activity of the CuAg catalyst exceeding the sum of Cu and Ag. In contrast, Qiu et al. [39] figured out that the main reason for the CuPd/SiO2 catalyst prepared via the incipient-wetness impregnation method to exhibit higher activity than Cu/SiO2 was the high dispersion of Cu and Pd and the geometric effect between the bimetals, while the electronic effect had weaker effect towards catalytic performance. To sum up, it is just the mechanism of the synergistic effects of bimetallic catalysts that allows the two metals doped in a bimetallic catalyst to produce higher performance (catalytic activity, selectivity, and stability) than either of its components would produce if they were present alone.The ideal ORR catalysts are expected to have high ORR catalytic activity. Currently, electrochemical testing is mostly performed on an electrochemical workstation (Pine, USA) with a standard three-electrode system. Among them, the working electrode is always the study electrode, the reference electrode (e.g. SCE, Ag/AgCl, RHE, etc.) is mainly used to determine the potential of the working electrode, and the counter electrode is used to form a series circuit with the working electrode to pass current. The most main ORR electrochemical measurement techniques are cyclic voltammetry (CV) and linear scanning voltammetry (LSV). Apart from these, various common characterization and analytical techniques such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), Brunauer-Emmet-Teller (BET), and electrochemical impedance spectroscopy (EIS), have been employed to observe, understand and analyze the surface morphology, microstructure, crystal structure, elemental, specific surface area, and internal resistance variation including charge transfer resistance (Rct) of bimetallic catalysts, respectively. These analytical technologies will be briefly expounded in the applications section of this review.CV testing of ORR performance is usually performed using the standard static three-electrode system in O2- and N2-saturated 0.1M KOH electrolyte for one cycle scan of the oxidation and reduction processes, which is aimed at avoiding gas over-disturbance of the solution and thus ensuring the stability of the system. By CV analysis (Fig. 7 a), Goswami et al. [70] observed that the presence of oxidation peaks could only be observed clearly for the Pd3Cu/C in O2-saturated 0.1 M KOH solution, and the more positive reduction potential (–0.26 V) than Pt/C indicated a superior ORR activity. LSV testing can further investigate the ORR activity of the catalyst with the same principle as the CV method. It is just missing a back sweep while still needing to provide a high-speed stirring environment to reduce or eliminate the effect of factors such as diffusion layers and accelerate the mass transfer rate of O2 [71]. Normally, LSV testing is carried out in O2-saturated electrolyte (KOH, NaOH, or H2SO4) using a rotating disc electrode (RDE) or a rotating ring disc electrode (RRDE) as the working electrode. LSV curve measured by Kisand et al. [72] displayed that FeCoNC catalyst had onset potential (Eonset) and half-wave potential (E1/2) similar to Pt/C catalyst (–0.048 and –0.18 V), confirming its good ORR activity (Fig. 7b).Tafel slopes obtained by LSV curve fitting can further explore the kinetics of electrocatalytic reactions, it is generally accepted that the smaller the Tafel slope, the better the catalyst has the potential for ORR activity [73]. Anwar et al. [74] revealed that the Tafel slope of the CuPt/NC catalyst (190 mV dec−1) was smaller than that of commercial Pt/C (213 mV dec−1), validating its superior ORR catalytic activity over the Pt/C (Fig. 7c). Due to the limitations of the RDE test in not fully capturing the intermediate products [75], to further explore the ORR mechanism of the catalysts, RRDE testing can be used to more accurately evaluate the electron transfer number (n) and H2O2 yield of catalysts during the entire ORR process. Based on this, You et al. [76] through RRDE testing showed that SrCO3/Fe3C (1:12) had a higher n value (3.94–3.96) and a lower H2O2 yield (less than 3%) compared to Pt/C, indicating a good selectivity for ORR (Fig. 7d).Improving the activity and durability of ORR catalysts simultaneously is a challenging task, as catalysts with higher activity are usually less stable [77]. Numerous studies have shown that ORR catalysts will inevitably deactivate to some extent during long-term use, mainly in the form of a gradual decline in catalyst activity or selectivity over time and a shortened working life. There are many reasons for the deactivation of the catalysts, which can be divided into three categories: “poisoning” deactivation [78], coking and blocking deactivation [79,80], sintering and thermal deactivation [81,82]. In particular, ORR catalysts applied to the single-chamber MFC are also susceptible to microbial contamination due to the lack of isolation by proton exchange membranes [83]. Therefore, the ideal ORR catalysts should not only have excellent ORR activity but also long-term stability and anti-inactivation stability, which means that the development of high-stability performance ORR electrocatalysts is imperative.The long-term stability of ORR catalysts is mostly assessed using the Cycle endurance test (CV and LSV), chronoamperometry (CA), thermogravimetric analysis (TGA), and others. Xiong et al. [84] assessed the stability of the 3D Pd–Cu catalyst and the commercial Pt/C using cyclic endurance experiments, as can be seen in Fig. 8 a, the 3D Pd–Cu catalyst maintained outstanding stability after the 1000-3000th CV cycle, Fig. 8b showed that its E1/2 only decreased by 3 mV and 12 mV after the 3000th and 5000th LSV cycles, whereas the activity and stability of the commercial Pt/C decayed significantly (39 mV decrease in E1/2 after 3000 cycles). CA tests are to observe the decay of catalyst current density at a constant potential, which can be used to detect the stability of ORR catalysts under harmful environmental conditions. Lim et al. [85] conducted methanol crossover tests using CA at a constant voltage of 0.6 V, the results showed that the addition of methanol at 200 s caused a sharp decrease in the current density of the commercial Pt/C catalyst, while the current density of NiCo2O4 did not change significantly and the initial current density decayed more slowly than Pt/C within 10,000 s (Fig. 8c and d), which indicated that NiCo2O4 has better methanol crossover resistance and long-term stability performance. The TGA method, which measures the change of catalyst mass over time under programmed temperature control, is commonly used to characterize the thermal stability of synthetic materials, which will be briefly expounded in the applications section.The economic viability of ORR catalysts is another major challenge. As mentioned earlier, the noble-metal catalysts represented by Pt have long been the core of catalytic research due to their high ORR electrocatalytic performance. However, they are limited by high prices and scarcity of earth resources, in particular, Bernsmeier et al. [86] showed that the cathode materials can account for more than 50% of the total cost of laboratory-scale MFC, making noble-metal catalysts economically inappropriate for large-scale wastewater treatment. The high maintenance costs also limit the large-scale use of MFC [87]. Since the primary role of MFC electricity production is for balancing energy consumption rather than generating additional economic benefits [88], the efficient and rational use of noble-metal materials or the search for non-expensive alternative catalysts has become a top priority in current ORR catalysis research.In the present work, scholars have primarily concentrated on the development of non-precious metal catalysts with new structures. Forming Pt-based alloys via the introduction of a second non-precious metal can improve the d-band center of the catalyst while increasing the cost-effectiveness and activity of the catalyst. Yan et al. [89] revealed that the addition of relatively inexpensive Sn reduced the Pt loading while decreasing the unoccupied d-band of neighboring Pt in the PtSn catalyst, and the Pt50Sn50 catalyst with the highest Sn content results in the highest mass activity. The synthesis of bimetallic catalysts using earth-abundant or cheaper metals is also an effective strategy to improve ORR catalyst activity and decrease costs. A study by Qiao et al. [90] discovered that low-cost Cu–Fe alloy synthesized from two non-precious metals exhibited high activity due to electronic effects that caused a change in the d-band center energy of Cu. Currently, the economic feasibility analysis of most bimetallic ORR catalysts in MFC is based on the cost of production (COP) as the main indicator, including material cost and processing cost (e.g., power consumption). Considering that the price of Mn salts was slightly lower than the same of Fe, the Fe–Mn–N–C catalyst using Mn as a second metal can slightly decrease the total cost, which was about 3.5 $ g−1 considering only consumables [91]. It was estimated that the cost of the Mn/Fe@WRC catalyst doped with Mn and Fe with watermelon rind as raw material is even less (about 0.15 $ g−1), which was much lower than Pt/C (33.0 $ g−1) [92]. In addition to the COP, In order to fabricate large-scale MFCs for practical application, an ideal ORR catalyst should also have cost-effectiveness associated with power generation, which can be interpreted as a maximum power density (MPD) normalized to the cost for comparison. The normalized MPD (mW $−1) is determined by the following equation [93]: N o r m a l i z e d M P D = M P D A O C × C O P Where MPD is the maximum power density (mW m−2), AOC and COP are the amount of catalyst (g m−2) and cost of production ($ g−1), respectively. Such analysis provides useful information for decision-makers to understand which type of catalyst is more cost-effective or more economically feasible to use. The CoNi alloy synthesized by Hou et al. [94] exhibited a superior normalized MPD (150 mW $−1) over the Pt/C catalyst. Similarly, the Cu–Sn-2/AB catalyst prepared by Noori et al. [95] was nearly 11 times more cost-effective (31.0 mW $−1) than the Pt/C catalyst, meeting the sustainability and economic requirements.According to current reports, various synthesis methods for bimetallic catalysts, among which the typical ones mainly include co-reduction, seed-mediated, impregnation, co-precipitation, electrodeposition, microemulsion, and microwave heating method (Fig. 9 a–d). In practice, the appropriate synthesis method should be selected according to the specific situation to achieve the desired synthesis effect. Furthermore, the properties of bimetallic catalysts can also be enhanced via the combination of different synthesis methods.Co-reduction aims to use reducing agents (e.g. NaBH4, N2H4) to simultaneously reduce the two metal salt precursors in the system to zero-valent metal atoms, then make them grow together to form the bimetallic catalyst. The alloys prepared by this method mostly have a relatively high alloying degree, such as Pd–Pb [100], CuFe [101] and AgPd [102], etc. It is generally more difficult to synthesize core-shell structured alloys due to the limitations of the applicable conditions (e.g. inherent properties of the metal), which require the assistance of other synthetic methods. Tsuji et al. [103] successfully gained the Ag@Ni catalyst by using the microemulsion-assisted polyol method to simultaneously reduce the mixture of AgNO3 and NiSO4·6H2O, NiCl2·6H2O or Ni(NO3)2·6H2O for 10 min.Seed-mediated is used to first reduce one metal salt precursor into a metal seed and deposit it as a nucleus, then another metal atom is deposited as a shell by reduction or thermal decomposition to attach to the surface of the formed metal seed, forming a core-shell structure. Bimetallic catalysts prepared using this method generally have larger nanostructure sizes, such as core-shell Au–Pt dendritic nanoparticles [104], Pd@Ir [105], etc. Nevertheless, the presence of strong reducing agents leads to too fast a reduction rate which breaks the conditions close to equilibrium, making it more difficult to synthesize high-quality bimetallic nanoparticles with a single structure. It was found that the independent nucleation of the second reduced metal could be avoided by surface modification [106] and surface replacement reactions(or galvanic replacement reactions) (Fig. 9a) [96].Utilizing metallic salt solutions to impregnate support materials with porous structure, removing the remaining liquid after reaching equilibrium, and then via heat decomposition and activation to make the bimetallic nanoparticles dispersed in the pores on the surface of the carrier. Using an amount of the precursor solution over the pore volume of the support, producing a thin slurry, is called wet impregnation, limiting the amount of precursor solution to just filling the pore volume is termed incipient wetness impregnation [107]. There are also co-impregnation and sequential impregnation methods that are commonly used for the synthesis of loaded bimetallic catalysts. The teams of Nourozi Rad [108] and Naicker [109] respectively synthesized Ni–Cu/TiO2 catalyst and Cu–Ag/Al2O3 catalyst using co-impregnation and sequential impregnation methods. The impregnation method has the advantages of simple operation and low cost, but the lack of induced interaction between the precursor and the support during the catalyst drying process may lead to agglomeration of metal particles and inhomogeneous elemental composition [110], thereby causing low metal utilization.The simultaneous precipitation of cations from two metal salt solutions with the aid of the precipitating agents to acquire the desired bimetallic catalysts. The use of ammonium hydroxide as a precipitant avoids an additional washing step (e.g. removal of residual impurity ions) [111]. Suitable pH and temperature are also key factors in realizing excellent sedimentation results. Yang et al. [112] used a low-temperature co-precipitation method to prepare FeMn catalyst with highly dispersed and selective. Compared with the impregnation method, the co-precipitation method can better control the particle size and distribution of elements in bimetallic catalysts, which is the most commonly used method to prepare composite metal oxides, such as NiCo2O4 (Fig. 9b) [97], CoMoO4 [113], etc. Although this method has the characteristics of a simple process and low calcination temperatures, unnecessary co-precipitation of impurities and analytes can occur at each stage, thereby leading to the formation of aggregates.Bimetallic catalysts prepared by sequential electrodeposition of metal cations in a metal salt solution using an electrode system with an applied constant voltage have good dispersion and chemical stability. The teams of Liu [114] and Vega-Cartagena [115] both used electrodeposition to synthesize a PtPd/rGO catalyst with highly dispersed and an Ag/Pd catalyst with excellent stability in a relatively short time. The catalyst composition and performance can be further regulated by adjusting factors such as potential, temperature, and the type of metal salt precursor. Ajmal et al. [116] synthesized a CuZn alloy with good selectivity by sequential electrodeposition of CuSO4–5H2O and ZnCl2 in the electrolyte at a constant voltage of −0.3 V. Xia et al. [98] synthesized a Fe–Cu alloy by electrodeposition in an electrolyte of FeSO4 and CuSO4 at 25 °C for the 20s (Fig. 9c). Although this is a green, simple and fast method, its preparation cost and conditions are relatively high and harsh.Under the action of surface and auxiliary active agents (e.g. alcohols), immiscible liquids (e.g. oil and water) are first mixed and emulsified to form thermodynamically stable microemulsion systems, then metal salt precursors react in the microbubbles to obtain bimetallic catalysts. Although the preparation is complex and tedious, the particle size of the catalysts can be well adjusted via micelle modification and varying the size of microbubbles, the concentration of reactants, and the pH value in the aqueous phase. Sheoran et al. [117] synthesized a series of spinel catalysts (MFe2O4; M = Co, Ni, Cu, and Zn) with small particle sizes, large specific surface areas, and sufficient magnetism by the microemulsion method. Currently, the water-in-oil (W/O) method is mostly used to prepare uniformly distributed bimetallic catalysts, Szumelda et al. [99] synthesized a series of alloy catalysts such as PdAu, PdPt, PdRu, and PdIr using the W/O microemulsion method (Fig. 9d), where the PdRu and PdIr systems exhibited almost uniform alloy microstructures.Microwave heating is a method that allows metal salt precursors to form bimetallic catalysts under microwave radiation. Compared to conventional heating methods, this method has milder synthesis conditions (uniform heating and fast reaction), which can not only accelerate the synthesis of particles but also obtain more dispersed catalysts. Galhardo et al. [118] successfully synthesized one highly active PtNi catalyst uniformly distributed on a carrier by reacting for 6 min under microwave radiation at 160 °C, reducing the time by 94% compared to the traditional heating methods. In addition, it is a generally feasible method to combine it with other methods, Lingaiah et al. [119] synthesized a series of SiO2-supported Pd–Fe catalysts by combining the impregnation method respectively with the conventional heating method (calcination for 5 h) and microwave synthesis (irradiation at 100% power for 5 min). However, this method cannot effectively regulate the structure of bimetallic nanoparticles.By classifying the currently applicable bimetallic catalysts for MFC, the key contributions of six bimetallic catalysts to MFC (excellent ORR activity, high stability and economic efficiency) and their application towards energy-efficient wastewater treatment in MFC are detailed in sequential order. In general, the values of Eonset, E1/2, n, Rct, BET surface area (SBET), and open-circuit voltage (OCV) are regarded as key indicators to evaluate the ORR performance of bimetallic catalysts suitable for MFC (Table 1 ). In the process of realizing wastewater treatment and energy conversion, MFC mainly uses synthetic wastewater, sludge, high-concentration organic wastewater, and general persistent or stubborn pollutants as substrates. The effectiveness of MFC configured with bimetallic catalysts for wastewater treatment is usually characterized by chemical oxygen demand (COD) removal and pollutant degradation rates. To explore the power generation of MFC under the action of different bimetallic catalysts, the maximum stable output voltage and the MPD are mainly used as evaluation indexes, where the MPD is further divided into area power density and volume power density.Alloying of Pt with relatively inexpensive 3d group transition metals (M = Fe, Co, Ni, etc.) to form Pt-M alloy catalysts can reduce the amount of Pt while significantly enhancing the ORR catalytic activity. This is mainly attributed to the geometrical effect [155], the electronic effect [156], the Raney effect (increasing the effective active area of Pt) [157], and the anchoring effect (allowing Pt to be better embedded in the carrier) [158] of Pt-M alloys, enabling the modulation of Pt catalytic activity.Graphene (G) with high surface area, good electrical conductivity, thermal stability, and durability [159], can be used as excellent support for electrocatalysts. Yan et al. [120] performed XRD analysis of the Pt–Co/G (15 wt% Pt) catalyst and revealed that its reflection peaks were shifted to higher angles than Pt/C, indicating that the addition of the Co causes the lattice contraction of Pt to lower its d-band center relative to the Fermi energy level, which was favorable to the ORR. CV testing demonstrated that the higher Eonset of Pt–OH generation on the surface of the Pt–Co/G catalyst compared to the Pt/C catalyst may be related to the enhanced ORR activity. The OCV (0.71 V) of the MFC configured with Pt–Co/G was similar to that of Pt/C (0.77 ± 0.01 V), indicating a comparable ORR rate, as well as obtained MPD (1378 mW m−2) is quite close to that of the Pt/C catalyst (1406 mW m−2). However, the long-term stability of Pt–Co/G catalysts is not as good as that of commercial Pt/C catalysts, and the maximum output voltage after the cyclic operation is lower than that of Pt/C catalysts. Due to low Pt loading, The total cost of Pt–Co/G was about three-quarters of that of the Pt/C catalyst, which can be used as low-cost cathode material for MFC. Studies have shown that the ORR activity sequence of the Pt-M/C catalyst was: Pt–Fe/C > Pt–Co/C > Pt–Ni/C > Pt/C [160]. Zhang et al. [121] synthesized the C/Pt–Fe catalyst with only 0.5 mg cm−2 loading amount of Pt, which was slightly higher than the optimum loading amount (0.1 mg cm−2) [161], effectively reducing the cost. CV experiments indicated that the C/Pt–Fe alloy produced the largest current density value (–1.53 mA cm−2) compared to Pt and Fe, indicating its highest catalytic activity for ORR. It could be attributed to the presence of Fe reduced the Pt–Pt bond length and increased the electron density in the 5d orbital holes. The MPD of 1148.8 mW m−2 produced from the MFC equipped with C/Pt–Fe was 6.5% higher than that of C/Pt (1078.2 mW m−2), which makes it potentially feasible in terms of power generation and economics to be an alternative catalyst to Pt.Since Stamenkovic et al. [162] found that the activity of Pt3Ni (111) alloy was 90 times that of Pt/C, then a boom in research on Pt–Ni alloy catalysts have been set off. Yan et al. [122] reported a Pt–Ni/MWNT (atomic ratio, Pt: Ni = 1:1,15 wt% Pt) catalyst with multi-walled carbon nanotubes (MWNT) as a support, where Pt–Ni particles were uniformly distributed on the MWNT as small particles (anchoring effect). The OCV of the MFC configured with Pt–Ni/MWNT was 0.74 ± 0.01 V, which was similar to that of Pt/C (0.76 ± 0.01 V), indicating that the ORR rate of the Pt/Ni/MWNT catalyst was comparable to that of the Pt/C catalyst. In MFC, the Pt–Ni/MWNT catalyst obtained an MPD (1.22 W m−2) similar to that of Pt/C (1.40 W m−2), but its price was about three-fourths of the Pt/C catalyst, effectively reducing costs. The maximum output voltage of Pt–Ni/MWNT was only 10–20 mV lower than Pt/C during the whole operation cycle of the MFC, and the maximum voltage of each cycle was stable around 0.57 V and 0.585 V. With the contributions of excellent ORR activity, low cost, and high power, the Pt–Ni/MWNT presented a highly promising air-cathode catalyst for MFC. For understanding the effect of the Pt/Ni ratio on the ORR activity of Pt–Ni catalysts, Wang et al. [123] synthesized three highly dispersed alloy catalysts with different Pt:Ni atomic ratios on carbon support. All of the Pt–Ni alloy catalysts with narrow particle size distribution and well-dispersed showed a single-phase face-centered cubic structure, which was beneficial for enhancing the ORR activity. CV experiments concluded that Pt2–Ni/C catalysts possessed the largest electrochemically surface area (ECSA) of 81 m2 g−1, thus may feature favorable ORR catalytic activity, which is known as the Raney effect. LSV measurements showed that the Pt2–Ni/C catalyst exhibited the most positive Eonset (0.630 V) and E1/2 (0.51 V), it can be seen that the Pt2–Ni/C catalyst exhibited the highest ORR activity. Consequently, the MFC modified with Pt2–Ni/C produced 22% higher MPD (1724 mW m−2) than the commercial Pt/C (1413 mW m−2) and obtained the maximum OCV (0.78 V). To generate electricity directly from dairy wastewater, Cetinkaya et al. [124] exploited CMI7000 membranes coated with Pt–Ni alloy as the cathode of MFC. The results showed that the addition of Ni as an alloying element in a certain proportion can reduce the particle size and content of noble metal in Pt–Ni alloy (reduce the cost) without losing the active surface area. When the Ni content was increased to Pt:Ni = 1:1, the inherent catalytic activity of the Pt:Ni(1:1) catalyst reached its maximum, reducing oxygen more easily than pure Pt. The MPD (637 mW m−2) produced by the MFC with Pt:Ni(1:1) as the cathode was much higher than that of pure Pt (180 mW m−2), and 85% COD removal was obtained after the MFC power output was stabilized. Overall, the Pt:Ni(1:1) catalyst can be used as an alternative catalyst for Pt in MFC applications.Transition metal alloys composed of two relatively inexpensive or nonprecious transition metals compared to Pt have the advantage of being easily available and synthesized, which solves to a certain extent the problem of still too high Pt burden in most Pt-M catalysts, thus serving as an efficient and low-cost cathode catalyst for MFC. Relatively inexpensive transition metal alloys can provide a viable method for wastewater pretreatment, Włodarczyk et al. [125] applied Ni–Co (85% Ni, 15% Co) to MFC for the treatment of highly concentrated pollutant wastewater from a yeast plant, requiring only an additional oxygen supply from the air, the value of which was 27 times lower than that of aeration reactor. The MFC configured with Ni–Co (85% Ni, 15% Co) alloy reduced COD by 90% within 20 days and achieved a power output of 6.1 mW. Although the electricity output was not significant, it could offer extra electrical energy to the subsequent wastewater treatment process equipment. To further investigate the electricity production performance of MFC configured with different cathodes. Thereafter, Włodarczyk et al. [126] utilized MFC modified with Ni–Co alloy to treat wastewater from municipal wastewater treatment plants. The alloy containing 15% cobalt exhibited higher catalytic activity, after 8 h of oxidation and three times of anodic loading, showing the most favorable electrode voltage. The MFC with Ni–Co (85% Ni, 15% Co) as a cathode catalyst obtained a slightly higher maximum power (7.19 mW) than the carbon cloth cathode (1.56 mW), and the COD reduction time to an assumed 90% was also shortened by 3 days compared to the carbon cloth electrode.By TGA testing, Papiya et al. [127] found that the Ni–Co/MGO catalyst had the lowest total mass loss (20%) when the temperature reached 600 °C (Fig. 10 b), demonstrating its good thermal stability. The MPD obtained (1003.18 mW m−2) by applying the Ni–Co/MGO catalyst to the MFC was much stronger than the Pt/C catalyst (483.48 mW m−2). Using conductive polymers as support carriers is a potential solution for improving chemical stability and catalytic activity [163,164], such as polyaniline (PANI), polypyrrole (Ppy), polythiophene (PTH), etc. Nguyen et al. [128] gained the Ni–Co(1:1)/SPAni catalyst using sulfonated polyaniline (SPAni) as the carrier, where catalyst particles were well-dispersed and distributed on SPAni without any agglomeration and sintering, increased the active site in favor of ORR. In single-chamber MFC equipped with Ni–Co(1:1)/SPAni, MPD acquired (∼659.79 mW m−2) was greater than Pt/C (∼483.48 mW m−2), with the COD removal rate is up to 91.5%.Combining carbon-based materials with conducting polymers as cathode catalysts is an effective means of increasing ORR rates [165]. Papiya et al. [129] synthesized another supported catalyst (Mn–Co/SGO-PAni) by combining sulfonated graphene oxide (SGO) with PAni, XPS analysis revealed the presence of Mn, Co, C, N, and S elements, indicating that Mn–Co/SGO-PAni was successfully prepared (Fig. 10c), and the MFC containing Mn–Co/SGO-PAni exhibited larger MPD (1392.68 mW m−2) than Pt/C (481.3 mW m−2). The factors affecting the ORR catalytic activity of alloys are not only related to the type and content of the transition metal, but also related to its surface stacking mode (the morphology of alloys) [166]. Yang et al. [130] utilized RDE testing to find that the n value of core-shell Au–Pd was similar to that of hollow Pt nanoparticles, the semicircle diameter in the EIS (Nyquist plots) also demonstrated that the Au–Pd catalyst features a far lower Rct value than the hollow Pt nanoparticles (Fig. 10d), MFC with Au–Pd as cathode produced the MPD of 16.0 W m−3, which synthetically confirmed its excellent ORR catalytic activity.Transition non-noble metal alloys (e.g. CuZn) have been successfully adapted to fuel cells due to their good catalytic properties [167]. Das et al. [131] employed carbon-loaded CuZn to field-scale MFC (septic tank slurry as substrate), obtaining an MPD (0.32 mW m−2) 64 times higher than without the cathode catalyst, with 68 ± 8% removal of COD, the 1000 cycles of CV showed that CuZn has particularly strong electrochemical stability. The positive effect of cathodic catalysts containing M-N-C structures towards ORR has been reported [168]. The Fe–Mn–N–C catalyst prepared by Kodali et al. [91] using the sacrificial support method with higher Eonset (0.28 V) and more positive E1/2 (0.10 V) than the Fe–N–C catalyst, as well as the MPD (228.4 μW cm−2) obtained by Fe–Mn–N–C as the MFC cathode was also higher than that of the Fe–N–C catalyst (196.4 μW cm−2), indicating that the Fe–Mn–N–C catalyst had an excellent ORR catalytic performance. The simpler preparation of Fe–Mn catalysts is beneficial for practical applications, Guo et al. [132] synthesized the FeMn2 catalyst via a simple hydrothermal method applied to MFC, obtained a higher MPD (1971 mW m−2) than both FeMn (1820 mW m−2) and FeMn4 (1580 mW m−2), the possible ORR mechanism for the Fe–Mn catalyst was shown in Fig. 11 .Transition metal oxides are characterized by high specific capacity, low cost, and commercial suitability [169], which can improve the ORR catalytic performance of catalysts by facilitating oxygen adsorption and O=O bond dissociation during the ORR process. It can be classified by their metal element composition into unitary transition metal oxides (e.g. manganese series oxides (MnOx)) and transition composite metal oxides (oxides in which two or more metals co-exist). Transition composite metal oxides exhibit better electrical conductivity and catalytic properties than monometallic oxides due to the synergistic effect between multiple metal species [170,171]. In recent years, using transition metal composite oxides to replace conventional Pt/C has become a research hotspot.Since most transition composite metal oxides are semiconductors, combining them with highly conductive carriers (e.g. carbon support) can effectively improve conductivity and catalytic activity [172]. Graphene oxide (GO) is produced by strong oxidation of G, and the functional groups generated on the surface confer its good electrical conductivity and catalytic activity [173]. The FeCoO/GO catalyst obtained more catalytic activity sites while reducing agglomeration due to the addition of GO, thereby exhibiting outstanding ORR electrochemical activity [133]. In the MFC-Fenton system, the gained MPD (461.2 mW m−2) was 4.5 times higher than that of carbon felt (102.5 mW m−2) and the removal rate of 80.34% for 20 mg/L oxytetracycline (OTC), indicating that the application of FeCoO/GO to MFC-Fenton was a feasible solution for the deep removal of antibiotic contaminants.Perovskite-type metal composite oxides, in general formulation of ABO3, with high conductivity, thermal stability, and ORR electrocatalytic activity. Compared with the GO/FeMnO3 catalyst, the C/FeMnO3 catalyst was homogeneously dispersed on the carbon support, with higher porosity and surface area, significantly accelerating ORR, electrochemical characterization also showed that it had a lower Rct value (69.8 Ω) [134]. Treat it as a single-chamber MFC cathode, a higher MPD (475 mW m−2) than a Pt-based catalyst (461 mW m−2) was acquired. The C/FeMnO3 catalyst maintained good temperature cycling resistance in the long-term temperature cycling mode of operation, indicating that it was a promising ORR catalyst with high-temperature resistance. Individual perovskite metal composite oxides are widely used as photocatalysts due to their photocatalytic features of semiconductors, Ahmadpour et al. [135] confirmed that a two-chamber MFC with NiTiO3 as the cathode obtained twice as much as the MPD (76.86 mW m−2) under visible illumination than under dark conditions, which was attributed to the increase of n value caused by the decrease of Rct value from 38.67 Ω to 31.42 Ω under light conditions. This result showed that NiTiO3 is an excellent photocatalyst in MFC-coupled photocatalytic systems.Spinel metal composite oxides with the general formula AB2O4 have the same ORR catalytic activity as ABO3, which is another promising class of ORR catalysts. Liu et al. [136] found that the hexagonal shape of the CoGa2O4 catalyst exposed more active sites (Fig. 12 a and b), which was conducive to O2 adsorption and the ORR 4e− pathway, leading to a superior catalytic effect. The high SBET (207 m2 g−1) and large pore size (46 nm) of the CoGa2O4 catalyst increased the active sites while promoting the diffusion of O2 and electron transfer. RDE testing demonstrated that the CoGa2O4 catalyst had an n value close to 4 (3.87), which was 48% higher than activated carbon (AC) (2.58). MFC equipped with CoGa2O4 modified AC produced 80% higher MPD (1960 mW m−2) than reported Pt/C. Its total cost was also far lower than that of Pt/C. Nano spinel rod-like CoFe2O4 doped in AC (AC-CoFe2O4) was successfully used as an air cathode catalyst for MFC by Ren and co-workers, as well as suitable O2 adsorption energy and the direct 4-electron ORR pathway on CoFe2O4 was verified by DFT calculations [137]. In addition, the AC-CoFe2O4 catalyst also exhibited a much higher n value (3.85) than the NiCo2O4/AC catalyst (3.72) [138], indicating its good ORR kinetic activity, and the MPD obtained for MFC configured with 10% CoFe2O4 reached as high as 1800 mW m−2, which was 1.99 times higher than that of bare AC. Huang et al. [139] generated a CoFe2O4@N-AC catalyst in situ on nitrogen-doped activated carbon (N-AC), which showed good ORR electrocatalytic activity due to the synergistic effect between CoFe2O4 and N-AC, furthermore, the obtained MPD by MFC modified with the CoFe2O4@N-AC catalyst reached as high as 1785.8 mW m−2, which was 2.39 times higher than that with a bare electrode.Reduced graphene oxide (rGO) possess large surface area and abundant surface oxygen-containing functional groups, utilizing it as one support carrier for catalysts not only can provide massive adsorption sites, but also leads to stable anchoring of the active metal, thus inhibiting its agglomeration [174,175]. The CoMn2O4/rGO-8 catalyst prepared by Hu et al. [140] exhibited better ORR electrochemical activity than the CoMn2O4 catalyst, which was attributed to the uniformly embedding of CoMn2O4 nanoparticles on the rGO surface (Fig. 12c), yielding 2.6 times larger SBET (78.4 m2 g−1) with pore volume of about 0.19 cm3 g−1, providing more abundant oxygen adsorption active sites and facilitating the mass transfer of reactants and reaction products in ORR. The CoMn2O4/rGO-8 catalyst in RDE test yielded the optimum Eonset (0.69 V) and current density (1.57 mA cm−2) than the CoMn2O4 catalyst (0.64 V and 1.40 mA cm−2). Loading CoMn2O4/rGO-8 onto modified graphite felt (GF) formed the GF-CoMn2O4/rGO catalyst that well bonded to the graphite fibers (Fig. 12d). The MFC with modified the GF-CoMn2O4/rGO catalyst obtained significantly higher MPD (361 mW m−2) than other materials. To obtain electrical energy while effectively decomposing refractory pollutants in a photocatalytic assisted MFC coupled system, Li et al. [141] certificated that immobilization of the CoFe2O4-rGO photocatalyst on photocatalytic composite membranes (PCM) for assisted application in MFC/MBR systems can accelerate the cathodic ORR rate, producing an MPD (942 mW m−3) higher than the light-free mode (871 mW m−3) under natural light irradiation, with a degradation rate of up to 95% for 50 mg/L tetracycline hydrochloride (TH). These achievements indicated the potential for better and wider application of photocatalyst-assisted MFC-MBR systems in wastewater treatment.N4-transition metal macrocyclic compounds (e.g. porphyrins, phthalocyanines, etc.) are favored owing to their ORR catalytic activity comparable to that of Pt and lower cost than metal oxides. Noteworthily, the ORR catalytic activity depends mainly on the transition metal centers in the macrocyclic N4 [176], especially Fe (mainly occurs 4e− ORR) and Co (mainly occurs 2e− ORR) [177]. Thus, iron phthalocyanines (FePc) [178], cobalt phthalocyanines (CoPc) [179], iron porphyrins (FePP) [180], and cobalt porphyrins (CoPP) [181] became the most studied N4-transition metal macrocyclic compounds ORR catalysts.Carbon nanotubes (CNT), with high specific surface area, excellent mechanical strength, high thermal conductivity, and electrical conductivity [182], possess high application prospects in the aspect of acting as carbon carriers for catalysts. Using N4-transition metal macrocyclic compounds as precursors for the preparation of bimetallic catalysts is a feasible approach. Deng et al. [142] discovered that the Co/Fe/N/CNT catalyst prepared by high-temperature pyrolysis of the CoTMPP/FePc functionalized CNT precursors mainly supported the ORR 4e− pathway. The CoTMPP and FePc attach to the CNT surface via π-stacking interactions, which facilitated the increase of catalytically active surface, while the presence of CNT also promoted electron transfer (3.75–3.81) and reduced the ORR overpotential. The OCV of the MFC with Co/Fe/N/CNT as cathode catalyst was 0.76 V, and the obtained MPD (751 mW m−2) was higher than that of Pt/C (498 mW m−2) and Co/Fe/N/graphite (618 mW m−2), with no significant change in output power after 2 months. These results indicate that Co/Fe/N/CNT was a suitable material for the preparation of MFC.In 2019, Noori et al. [143] successfully synthesized the Co-FePc/carbide-derived carbon (CDC) catalyst by introducing FePc into Co/CDC, TEM analysis revealed that FePc extensively covered the surface of Co/CDC (Fig. 13 a), the imperfect molecular arrangement of Co-FePc (Fig. 13b) indicated the stacking of different materials on CDC, confirming the successful doping of Co-FePc in CDC. Based on the analysis of the N2 physisorption technique, the Co-FePc/CDC had the most suitable pore volume (0.159 cm3 g−1) and SBET (212 m2 g−1) to effectively promote the electron transfer to the catalyst layer, indicating the kinetic propensity of the catalyst for ORR. Due to the synergistic effect of Co and FePc in CDC, the Co-FePc/CDC catalyst could improve the ORR catalytic activity through the 4e− ORR pathway. Electrochemical tests confirmed that Co-FePc/CDC possessed a significant oxygen reduction peak and a more positive Eonset. When it was applied to MFC (external resistance of 500 Ω) containing acetate-based synthetic wastewater, the maximum MPD and OCV obtained were 1.57 W m−2 and 741 mV, respectively (Fig. 12c), and the COD removal rate was 86%. CA tests showed that The current density response of Co-FePc/CDC at each applied voltage was almost constant for 20,000 s. The RDE test also demonstrated that the diffusion current density of Co-FePc/CDC remained almost constant until 250 cycles and varied slightly after 500 cycles, indicating the high stability of the catalyst.Metal-organic framework (MOF) is one zeolite-like material with a three-dimensional microporous network structure formed by the self-assembly of a metal source (inorganic metal ions or metal clusters) and a carbon source (organic ligands) through coordination bonds [183]. Compared with conventional porous materials, it has a greater specific surface area, a more developed and regular pore structure, and superior chemical and thermal stability [184]. Currently, MOF is widely used in catalytic reactions [185], sensing [186], biomedicine [187], and other fields, which are favored by many research scholars. MOF-based catalysts prepared with MOF as the base or sacrificial templates are prone to defective oxygen vacancies (exposing more active sites) during calcination, thus having the potential to promote ORR.N-doped MOF has been successfully applied in MFC to improve catalytic activity [188]. In 2019, Using the Cu/Co/N–C#2 catalyst with hexagonal hollow structures that were synthesized via Cu and N co-doped with ZIF-67 (Co) as the cathode of MFC, which was a feasible approach to improve the catalytic performance of ORR, yielding an MPD of 1008 mW m−2 and a maximum stable output voltage of 677 mV, which are 1.25 and 1.31 times higher than the 20% Pt/C catalysts [144]. The analysis revealed that outstanding performance for Cu/Co/N–C#2 was mainly attributed to the high SBET (286 m2 g−1) and pore size (approximately 8 nm) of both larger than Co–N–C, which increased the effective catalytic active site and facilitated electron transfer and mass transfer. The successful doping of Cu and Co elements, especially the presence of Cu ions increases the N content through Cu–N, also made an important contribution to the improvement of ORR catalytic activity. The values of the Eonset (0.25 V) and E1/2 (0.14 V) of the Cu/Co/N-C#2 catalyst are better than those of Co/N–C (0.02 V and 0.24 V). CA tests revealed that with the addition of methanol, the current drops of Cu/Co/N–C#2 and Pt/C were 2.6% and 6.8%, confirming that Cu/Co/N–C#2 had better resistance to methanol neutrality. In addition, the current density of Cu/Co/N–C#2 decreased by only about 15.4% after 8000 s, with better long-term stability than Pt/C (24.9%). In 2020, the Mn–Fe@g-C3N4 catalyst was fabricated via pyrolyzing Mn-doped g-C3N4 assisted Fe-based MOFs (MIL-101), which was an outstanding air-cathode in MFC with an MPD (420 mW m−2) and maximum stable output voltage (0.450 V), superior to 20 wt% Pt/C (333.9 mW m−2 and 0.422 V) [145]. At high temperatures, Mn ions and N atoms can interact with the carbon layer to optimize the SBET (268.6 m2 g−1) and pore volume (0.119 cm3 g−1) of the Mn–Fe@g-C3N4 catalyst with a graded porous structure, accelerating O2 transport and internal proton and external electron transfer, as well as exposing more potential catalytically active sites. The Mn–Fe@g-C3N4 catalyst exhibited superior Eonset (0.393 V) and E1/2 (−0.042 V) over the state-of-the-art Pt/C catalysts (0.343 V and −0.067 V). The results indicated that its excellent ORR catalytic activity was mainly attributed to the 3D interconnected porous structure, the high conductivity framework, and the synergistic effect of the N ions with the metal ion centers. Based on CA tests, the current density of the Mn–Fe@g-C3N4 catalyst decayed to 90.1% after 30,000 s with long-term durability, while the Pt/C catalyst maintained only about 59.4% of the original activity. With the addition of methanol at 1000 s, the Mn–Fe@g-C3N4 catalyst performed well tolerated, no significant change in current density. In the same year, Xue et al. [146] discovered that FeCo nanoparticles were completely dispersed in the carbon matrix without significant aggregation in FeCoNC-900 catalysts prepared with ZIF-67 (Co) as a precursor, and the metal-bound state nitrogen (M−N) and pyridine state nitrogen in FeCoNC-900 promoted plasmid transport. Although the increase of the pyrolysis temperature led to the transformation of the catalyst micro-to mesoporous structure, the FeCoNC-900 with the presence of a large number of mesopores still had a relatively high SBET (864.13 m2 g−1) and pore volume (2.83 cm3 g−1), favoring the increase of ORR active sites and promoting electron transfer, exhibiting a significantly higher 4-electron transfer number than the other samples. The MPD (1769.95 mW m−2) obtained by the FeCoNC-900 in MFC was superior to Pt/C (1410.31 mW m−2) and exhibited superior durability and stability after 4 months of continuous operation with no significant change in MPD.Based on the synergistic effect between bimetallic components and the features of stability and functionalization of MOF, bimetallic MOF can be used as precursors for the preparation of efficient catalysts [189]. In 2020, Wang et al. [147] obtained Cu/Co/N-HS-3 catalyst with uniform hollow structures by pyrolysis of polystyrene@Cu/CoZIFs composite precursors, which were free from aggregation and structural collapse and existed micropores on the surface. N2 adsorption isotherm characterization illustrated that the Cu/Co/N-HS-3 catalysts had a large SBET (708 m2 g−1) and a suitable pore structure, which is favorable for active site exposure and proton transport. The pyridine-N, graphite-N, Co–N and Cu–N are considered as effective ORR active sites. The Eonset (0.25 V) and E1/2 (0.13 V) values of Cu/Co/N-HS-3 were greater than 20% Pt/C (0.24 and 0.12 V). The MPD (1016 mW m−2) gained also better than commercial Pt/C (908 mW m−2) when applied to MFC. When methanol was added, the ORR activity of Cu/Co/N-HS-3 catalyst and Pt/C decreased by 7.5% and 18.4%, while Cu/Co/N-HS-3 maintained 84.5% electrocatalytic ORR activity after 10,000 s, confirming its superiority over Pt/C catalyst in terms of methanol poisoning resistance and long-term stability. In 2021, Li et al. [148] used the Fe–Co–C/N catalyst synthesized with ZIF-L&FeTPP@ZIF-8 as precursors as an air-cathode catalyst, and exhibited a degradation efficiency of 61.64% toward 6 mg/L sulfamethoxazole (SMX) within two days and obtained an MPD of 219.45 mW m−2. RDE tests revealed that Fe–Co–C/N had an excellent electrocatalytic activity with n values ranging from 4.01 to 4.24, indicating an ideal 4-electron pathway for its ORR process, mainly attributed to the high content of pyridine-N (62.5%) and Co (63.7%), which facilitated electron transfer and enhanced ORR catalytic activity. In the long-term stability test, although the output voltage of the MFC decayed with increasing SMX concentration, it remained stable overall (nearly 1400 h), indicating that the Fe–Co–C/N is a feasible cathode catalyst for antibiotic wastewater treatment of MFC.Research demonstrated that heteroatom doping and oxygen vacancy formation (at high temperatures) can lead to a reduction in active sites [190], so catalysts with higher ORR activity and stability need to be developed. In 2020, Yan et al. [149] revealed that direct contact of FeCoS(MOF) with the substrate material (foam Ni) produced abundant oxygen defects (increased active sites) and reduced interfacial resistance (accelerated electron transfer). The layered porous FeCoS(MOF) present a multi-energy microspheres (average diameter of 1 μm) morphology, which facilitated the exposure of active sites. The microporous structure of the surface facilitated electron transfer and mass transfer, where N doping significantly improves the oxygen adsorption and conductivity, exhibiting excellent ORR activity. The FeCoS(MOF) exhibited the highest E1/2 (−0.208 V), which was comparable to that of Pt/C (−0.124 V). Poisoning tests with the addition of S2− and SCN− revealed that the FeCoS(MOF) catalyst maintained a stable current output with an insignificant current drop in both cases (15% and 12%), which was superior to that of Pt/C catalysts (more than 30%). The MPD of the FeCoS(MOF) catalyst-based MFC was enhanced to 1008 mW m−2, which was 2.55 times higher than Pt/C. After 2 months of MFC operation, the output voltage of FeCoS(MOF) remained steady, while that of the Pt/C catalyst dropped to 67% of the initial voltage, revealing superior toxicity resistance and stability than that of the Pt/C cathode. The mechanical strength of the MFC with FeCoS(MOF) as the cathode decreases significantly after a long subsequent reaction, in 2022, Yan et al. [150] also implanted B-doped graphene quantum dots (BGQDs) into FeCoMOF to obtain the BGQDs/MOF-15 catalyst with flower-like morphology (Fig. 14 a), which facilitated fast electron transfer and mass transfer. The BGQDs implantation provided an efficient charge transfer rate and abundant edge active centers, which were potential active sites for ORR. Based on electrochemical tests, the value of the Eonset of BGQDs/MOF-15 (0.014 V) was significantly higher than that of other catalysts, which was even close to that of Pt catalyst (0.04 V). The MFC with BGQDs/MOF-15 catalyst as cathode produced 1.53 times more MPD (704.24 mW m−2) than Pt/C (460.29 mW m−2). After 50 h of operation, the current of the BGQDs/MOF-15 was maintained at 91.2%, much higher than that of Pt/C (57.5%) (Fig. 14b). After 800 h, the maximum voltage output of BGQDs/MOF-15 was maintained at approximately 0.6 V compared to 0.51 V for the Pt/C electrode (Fig. 14c), demonstrating the long stability of the BGQDs/MOF-15 catalyst in the MFC.During the long-term operation of MFC, microorganisms attached to the cathode and the harmful substances produced by their secretions can block the active site and lead to catalyst “poisoning” deactivation, thus increasing the diffusion resistance and internal resistance of the MFC system and impeding the ORR process [191]. Combining ORR catalysts with antimicrobial components and developing bifunctional catalysts (increasing ORR activity while inhibiting microbial contamination) is an effective way to improve MFC output power and long-term stability. It has been demonstrated that Ag and Cu, two heavy metal elements, have certain antibacterial activity, reducing efficiently the adhesion and viability of microorganisms [192,193], which are the most widely used antibacterial active metal materials.Based on the sterilization performance of Ag NPs and the principle that alloying with any element does not lose its bactericidal properties [194], In 2018, Noori's team reported a C–Ag3–Pt catalyst with a high ORR current response (5.2 mA) and positive reduction potential (−0.06 V) than the C–Pt catalyst, which can reduce the activation energy barrier of ORR due to the striking molecular arrangement and d-band electron sharing ability. The MFC decorated with C–Ag3–Pt obtained MPD up to 1030 mW m−2, which was superior to C–Pt (963 mW m−2) and carbon catalysts (111 mW m−2) [151]. After 40 days of MFC operation, no obvious signs of cracks and biological contamination were observed on the cathode (C–Ag3–Pt) or membrane surface, and no significant changes in output voltage. This high activity and bacterial inhibition were attributed to the synergistic effect between the Pt and Ag components and the poisoning of microorganisms via Ag NPs. The cost-benefit assessment analysis showed that the C–Ag3–Pt catalyst had an excellent normalized MPD (39.1 mW $−1), reflecting its potential economic viability for scale-up applications. Modification of Ag NP onto Co–N–C may be a strategy to improve ORR activity and antibacterial capacity. This is well demonstrated by Jiang and co-workers, who observed that the two-dimensional Ag/Co–N–C-30 catalyst with optimal Ag content presented suitable ECSA (286.8 m2 g−1) and graded porous structure varying from micropores to mesopores, facilitating the transfer of reactive species and electrons. The synergistic effect of the higher conductivity Ag NP and Co, N to provide antibacterial and ORR activity for the Ag/Co–N–C-30 catalyst, which had an Eonset of 0.88 V close to 20% Pt (0.91 V) and showed the lowest Tafel slope (51.25 mV dec−1). The MFC equipped with Ag–Co–N–C-30 obtained an MPD of 560.6 mW m−2 and maintained an excellent output voltage (468 ± 17 mV) even after 1600 h of operation [152]. Using the E. coli as the antibacterial model to quantitatively investigate the antibacterial activity of different concentrations of Ag/Co–N–C-30 catalyst, the results showed that the numbers of colonies and colony-forming units (CFU) decreased significantly with increasing catalyst concentration (Fig. 15 a and b), among them, the best inhibition of aerobic bacteria in cathode biofilm was attributed to the selective antibacterial ability of Ag/Co–N–C-30.Combining Ag NPs with MOF is an effective strategy for improving the performance and bacteriostatic ability of catalysts [195]. In 2021, Zhong and co-workers produced a porous carbon catalyst (Ag/Fe–N–C-2:1) with relatively regular octahedral morphology via Ag/Fe co-doping UiO-66-NH2 (Zr-based MOF). The N2 adsorption isotherm and pore structure distribution indicated that Ag/Fe–N–C-2:1 had a hierarchical pore structure (coexistence of micro- and mesopores) that mainly developed at approximately 0.5–2 nm, and the maximum pore volume (0.129 cm3 g−1) facilitated the mass transfer of reactive species at the interface to the catalytic site. Its possession of the largest SBET (311.7 m2 g−1) increased the exposure of the active center and was highly conducive to facilitating the ORR process [153]. Based on electrochemical tests, the Ag/F-N-C-2:1 catalyst showed excellent Eonset (1.14 V) and Tafel slope (78 mV dec−1), even comparable to commercial Pt/C catalysts (1.14 V and 65 mV dec−1). Such excellent ORR activity was mainly attributed to the synergetic effect between Ag and Fe to optimize the d-band center (promoting O–O bond breaking and charge transfer rates) (Fig. 15c). Moreover, the number of E. coli colonies decreased significantly with increasing Ag/Fe–N–C-2:1 concentration and was significantly inactivated within 12 h, demonstrating its excellent antibacterial activity. The MPD of Ag/Fe–N–C-2:1 catalyst-based MFC is raised to 1285.1 mW m−3, which was superior to that of Pt/C (1101.5 mW m−3), and a steady maximum output voltage of 0.425 V was maintained after eight cycles of operation. Based on the fact that the application of Cu-based MOFs as cathode catalysts with antimicrobial function is not common, Wang et al. [154] prepared CuCo@NCNTs catalyst by impregnation and pyrolysis, in which doping of Co and N promoted ORR via forming highly active sites. The N2 adsorption isotherm and pore structure distribution showed that the flowing gas could form pores to increase the SBET of the CuCo@NCNTs catalyst, with a significant increase in SBET to 96 m2 g−1, as well as also exhibiting a multistage pore structure with the coexistence of micropores and mesopores, contributing to mass transfer and exposure of more active centers. The CuCo@NCNTs catalyst exhibited excellent Eonset (0.91 V) and Tafel slope (44 mV dec−1), even better than commercial Pt/C catalysts (0.82 V and 92 mV dec−1). Based on antibacterial tests, this catalyst was dissolved in deionized water and dropped into the bacteria-containing culture medium, which showed a significant bacteria-inhibiting ring. Furthermore, the MPD obtained by CuCo@NCNTs as MFC cathode was up to 2757 mW m−3, and the biomass of the cathode was only 0.35 ± 0.048 mg cm−2 after one month of MFC operation, which was significantly lower than that of Pt/C cathode (0.57 ± 0.061 mg cm−2). After 3 months of operation, a stable output voltage of 0.51 V was maintained. These results collectively indicated that CuCo@NCNTs exhibit strong stability and antibacterial ability.Throughout the history of MFC development, it is not hard to observe that the ultimate goal of MFC is to achieve energy-efficient wastewater treatment, not to add additional costs. Table 2 summarized the key contributions and application effects of bimetallic catalysts for MFC. It can be concluded that most bimetallic catalysts are durable after long-term operation, stable in wastewater, and economically feasible, which enables MFC to treat wastewater while obtaining higher and stabler output power better than or comparable to Pt/C catalysts. Specifically, compared to simple Pt catalysts, due to electronic and geometrical effects, Pt-M (M = Fe, Co, Ni) alloyed with transition metals significantly improves the catalytic performance of ORR with reduced Pt utilization, obtaining appreciable MPD, but tolerance and long-term stability are still not specifically discussed. To further improve cost-effectiveness, ORR activity and stability, transition metal alloys rich in Earth-rich elements are first considered. Transition metal alloys are simple to prepare and inexpensive, maintaining good stability and high MPD even during actual wastewater treatment. Transition composite metal oxides with oxidation states and unique structures, combined with highly conductive carriers, can better improve the electrocatalytic activity, showing the incomparable advantages of metal catalysts, i.e., unique photocatalytic properties and high stability, and high output MPD makes it stand out, but it does not seem to be competitive in terms of durability and cost. Transition metal macrocyclic compound-based bimetallic catalysts possess low manufacturing cost, exhibit excellent catalytic activity, tolerance and high stability, achieving more stable MPD. Finally, MOF-based bimetallic catalysts with inherent porous structure, abundant pores and high SBET are favorable for mass transfer and improved catalytic activity, and when used as bifunctional catalysts, they achieve increased ORR activity while inhibiting microbial contamination, resulting in highly stable MPD and long-term durability, which makes it probably the most promising candidate for practical applications, but the adaptability to high-temperature pyrolysis still needs to be improved.In recent years, with the continuous deepening of the applied research on bimetallic catalysts, we can find that a large number of problems remain to be solved if the MFC decorated with bimetallic catalysts is to achieve efficient wastewater treatment and scale-up application. To facilitate future research and practical applications of bimetallic catalysts, this review offers the following insights into the existing challenges. (1) Bimetallic synergistic catalysis is an important way to regulate and control the catalytic performance of bimetallic catalysts. The mutual accompaniment of electronic and geometric effects between bimetallic components is challenging to identify the key causes leading to enhanced catalytic performance of bimetallic catalysts. The use of more advanced in situ characterization techniques (e.g. In situ XAS) [38] combined with theoretical calculations (e.g. DFT) and experimental methods is vital to further essentially reveal the dominant effect of synergistic catalysis of different bimetallic catalysts during ORR, as well as to clarify the real active sites of bimetallic catalysts, which may provide theoretical guidance for the design of bimetallic catalysts with higher activity and stability in the future. (2) It is well known that the synthesis methods of bimetallic catalysts commonly play a key role in their catalytic performances and behaviors. Each of the current typical synthesis methods has certain weaknesses, indicating that it is still a developing field. Among them, relatively difficult to achieve highly accurate regulation of the microstructure of bimetallic nanoparticles becomes a key limiting factor, which implies that the modification of these methods is imperative, i.e., the development of new synthesis methods should make the above problems possible and provide some reference for the future design of bimetallic ORR catalysts with accurate regulation of structure, components, and morphology on this basis. (3) At present, the application of bimetallic catalysts in MFC coupled systems is comparatively less studied, even though applied to MFC, the substrates are also mostly from synthetic wastewater or high concentrations of organic pollutants, and there are still few studies on the targeted degradation of emerging pollutants. Therefore, in the future, the application research of bimetallic catalysts can be expanded more to MFC coupled with photocatalytic, electro-Fenton, and MBR systems, especially for the degradation of emerging contaminants (e.g. antibiotics, PAHs, personal care products, endocrine disruptors, etc.), which still need to be further strengthened. (4) In recent years, the degradation of antibiotics in MFC decorated with bimetallic catalyst has gradually become a hotspot and gained achievements to some extent, however, the long-term stable operation research is still in the initial stage, how to achieve long-term stable operation (with high output voltage) and practical application of MFC should be the key direction of subsequent research. Given the future development trend of MFC is from laboratory scale to field scale, or even towards actual industrial power generation, the formation of strong metal-carrier interactions through the selection of suitable catalyst carriers (e.g. heteroatom-doped carbon materials, conductive polymers, etc.) may be an effective strategy to enhance the long-term stability of MFC. (5) MFC configurated with bimetallic catalysts is still facing the problem of relatively low output power when applied to on-site scale wastewater treatment and does not achieve the expected effects. Future practical applications are more likely to be as an alternative energy source: pretreating wastewater from a wastewater treatment plant while providing a small amount of electricity for subsequent low-power equipment (e.g. aerators, etc.) and for daily use in nearby rural areas (e.g. night-time toilet lighting, etc.). Alternatively, this electrical energy could also be used to build MFC self-powered biosensors for real-time monitoring of pollutants in wastewater. (6) Currently, MOF-based bimetallic catalysts have achieved gratifying results in MFC applications owing to their remarkable specific surface area and outstanding porosity. There are still some issues such as the limited variety of MOF precursors, the relatively complicated preparation process, the adaptability to high-temperature pyrolysis, and the over-generalized cost estimation. Considering those, exploring other effective MOF precursors or appropriately broadening the choice of their organic ligands will help to expand the diversity of MOF derivatives. Developing simpler synthesis methods will be an important step toward achieving commercial production and application of MOF-based bimetallic catalysts. Besides, further cost estimation using normalized MPD, which characterizes power production costs, is significant for the overall evaluation of cost-effectiveness. Bimetallic synergistic catalysis is an important way to regulate and control the catalytic performance of bimetallic catalysts. The mutual accompaniment of electronic and geometric effects between bimetallic components is challenging to identify the key causes leading to enhanced catalytic performance of bimetallic catalysts. The use of more advanced in situ characterization techniques (e.g. In situ XAS) [38] combined with theoretical calculations (e.g. DFT) and experimental methods is vital to further essentially reveal the dominant effect of synergistic catalysis of different bimetallic catalysts during ORR, as well as to clarify the real active sites of bimetallic catalysts, which may provide theoretical guidance for the design of bimetallic catalysts with higher activity and stability in the future.It is well known that the synthesis methods of bimetallic catalysts commonly play a key role in their catalytic performances and behaviors. Each of the current typical synthesis methods has certain weaknesses, indicating that it is still a developing field. Among them, relatively difficult to achieve highly accurate regulation of the microstructure of bimetallic nanoparticles becomes a key limiting factor, which implies that the modification of these methods is imperative, i.e., the development of new synthesis methods should make the above problems possible and provide some reference for the future design of bimetallic ORR catalysts with accurate regulation of structure, components, and morphology on this basis.At present, the application of bimetallic catalysts in MFC coupled systems is comparatively less studied, even though applied to MFC, the substrates are also mostly from synthetic wastewater or high concentrations of organic pollutants, and there are still few studies on the targeted degradation of emerging pollutants. Therefore, in the future, the application research of bimetallic catalysts can be expanded more to MFC coupled with photocatalytic, electro-Fenton, and MBR systems, especially for the degradation of emerging contaminants (e.g. antibiotics, PAHs, personal care products, endocrine disruptors, etc.), which still need to be further strengthened.In recent years, the degradation of antibiotics in MFC decorated with bimetallic catalyst has gradually become a hotspot and gained achievements to some extent, however, the long-term stable operation research is still in the initial stage, how to achieve long-term stable operation (with high output voltage) and practical application of MFC should be the key direction of subsequent research. Given the future development trend of MFC is from laboratory scale to field scale, or even towards actual industrial power generation, the formation of strong metal-carrier interactions through the selection of suitable catalyst carriers (e.g. heteroatom-doped carbon materials, conductive polymers, etc.) may be an effective strategy to enhance the long-term stability of MFC.MFC configurated with bimetallic catalysts is still facing the problem of relatively low output power when applied to on-site scale wastewater treatment and does not achieve the expected effects. Future practical applications are more likely to be as an alternative energy source: pretreating wastewater from a wastewater treatment plant while providing a small amount of electricity for subsequent low-power equipment (e.g. aerators, etc.) and for daily use in nearby rural areas (e.g. night-time toilet lighting, etc.). Alternatively, this electrical energy could also be used to build MFC self-powered biosensors for real-time monitoring of pollutants in wastewater.Currently, MOF-based bimetallic catalysts have achieved gratifying results in MFC applications owing to their remarkable specific surface area and outstanding porosity. There are still some issues such as the limited variety of MOF precursors, the relatively complicated preparation process, the adaptability to high-temperature pyrolysis, and the over-generalized cost estimation. Considering those, exploring other effective MOF precursors or appropriately broadening the choice of their organic ligands will help to expand the diversity of MOF derivatives. Developing simpler synthesis methods will be an important step toward achieving commercial production and application of MOF-based bimetallic catalysts. Besides, further cost estimation using normalized MPD, which characterizes power production costs, is significant for the overall evaluation of cost-effectiveness.In a nutshell, bimetallic catalysts can become the research hotspots for MFC cathode ORR catalysts in recent years mainly due to two points: firstly, the synergistic effects between the bimetallic components endow them with better catalytic performance than monometallic catalysts, and secondly, they are simpler to prepare than multimetallic catalysts, making them more feasible for practical applications. Although bimetallic catalysts currently suitable for MFC have been studied extensively, the lack of a collective summary of them hinders the prediction of future research directions. Therefore, this review is designed to address the aforementioned significant issues. It is noteworthy that the trend of bimetallic catalyst research from Pt-M alloys to MOF-based bimetallic catalysts is obvious, especially when using MOF-based bimetallic catalysts as ORR and bifunctional catalysts, the overall performance of MFC has been significantly improved. However, there is still much work to be done if MFC is to move from experimental scale to actual industrialization. These include the broadening of MOF precursors, simplification of synthesis methods, cost estimation using normalized MPD, and the selection of suitable catalyst carriers to improve MFC output power and long-term stability (especially in the degradation of emerging contaminants). By updating the above conclusions, MOF-based bimetallic catalysts are expected to become one of the most desirable bimetallic ORR catalysts for MFC in the future.We have consulted the Guide for Authors in preparing this manuscript and confirm that the manuscript is prepared in compliance with the Ethics in Publishing Policy as described in the Guide for Authors. No conflict of interest exists in the submission of this manuscript and is approved by all authors for publication. The work has not been published previously and is not under consideration for publication elsewhere.This study was supported by the National Key R&D Program of China (2019YFC1804102) and the National Natural Science Foundation of China (32171615).

Microbial fuel cell (MFC) is one synchronous power generation device for wastewater treatment that takes into account environmental and energy issues, exhibiting promising potential. Sluggish oxygen reduction reaction (ORR) kinetics on the cathode remains by far the most critical bottleneck hindering the practical application of MFC. An ideal cathode catalyst should possess excellent ORR activity, stability, and cost-effectiveness, experiments have demonstrated that bimetallic catalysts are one of the most promising ORR catalysts currently. Based on this, this review mainly analyzes the reaction mechanism (ORR mechanisms, synergistic effects), advantages (combined with characterization technologies), and typical synthesis methods of bimetallic catalysts, focusing on the application effects of early Pt-M (M = Fe, Co, and Ni) alloys to bifunctional catalysts in MFC, pointing out that the main existing challenges remain economic analysis, long-term durability and large-scale application, and looking forward to this. At last, the research trend of bimetallic catalysts suitable for MFC is evaluated, and it is considered that the development and research of metal-organic framework (MOF)-based bimetallic catalysts are still worth focusing on in the future, intending to provide a reference for MFC to achieve energy-efficient wastewater treatment.

Hydrogen (H2) is becoming increasingly important as a future fuel compared with fossil fuels because of its advantages of clean and renewable energy generation (Dresselhaus and Thomas, 2001; Turner, 2004). Electrochemical water splitting provides an effective approach for H2 production. Water splitting consists of the hydrogen evolution reaction (HER) and oxygen evolution reaction (OER), both of which require efficient catalysts to reduce the overpotential for practical applications (Jin et al., 2015; Zhang et al., 2017a; Balogun et al., 2016). Although platinum (Pt) is regarded as a conventional HER catalyst in acidic solutions owing to its highest exchange current density and low Tafel slope, it shows an “incomparable” HER activity in alkaline solutions owing to the sluggish reaction kinetics (Ma et al., 2017; Mahmood et al., 2017; Zheng et al., 2016). Even though non-noble metal materials have been widely explored as enhanced catalysts for HER, the greatest challenge for the use of non-noble metal materials so far is that their HER activities still underperform Pt-based catalysts, and they are susceptible to acid corrosion (Zhang et al., 2017b; Conway and Tilak, 2002). Similar obstacles are still unavoidable for non-noble metal materials for OER applications owing to their relatively high overpotentials for driving the OER process and the low energy conversion efficiencies. To date, pursuit of effective catalysts for both OER and HER in the same electrolyte, not to mention under universal pH conditions, has been extremely desirable (Zheng et al., 2014; Wang et al., 2018a, 2018b; Ellis et al., 2010). Therefore, the development of efficient and stable bifunctional catalysts for the simultaneous production of H2 and oxygen (O2) under universal pH conditions is still a significant challenge.It has been generally considered that noble metal materials, such as Ru-based catalysts, are the most promising catalysts for use as overall water-splitting catalysts owing to their promising activities for the two half-reactions in both acidic and alkaline solutions as well as their high stability under extreme conditions (Lu et al., 2014; Jin et al., 2016; Petrykin et al., 2010; Seitz et al., 2016; Kong et al., 2016). However, the water-splitting performances of the reported Ru-based catalysts are still far from satisfactory, particularly under universal pH conditions. From the viewpoint of the structure, a two-dimensional (2D) structure can provide great opportunities for enhancing the electrochemical performance because it largely exposes the surface area (Hang et al., 2014; Gao et al., 2012). However, undesirable drawbacks arise from the severe aggregation or fracture that usually occurs during the electrochemical process, inevitably leading to the obvious activity decay. This renders the conventional 2D structure not an ideal candidate for efficient electrocatalysis (Zheng et al., 2014; Chhowalla et al., 2013; Hwang et al., 2011; Chen et al., 2015). Based on this, the assembly of 2D structures into unique 3D structures may provide an effective strategy to achieve efficient catalysts for water splitting under universal pH conditions because the structures can achieve a large exposure of the active sites while stabilizing the structure.To surmount this challenge, we report an efficient wet chemical approach for the synthesis of 3D hierarchical Ru-Ni nanosheet assemblies (NAs) consisting of ultrathin nanosheets as subunits and explore their high performances for overall water splitting under universal pH conditions. The distinctive hierarchical NA structures are highly beneficial for enhancing electrochemical energy conversion. We found that the introduction of Ni into Ru largely downshifts the d-band center of the Ru-Ni NAs and effectively modulates the surface environment for HER. After air treatment at 350°C, the newly generated abundant RuO2 provides effective active sites for OER. As a result, the Ru-Ni NAs deliver high HER and OER activities as well as outstanding stability under a broad range of pH conditions. More interestingly, Ru3Ni3 NAs demonstrated potential applications for overall water splitting with a lower overpotential, smaller Tafel slope, and better stability than the reference Ir/C-Pt/C catalyst.A typical preparation of Ru-Ni NAs was introduced by adding ruthenium(III) acetylacetonate (Ru(acac)3), nickel(II) acetylacetonate (Ni(acac)2), phloroglucinol, tetramethylammonium bromide, polyvinylpyrrolidone (PVP), and benzyl alcohol into a glass vial. After capping the vial, the mixture was ultrasonicated for approximately 1 h. The resulting homogeneous mixture was then heated from room temperature to 160°C and maintained at 160°C for 5 h using an oil bath. Ru-Ni NAs with different Ru/Ni ratios (i.e., Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs) have been readily achieved by tuning the Ru/Ni precursor amount ratios (Figures S1A–S1C).The detailed characterizations of Ru3Ni3 NAs were further carried out to determine the 3D assembly structure (Figures 1 , S1D, and S1E). The high-angle annular dark-field scanning transmission electron microscopic (HAADF-STEM) image (Figure 1A) showed at first glance that all the products had a spherical outline, which indicated the high purity of product. For a close view of the Ru3Ni3 NAs, enlarged HAADF-STEM was performed, and a 3D flower-like structure assembled by hierarchical 2D nanosheet subunits was clearly observed (Figure 1B). Elemental mappings and line scans showed that the flower-like Ru3Ni3 NAs had a typical core-shell structure consisting of a Ru-Ni core and Ru shell (Figures 1C and 1D). Compared with those of the pure Ru NAs, additional X-ray diffraction (XRD) peaks in the Ru-Ni alloy were observed for 3D Ru3Ni3 NAs, which further confirmed the core-shell structure of the Ru3Ni3 NAs with the Ru phase and Ru-Ni alloy phase (Figures S1D and S1E). As revealed by the high-resolution transmission electron microscopic (TEM) image of the Ru3Ni3 NAs, lattice fringes with interplanar distances of 0.204 and 0.230 nm were observed, which correlated well with the (101) plane of Ru and the (100) plane of the Ru-Ni alloy, respectively (Figures 1E–1G).Notably, the morphologies of Ru3Ni2 NAs and Ru3Ni1 NAs with different Ru/Ni ratios were similar (Figures S2A, S2B, S2E, S2H, S2I, and S2L). The XRD results show that as the amount of Ni increased, the main peaks of the Ru-Ni alloy approach the standard pure Ni XRD peaks (PCPDS No. 89–7,129), which suggested the successful alloying of Ni into Ru. The energy dispersive spectroscopy (EDS) elemental mapping images and line scans confirm that the alloys have a core-shell structure similar to that of the Ru3Ni3 NAs (Figures S2C, S2D, S2J, and S2K). The same lattice fringes with an interplanar distance of 0.204 nm were found in the Ru3Ni2 NAs and Ru3Ni1 NAs, which correlated well to the (101) plane of Ru. Lattice fringes of the (100) Ru-Ni alloy with interplanar distances of 0.231 and 0.232 nm were also observed in the Ru3Ni2 NAs and Ru3Ni1 NAs, respectively (Figures S2F, S2G, S2M, and S2N).The direct creation of unique, 3D Ru-Ni superstructures with ultrathin building blocks is the most striking feature of the synthesis reported here, which has never been reported previously. To gain a better understanding of the growth mechanism behind the successful synthesis, characterizations of the intermediates collected at different reaction times were also carefully performed (Figures S3A–S3J). At the beginning of the reaction (25 min), intermediates with messy and irregular multi-branched structures were observed (Figures S3A and S3B). Nanosheets began to form, and a portion of the assembled flower-shaped intermediates appeared at a reaction time of 40 min (Figures S3C and S3D). When the reaction reached 1.5 h, the diameter of the flower-shaped intermediates increased (Figures S3E and S3F). After the reaction progressed for 3 h, the monodispersed, hierarchical assembly became obvious (Figures S3G and S3H). A further increase in the size of the Ru3Ni3 NAs was observed after the completion of the reaction (Figures S3I and S3J). The different reaction intermediates were also further analyzed by XRD (Figure S4), and the peaks of Ru and small peaks of the Ru-Ni alloy were detected during the initial 25 min. With the prolonged reaction time, the peak indexed to the Ru-Ni alloy became increasingly obvious and shifted to a higher angle, which indicated that more Ni was reduced and alloyed with Ru (Figure S3K).To further understand the formation progress behind the successful synthesis, the effect of various experimental parameters on the morphology of Ru-Ni NAs was carried out. The results reveal that the combined use of PVP, phloroglucinol, and tetramethylammonium bromide was essential for the successful creation of Ru-Ni NAs. The Ru-Ni NAs could not be obtained in the absence of any PVP or phloroglucinol (Figures S5A, S5B, S7A, and S7B). Further detailed control experiments show that high-quality Ru-Ni NAs could only be obtained in the presence of specific amount of phloroglucinol and tetramethylammonium bromide. For example, irregular morphology was obtained when the amounts of phloroglucinol and tetramethylammonium bromide were changed (Figures S5 and S6), and a layered product with low yield was obtained when benzyl alcohol was replaced by ethylene glycol (Figures S7C and S7D). The morphology of assemblies has changed greatly without using Ni(acac)2 (Figure S8).Considering that Ru is expected to have high activities for HER and OER, the design of Ru-based catalysts for overall water splitting is highly significant from the viewpoint of practical applications (Pu et al., 2017; Jiang et al., 2015), but the systematic study of Ru-based catalysts is still very limited, especially in a broad pH range. To this end, detailed HER and OER measurements were carried out in electrolytes with different pH values using Ru-Ni NAs as the candidate catalyst. All electrochemical measurements were performed in a standard three-electrode system with a saturated calomel electrode as the reference electrode and a carbon rod as the counter electrode. The reference electrodes were calibrated before the electrochemical measurements (Figure S9). All polarization curves were recorded without iR compensation. Before the electrocatalytic measurements, all different Ru-Ni NAs were loaded on a carbon support (Vulcan XC72R carbon) by sonication. Ru loading of 20 wt % was maintained in each catalyst, and no obvious morphological changes were observed after heat treatment (Figure S10). The resulting Ru-Ni NAs/C were then dispersed in a mixture solvent containing isopropanol and Nafion (5%) and sonicated for 30 min to form a homogeneous catalyst ink. The concentration of the Ru-Ni NAs loading on the carbon powder was controlled at 2 mg mL−1; 10 μL catalyst ink was uniformly dropped onto a glassy carbon electrode and dried naturally at room temperature.The HER performance of the Ru-Ni NAs/C was first explored at a slow scan rate of 5 mV s−1 to ensure steady-state behavior on the electrode surface. To obtain the best performance of the Ru-Ni NAs/C in HER, we first determined the effects of the annealing temperature and atmosphere on HER performance by using Ru3Ni3 NAs as the candidate material. As shown in Figures S11A and S11B, the sample annealed at 250°C for 1 h exhibited the best HER activity in both alkaline and acidic conditions (0.5 M H2SO4 and 1 M KOH solutions). Figure 2 A shows the polarization curves of the Ru-Ni NAs and Ru NAs and commercial Pt/C in 1 M KOH. In detail, at a current density of 10 mA cm−2, the overpotentials of Ru3Ni3 NAs, Ru3Ni2 NAs, Ru3Ni1 NAs, Ru NAs, and commercial Pt/C were 39, 42, 44, 62, and 90 mV, respectively, versus the reversible hydrogen electrode (RHE), and the Ru3Ni3 NAs showed the smallest value. The Tafel slope is an intrinsic property of the catalyst that is determined by the rate-limiting step of the HER (Cherevko et al., 2016). Importantly, the Tafel slopes of the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs were calculated to be 26.9, 29.9, and 30.5 mV dec−1, respectively (Figures 2C and S12A). In contrast, the Ru NAs and commercial Pt/C showed relatively high Tafel slopes of 58.3 mV dec−1 and 46.8 mV dec−1. The electrocatalytic stability of the Ru3Ni3 NAs was further studied by both long-term cycling and chronopotentiometry tests, and the polarization curves of Ru3Ni3 NAs exhibited no obvious change after 12,000 cycles (Figure 2E). The Ru3Ni3 NAs showed only a slight potential increase after 10 h of chronopotentiometry at a current density of 5 mA cm−2 (Figure 2E, inset).With the change in the electrolyte to 0.1 M KOH, the Ru-Ni NAs still showed promising HER activities. At 10 mA cm−2, the overpotentials of the Ru3Ni3 NAs, Ru3Ni2 NAs, Ru3Ni1 NAs, Ru NAs, and commercial Pt/C were 119, 127, 123, 152, and 132 mV, respectively (Figure 2B). In addition to the low overpotentials, the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs also exhibited lower Tafel slopes than Pt/C (99.7 mV dec−1) and Ru NAs (76.0 mV dec−1) (Figures 2D and S12B). The Ru3Ni3 NAs also exhibited excellent durability after 12,000 cycles and in the chronopotentiometry test in 0.1 M KOH (Figure 2F), which indicated that the Ru3Ni3 NAs exhibit a superior HER activity and durability under alkaline conditions.The HER properties of the Ru-Ni NAs under acidic conditions were further investigated. Figure S13 shows that the overpotentials of the Ru3Ni3 NAs, Ru3Ni2 NAs, Ru3Ni1 NAs, and Ru NAs were 39 and 96 mV, 39 and 115 mV, 46 and 112 mV, and 55 and 122 mV at a current density of 10 mA cm−2 in 0.5 M H2SO4 and 0.05 M H2SO4, respectively. The Tafel slopes of the Ru3Ni3 NAs, Ru3Ni2 NAs, Ru3Ni1 NAs, and Ru NAs were 53.9 and 67.1 mV dec−1, 53.5 and 64.0 mV dec−1, 54.2 and 66.8 mV dec−1, and 81.6 and 79.6 mV dec−1 in 0.5 M H2SO4 and 0.05 M H2SO4, respectively. The Ru-Ni NAs showed a much better HER performance than the Ru NAs, indicating the vital role of Ni in improving the HER performance. After the working electrode was cycled for 6,000 cycles, the Ru3Ni3 NAs exhibited the best durability under acidic conditions with potential increases of only 62 and 39 mV in 0.5 M H2SO4 and 0.05 M H2SO4, respectively. In addition, after the 12-h chronopotentiometry test at 5 mA cm−2 in 0.5 M H2SO4 and 0.05 M H2SO4, the Ru3Ni3 NAs showed only potential increases of 36 and 49 mV, respectively (Figures S13E and S13F).The obtained Ru-Ni NAs were also successfully applied as efficient OER catalysts. Before the OER tests, the Ru-Ni NAs were also subjected to thermal annealing in air at different temperatures because Ru oxide has been discovered to be an active component for the OER (Petrykin et al., 2010; Reier et al., 2012). As shown in Figures S11C and S11D, the catalyst after heat treatment in air (350°C, 2 h) showed the best performance under both acidic and alkaline conditions (0.5 M H2SO4 and 1 M KOH). The TEM images show that the hierarchical structures were largely preserved (Figures S10C and S10D). We also studied the structural characterization of NAs after heat treatment by STEM image, elemental mapping, and line scan, where the core-shell structures of Ru3Ni3 NAs are largely reserved (Figure S14). We also showed that the carbon can enhance both the electrical conductivity and the dispersion of Ru3Ni3 NAs, and thus improve the electrocatalysis (Figure S15). To evaluate the OER performances of Ru-Ni under universal pH conditions, we tested the OER performances in both acidic (0.5 and 0.05 M H2SO4) and alkaline (1 and 0.1 M KOH) electrolytes. The commercial Ir/C catalyst was chosen as the reference because Ir is considered to be the benchmark catalyst for OER (Lettenmeier et al., 2016; Zhang et al., 2017c).Examination of the OER polarization curves in 0.5 and 0.05 M H2SO4 shows that the Ru-Ni NAs showed much better OER activities than the Ru NAs and commercial Ir/C. To drive a current density of 10 mA cm−2, the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs required overpotentials of 252 mV, 260 mV, and 268 mV in 0.5 M H2SO4, respectively (Figure 3 A). The Tafel slopes of the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs derived from Figure 3A were 45.8, 46.1, and 46.0 mV dec−1 in 0.5 M H2SO4, respectively. In contrast, the commercial Ir/C and Ru NAs required larger overpotentials of 328 and 277 mV in 0.5 M H2SO4, respectively. The Tafel slopes of the commercial Ir/C and Ru NAs were also larger than those of the Ru-Ni NAs (Figures 3C and S16A). Similar trends were also obtained in 0.05 M H2SO4, and the Ru3Ni3 NAs showed the lowest overpotential and Tafel slope of 312 mV and 70.8 mV dec−1, respectively (Figures 3B, 3D, and 6B).We further measured the OER activities in different alkaline electrolytes. The overpotentials of the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs were 304, 309, and 301 mV in 1 M KOH, whereas the Ru NAs and commercial Ir/C showed larger overpotentials of 351 and 311 mV, respectively (Figure S17A). The Tafel slopes of the Ru3Ni3 NAs, Ru3Ni2 NAs, Ru3Ni1 NAs, Ru NAs, and commercial Ir/C derived from Figure S15A were 91.7, 67.9, 73.4, 111.1, and 47.1 mV dec−1, respectively (Figure S17C). When the solution is replaced by a dilute alkaline solution (0.1 M KOH), in which it is more difficult for the OER to proceed (Lu and Zhao, 2015), the Ru-Ni NAs also exhibited a high activity. The overpotentials of the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs were 394, 390, and 384 mV, respectively, which were smaller than those of the Ru NAs (439 mV) and commercial Ir/C (407 mV) (Figure S17B). The Tafel slopes of the Ru3Ni3 NAs, Ru3Ni2 NAs, Ru3Ni1 NAs, Ru NAs, and commercial Ir/C derived from Figure S17B were 133.8, 131.4, 130.2, 140.7, and 111.1 mV dec−1, respectively (Figure S17D). All these results confirmed that the unique Ru-Ni NAs show excellent OER performances compared with the Ru NAs. In addition, in the 10-h chronopotentiometry test, the Ru3Ni3 NAs showed limited degradation after continuous electrolysis at 5 mA cm−2 in 0.5 M H2SO4, 0.05 M H2SO4, 1 M KOH, and 0.1 M KOH (Figures 3E, 3F,S17E, and S17F). No obvious morphological changes were observed in 0.5 M H2SO4 and 1 M KOH after the chronopotentiometry test (Figure S18), which demonstrated that the Ru-Ni NAs are indeed “acidic- and alkaline-stable” OER catalysts. To further demonstrate the OER and HER stability, chronopotentiometry test at higher current density was also performed, where the Ru3Ni3 NAs still showed limited degradations after continuous OER and HER electrolysis at 10 mA cm−2 in 0.5 M H2SO4 and 1 M KOH (Figure S19).As we explored the best catalysts for HER and OER under both acidic and alkaline conditions, a two-electrode setup with anodic catalyst Ru3Ni3 NAs after air treatment at 250°C for 1 h and cathodic catalyst Ru3Ni3 NAs after air treatment at 350°C for 2 h was used to study the potential application of Ru-Ni NAs in overall water splitting under universal pH conditions. The Linear Sweep Voltammetry (LSV) plots of Ru3Ni3 NAs and Ir/C-Pt/C under different pH conditions are presented in Figure 4 A. The data clearly show that both the potentials and Tafel slopes of the Ru3Ni3 NAs are much lower than those of Ir/C-Pt/C. The Ru3Ni3 NAs show an overpotential of 280 mV in 0.5 M H2SO4, which is considerably lower than that of Ir/C-Pt/C (370 mV). The Tafel slope of the Ru3Ni3 NAs is only 96.9 mV dec−1, whereas that of Ir/C-Pt/C is as high as 150.1 mV dec−1 (Figures 4B and S20A), indicating that the reaction kinetics of the Ru3Ni3 NAs are much faster than those of Ir/C-Pt/C. Significantly, the Ru3Ni3 NAs showed excellent durability with limited degradation after a 10-h chronopotentiometry test at 5 mA cm−2 in 0.5 M H2SO4, 0.05 M H2SO4, 1 M KOH, and 0.1 M KOH (Figure 4C). Overall, these results confirmed that the Ru-Ni NAs can serve as excellent water-splitting catalysts under universal pH conditions.It should be noted that both the HER and OER activities of the Ru-Ni NAs in different electrolytes are higher than those of most catalysts reported to date (Tables S1–S3). To explore the reasons behind the high performance, the surface structures of the different catalysts were first explored in detail. As shown in Figure S10, no obvious morphological changes were found in the Ru-Ni NAs after heat treatment. However, the XRD peaks assigned to RuO2 appeared in the catalysts processed at 350°C in air, and the Ru3Ni3 NAs showed the highest peak for RuO2 (Figure S21A). Considering that RuO2 plays an important role in enhancing the OER activity, the formed RuO2 greatly enhances the OER activity in the Ru-Ni NAs (Fang and Liu, 2010). XPS was also carried out to explore the surface properties of the Ru-Ni NAs. Figure S22 shows the full scan curves of the different Ru-Ni NAs, and the positions of the Ru and Ni peaks were consistent with the literature results (Folkesson et al., 1973). Furthermore, the XPS peaks of Ru in different catalysts after treatment at 350°C in air for 2 h were divided into Ru 3p3/2 and Ru 3p1/2 peaks, which can be further split into three peaks, corresponding to Rux+ (purple line), Ru4+ (orange line), and Ru0 (dark yellow line) (Figure 5 A) (Li et al., 2016). It was calculated that the Ru4+ fractions in the Ru3Ni3 NAs (57.00%), Ru3Ni2 NAs (44.46%), and Ru3Ni1 NAs (44.57%) were much higher than those in the Ru NAs (29.89%) (Table S4), which confirmed the higher concentrations of RuO2 in the Ru-Ni NAs. As shown in Figure 5B, the Ni 2p peaks in the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs were composed of Ni 2p1/2 and Ni 2p3/2 peaks, which both split into two oxidized Ni peaks, namely, Ni2+ (dark yellow line) and Ni3+ (orange line) (Zhang et al., 2007; Gong and Dai, 2015). Ni3+ is helpful for the formation of NiOOH on the catalyst surface, resulting in a better OER performance (Lee et al., 2012). This result indicates that both Ru and Ni in high oxidation states are generated in the Ru-Ni NAs by treatment at 350°C for 2 h in air, and that they are beneficial for the enhanced OER performance.Compared with the peaks of Ru-Ni NAs treated at 350°C for 2 h in air, no additional peaks were generated for the Ru-Ni NAs treated at 250°C for 1 h in air (Figure S21B). Based on XPS analysis, Ru can be successfully split into three peaks, namely, Rux+ (purple line), Ru4+ (orange line), and Ru0 (dark yellow line) (Figure S23). It was calculated that the area ratios of the metallic Ru0 were 59.74%, 56.77%, and 58.06% in the Ru3Ni3 NAs, Ru3Ni2 NAs, and Ru3Ni1 NAs, respectively, which were higher than 56.08% in the Ru NAs. (Table S5) and indicated a large number of active sites of metallic Ru present on the surface of the Ru3Ni3 NAs. Surface valence band XPS spectra were also obtained to determine the d-band centers of the Ru-Ni NAs treated at 250°C in air (Figures 5C and 5D). The d-band center downshifted with the increasing concentration of Ni. The reported d-band centers of Pt and Ru are located at −2.32 and −1.49 eV, respectively, corresponding to hydrogen binding energies of −0.32 and −0.64 eV, respectively, and suggesting that Ru shows a stronger hydrogen adsorption than Pt (Jiao et al., 2015). Pt is regarded as the best catalyst for HER performance owing to the suitable binding energy between the catalysts and adsorbates. Here, by alloying the catalyst with Ni, a downshift of the d-band center was observed in the Ru-Ni NAs (Figures 5C and 5D), which results in a suitable binding energy between the Ru-Ni NAs and adsorbates and boosted the HER activity of the Ru-Ni NAs (Stamenkovic et al., 2007).We further carried out density functional theory (DFT) calculations to elucidate how the downshift effect of the Ru-Ni NAs is related to the high performance of water splitting for both the OER and HER. The Ru-Ni NA system was modeled by a hexagonal lattice (hex-Ru-Ni) based on the Ru local symmetry. It shows a good metallic behavior with uniform isotropic conductivity across the Fermi level (EF) (Figure 6 A). The d-orbital projected density of states (PDOS) were compared and showed that Ni-3d downshifted to a value lower than that obtained for the bulk face-centered cubic Ni metal (Figure 6B) due to repulsion with the overlapping Ru-4d orbital, which implied a weakening in the Ni-O and Ni-H bonding. In addition, this downshifting effect appeared to be even more pronounced within the hexagonal local lattice than in the cubic lattice. Meanwhile, the Ru-4d states also downshifted compared with those in the hex-Ru metal, especially for the 4d-eg component above the EF (Figure 6B), regardless of the different local symmetries. This occurs because the eg-level component is essential for the adsorption of the bond of the p-π lone pair electrons in molecules such as H2O, O, or O2. This is because they almost remain in the non-bonding orbitals, and the adsorption stabilities are dominated by the Coulomb repulsion between 4d-eg in such p-π orbitals. Accordingly, the Ru in hex-Ru-Ni will easily transfer electrons between the catalysis substrate and intermediate molecules and facilitate O-O bond formation. The simulated OER pathway (Figure 6C) shows that the system is an energetically favorable catalyst even under U = 0 and U = 1.23 V, showing that water splitting with such Ru-Ni NAs would be a substantially low-barrier process. The splitting of H2O results in an increase in energy of 1.49 eV, guaranteeing that the initiation would be very reactive within a low overpotential. Meanwhile, there is no evident change in the energy for the evolution reaction [HO*+(H++e−)]→[O*+2(H++e−)] (∼0.4 eV). An additional similarly energetic increase (1.50 eV) was found for the formation of *OOH, indicating that the O* on the Ru-Ni still stays active to oxidize OH under lower overpotential. The splitting of H for the [HOO*+3(H++e−)]→[O2+4(H++e−)] transformation is very active. Compared with the pathway at U = 1.23 V, we confirm the overall overpotential (i.e., η = max{[barrier-1.23 eV]/e = 0.306 V}) is almost the same within the range of 0.200–0.300 V. Further calculations of the O2 dissociation confirmed that the combined O-O on the Ru-contained surface will be easily dissociated and enter into the surrounding solution conditions (Figure S24, Tables S6 and S7). Therefore, the OER on the Ru-Ni surface can achieve a very high performance supported by an energetic barrier-free water-splitting process. We further gain energetic insights on the alkaline HER. In the Ru-Ni surface system without partial oxidations by O-coverage, the alkaline HER performance overall is energetically downhill and the whole process gains a reaction heat of −0.48 eV with a small barrier of 0.16 eV. Activation barrier for the HER on this system may arise due to barrier of [H2O→H + OH]. As found by our experimental observation, partial oxidation states were found on the surface. We further conducted the reaction energy calculation. The overall reaction heat released is found to be −0.97 eV, showing it to be rather more energetically favorable than the case without oxidation. The process of [H2O→H + OH] is also energetically preferred gaining −0.28 eV during the bond cleavage on the partially oxidized Ru-Ni surface (Figure 6D). At the same time, a comparison of the chemisorption energies sheds light on the high HER/OER performance (Figure 6E). We also determined that the HER on the Ru-Ni system favors high H coverage with easy chemisorption of the 2H, and the formation of 2H→H2 is energetically favorable. Meanwhile, the low O coverage will easily facilitate water splitting and further accelerate further 2O chemisorption and O2 desorption. The kinetics of possible oxygen absorption or oxygen-related intermediates (OH−) is shown in absorption process in Figure 6F, which will result in the formation an intermediate distorted octahedral unit. The overlapping between eg orbital of Ru2+ and O-pσ orbitals will facilitate the ion transfer. The distorted structure prompts Ru2+ (d6) to change from a low-spin state (t2g 6eg 0) to an intermediate-spin state (t2g 5eg 1), where the eg1 can point to the intermediate with high bonding possibility. We also find that the absorption energy of further absorption on vertical oxygen molecule will be lowered nearly 1 eV, which can be attributed to the Jahn-Teller effect from the extra oxygen molecule to the c-axis of the distorted octahedral unit, which decreases the whole energy. Electrons on t2g can be further excited to eg and then form a high-spin state (t2g 4eg 2) with energy decrease. Overall, the Ru-Ni catalytic system is found to be efficient in HER performance from acidic to the basic condition. Thus, the Ru-Ni (NAs) system exhibits a high catalytic reactivity for water splitting based on the DFT calculations. We have also made a detail comparison for the preliminary absorption behavior on the cubic Ru-Ni (111) and hexagonal close packed (hcp) Ru-Ni (001) surface to elucidate the experimental treatment and related analysis. The discussions and analysis cover the following sections: energetics, electronic structures, orbital energetic behaviors, and adsorption analysis (Figures S24–S28 in Supplemental Information).In summary, for the first time, we have demonstrated a facile method for the synthesis of 3D Ru-Ni NAs, which leads to favorable 3D Ru-Ni superstructures with fully exposed active sites. The valence band spectra and DFT calculations revealed a change in the d-band center in the Ru-Ni NAs after the introduction of Ni, resulting in the transformation to a favorable surface environment for the OER and HER. The RuO2-decorated Ru-Ni NAs treated at 350°C in air provided additional active sites for the OER. The combined structural and electronic engineering leads to superior electrocatalytic performance for overall water splitting under universal pH conditions, and the performance is much better than that of the commercial Pt/C and Ir/C, demonstrating an unprecedented class of nanocatalysts with exceptional activity and excellent stability for electrochemical water splitting.Our work has demonstrated a novel bifunctional catalyst for water splitting in the universal environment from experimental and theoretical perspectives. Based on the combination of XPS and DFT as an effective approach, electronic environment modulation has been interpreted as the key factor that facilitates both HER and OER. However, an in-depth understanding of the oxidation states of the catalyst is still an open challenge because of the complex charge transfer induced by the overlap between the metal orbitals as well as the correspondingly accurate characterization. The site-to-site sampling and analysis of surface oxidation sites is of great significance for precise understanding of the catalyst reactivity. Therefore, we will keep working on further development and perfection on related theoretical exploration and advancement.All methods can be found in the accompanying Transparent Methods supplemental file.This work was financially supported by the Ministry of Science and Technology (2016YFA0204100, 2017YFA0208200), the National Natural Science Foundation of China (21571135), Young Thousand Talented Program, Jiangsu Province Natural Science Fund for Distinguished Young Scholars (BK20170003), the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD), and the start-up supports from Soochow University. B.H. thanks the support from the Natural Science Foundation of China (NSFC) for the Youth Scientist grant (Grant No.: NSFC 11504309, 21771156) and the Early Career Scheme (ECS) fund from the Research Grants Council of Hong Kong (Grant No.: PolyU 253026/16P).X.H. conceived and supervised the research. X.H., J.Y. and Q.S. designed the experiments. X.H., J.Y., and Q.S. performed most of the experiments and data analysis. X.H., J.Y., and Q.S. participated in various aspects of the experiments and discussions. B.H. and M.S. performed the DFT simulations. X.H., Q.S., J.Y., B.H., and M.S. wrote the paper. All authors discussed the results and commented on the manuscript.The authors declare no competing interests.Supplemental Information includes Transparent Methods, 28 figures, and 7 tables and can be found with this article online at https://doi.org/10.1016/j.isci.2019.01.004. Document S1. Transparent Methods, Figures S1–S28, and Tables S1–S7

Although electrochemical water splitting is an effective and green approach to produce oxygen and hydrogen, the realization of efficient bifunctional catalysts that are stable in variable electrolytes is still a significant challenge. Herein, we report a three-dimensional hierarchical assembly structure composed of an ultrathin Ru shell and a Ru-Ni alloy core as a catalyst functioning under universal pH conditions. Compared with the typical Ir/C-Pt/C system, superior catalytic performances and excellent durability of the overall water splitting under universal pH have been demonstrated. The introduction of Ni downshifts the d-band center of the Ru-Ni electrocatalysts, modulating the surface electronic environment. Density functional theory results reveal that the mutually restrictive d-band interaction lowers the binding of (Ru, Ni) and (H, O) for easier O-O and H-H formation. The structure-induced eg-dz2 misalignment leads to minimization of surface Coulomb repulsion to achieve a barrier-free water-splitting process.

There is an urgent need to develop highly efficient catalysts and technology to obtain butyl alcohol with their production based on local secondary resources of significant national economic importance.Butyl aldehyde, a limiting aliphatic aldehyde of the acyclic series, is obtained by the oxosynthesis of hydrocarbon raw materials. Its byproducts have long attracted the attention of researchers as a raw material – a starting object for the synthesis of new oxygen-containing organic compounds [1,2].In industry, aldehydes are obtained by hydroformylation (oxosynthesis) of unsaturated ethylene hydrocarbons of oil [3]. During the oxosynthesis of propylene, two isomers of butyl aldehydes are formed: Unlabelled Image One of the most effective methods of chemical processing of carbonyl-containing compounds is heterogeneous catalytic liquid-phase hydrogenation into the corresponding alcohols, which are in huge demand in the industry [4–9].Skeletal nickel catalyst, characterized by high activity and low cost, has been widely used in industry as well in several studies on the liquid-phase catalytic hydrogenation of oxygen- and nitrogen-containing organic compounds [10,11].To increase the stability and activity of catalysts in the hydrogenation of saturated aldehydes, promoting additives are employed [11].Analysis of the characteristics of industrial processes for the hydrogenation of butyl aldehyde, used in various countries, showed that each of these processes, along with advantages, has several disadvantages. Butyl alcohols are produced on an aluminum‑nickel‑titanium catalyst. However, the specified catalyst has low activity and productivity. The hydrogenation of butyl aldehyde at this contact occurs under severe conditions – high temperature (up to 170 °C) and pressure (up to 25 MPa), leading to a decrease in the process selectivity and an increase in the process cost [12,13].To address this problem, the search continues for effective catalytic systems and the selection of optimal technological parameters, conditions close to the environment (T, P), and a solution of harmless environment-friendly supercritical liquids for the hydrogenation of butyl aldehyde, considering the requirements of the principles of green technology [14].The authors studied promoted-skeletal nickel catalysts with the addition of platinum group metals [15,16]. They found that the rate of hydrogenation of carbonyl-containing compounds depends on the amount of promoter and passes through a maximum rate. Ruthenium is the most active catalyst in the hydrogenation of carbonyl compounds.The development of new selective modified catalysts for the hydrogenation of aldehydes contributes to improving the technology for the production of butyl alcohols, which are used as solvents for organic substances as well as for the synthesis of pharmaceuticals.The authors studied skeletal nickel [17,18] and supported [19–21] catalysts in the reactions of liquid-phase hydrogenation of carbonyl-containing compounds.In this study, the kinetic regularities of butyl aldehyde hydrogenation in solution at atmospheric pressure of hydrogen on skeletal Ni-Ru and Ni-Rh catalysts were investigated.This work aims to improve industrial processes for the selective hydrogenation of normal butyl aldehyde to primary butyl alcohol by developing new and modified heterogeneous catalysts considering various technological and economic requirements.Catalyst alloys were prepared in a high-frequency furnace. Nickel and aluminum granules and ruthenium or rhodium powder were used as metals for the preparation of alloy catalysts. To examine the rate of butyl aldehyde hydrogenation, alloyed nickel catalysts with the addition of ruthenium or rhodium were prepared. The amount of ruthenium or rhodium in the catalyst varied from 0.5% to 10%. The aluminum content in the initial alloy remained constant at 50%. The activation technique for promoted skeletal nickel catalysts consists of leaching 1.0 g of the alloy in a 20% NaOH solution at 100 °C for 1 h. The resulting catalyst was washed from alkali with distilled water to a neutral medium with respect to phenolphthalein.The n-Butyl aldehyde of the grade (chemically pure) TU 6-09-3828-74 was purified before the use by simple distillation and the purity of butyl aldehyde was controlled by chromatography. Distilled water was used as a solvent. The characteristics of the substances used are consistent with the reference data [22]. The hydrogenation reaction of butyl aldehyde was carried out in a glass thermostated duck-type reactor in water at the atmospheric pressure of hydrogen (700 bar) and a temperature of 20 °C [23]. To ensure that the reaction proceeds in the external kinetic region, the stirring speed was maintained at 600–700 unidirectional oscillations per min. The reactor was charged with 0.2 g of catalyst and 25 ml of solvent (H2O). Air was displaced from the reactor with a stream of hydrogen. The rocking chair was then switched on, and the catalyst was saturated with hydrogen at the temperature of the experiment for 30 min. Butyl aldehyde (0.5 ml) was added to a stream of hydrogen to the reactor. Catalytic activity was measured by the amount of hydrogen absorbed in one min. The state of the catalyst surface was monitored by the change in potential.The butyl aldehyde hydrogenation products were analyzed by gas-liquid chromatography (GLC) on a Chrom-4 chromatograph with a flame ionization detector in isothermal mode using a capillary column with a polar phase 50 m in length and 0.32 mm in inner diameter. The temperature in the column was maintained at 90 °C; temperature in the evaporation chamber was 200 °C; helium served as the carrier gas; the volume of the injected sample was 0.2 μl(microliter). Samples of the liquid reaction mixture were collected 2–3 times during the experiment [25]. GLC was based on the physicochemical separation of the analyzed components in the gas phase as they passed along a non-volatile liquid deposited on a solid sorbent. The GLC is caused as the components allow to separate and quantify substances in a complex mixture even if they are similar in chemical properties [26].The hydrogenation of primary aldehyde to alcohols takes place as per the following scheme: Unlabelled Image The catalytic properties of Ni-Ru and Ni-Rh catalysts in the butyl aldehyde hydrogenation reaction were studied in water at 20 °C. The preliminary results showed that the aldehyde hydrogenation rate slightly increases with an increase in its concentration in the solution by a factor of 2, and then remains constant. The hydrogenation process is limited by the activation of hydrogen. In all cases, hydrogen is consumed for the hydrogenation reaction, which is theoretically calculated. The kinetic curves consist of two sections. In the first section, a sharp decrease in the rate of hydrogen absorption was observed and in the second section, a smoother decrease in the rate occurred. Fig. 1 shows kinetic and potentiometric curves of butyl aldehyde hydrogenation on skeletal Ni-Ru and Ni-Rh catalysts in water. The ordinate axis shows the rate of butyl aldehyde hydrogenation (W), expressed in hydrogen milliliters absorbed per minute, and the potential of the catalytic system (E) is shown on the down of the ordinate axis. The abscissa axis shows the number of ml of hydrogen (V) absorbed during the reaction.The saturation potentials of nickel skeletal catalysts and promoted skeletal nickel catalysts are 700 mV and 650 mV, respectively. The ratio of the saturation potentials of the catalysts shown in Fig. 1 and Table 1 indirectly indicates the higher activity of Ni-Ru and Ni-Rh catalysts as compared to skeletal nickel catalysts (Niskel). The course of the kinetic curves shows that hydrogenation both on skeletal nickel and on promoted catalysts occurs at a decreasing rate. Varying the content in the alloy from 0.5% to 10% significantly increases the activity of the obtained catalysts. Table 1 shows the characteristics of Ni-Ru and Ni-Rh catalysts for the butyl aldehyde hydrogenation in water. Analysis of the data in the table shows that all the Ru and Rh additives increase the catalytic activity of the skeletal nickel catalyst during the butyl aldehyde hydrogenation by 1.7–1.9 times.When the hydrogenated substance is introduced into the reaction zone, the potential shift occurs, which indicates the competitive adsorption of hydrogen molecules and the hydrogenated aldehyde. The potential drop (△E) (as evident in Fig. 1 and Table 1) ranges from 105 to 120 mV. At the end of the hydrogenation reaction, the potential of the catalysts does not reach the saturation potential by 30–40 mV, due to irreversible adsorption by the reaction products on the surface of the catalysts.Notably, the character of the change in the potentiometric curves of the hydrogenation correlates with that in the process rate. To determine the comparative activity of alloyed nickel catalysts depending on the number of additives, the butyl aldehyde hydrogenation rate on the Ni-Ru and Ni-Rh catalysts was studied. This dependence is shown in Fig. 2 . The butyl aldehyde hydrogenation rate (W), expressed in ml of hydrogen absorbed per minute, is plotted up the ordinate axis, and the catalyst potential shift (△E) down the ordinate axis. The abscissa axis exhibits the amount of additive (Me / Ni-Al-Me, %) on the catalyst. Fig. 2 shows the catalysts' activity dependence after the absorption of 100 ml of hydrogen on the content of the addition of metals in the alloy during the butyl aldehyde hydrogenation in water. Fig. 2 shows that the optimal content of the additive in the alloy during the butyl aldehyde hydrogenation in the solvent (Н2О) is different. Thus, the maximum reaction rate on the Ni-Ru catalysts is observed at the contact with the content of 1.0% Ru. The butyl aldehyde hydrogenation rate on this catalyst is 1.9 times higher in water than in skeletal nickel without additives. The maximum activity of the Ni-Rh catalysts in water corresponds to 1.0% Rh in the alloy, the butyl aldehyde hydrogenation rate on this composition is 1.7 times higher than the corresponding value for skeletal nickel without additives. A further increase in the concentration of Ru and Rh in the alloy leads to a decrease in the activity of the contacts.An increase in the reaction rate is mainly accompanied by a decrease in the value of the anodic shift of the catalyst potential (Fig. 2), which characterizes a decrease in the degree of butyl aldehyde adsorption on the catalytic surface. These Ru and Rh additives help to better activate the reactants from the surface of the mixed catalyst. This conclusion is confirmed by the fact that the surface of promoted skeletal catalysts with an increase in the number of additives in the alloy changes slightly from 28 to 40 m2/g nickel.According to the data of X-ray diffraction studies, the promoting components have a significant effect on the composition and structure of the starting alloys and catalysts. The additives create, in addition to the phases usual for the alloy – NiAl3, Ni2Al3, and eutectics (NiAl3 + Al), new phases – Fх have not yet been deciphered. The ratio NiAl3 / Ni2Al3 in the promoted alloys is 1.25 times higher than in the Ni-Al alloy without an additive. Electron microscopic research methods indicate that the additives do not affect the crystal lattice parameter of nickel; however, these additives significantly grind nickel crystals (from 5.4 to 4.7 nm); increase the specific surface of the catalyst within 110–120 m2/g Investigations of the particle size distribution using optical microscopy and an electron microscope confirmed that Ru and Rh additives increase the fraction of dispersed catalyst particles, decrease diffusion inhibition in H2, leading to an increase in the reaction rate. The degree of its fatigue and deactivation depends on the ability of the catalyst to renew the initial concentration of adsorbed hydrogen and maintain it at a high level. This conclusion is supported by our data on the activation and stability of promoted catalysts.The hydrogen adsorption capacity was characterized using thermal desorption and conductometry methods.The conductometric method was used to characterize the state of hydrogen adsorbed by the catalysts. The promotion of skeletal nickel with Ru and Rh increases the fraction of weakly adsorbed hydrogen, leading to an increase in the binding energy of adsorbed hydrogen with its surface, thereby causing an increase in the total amount of adsorbed hydrogen. Large quantities of Ru and Rh in the alloy lead to a certain slowdown in the rate of hydrogen diffusion over the surface.Promotion on skeletal nickel leads to a decrease in the activation energy; the lowest value of the activation energy corresponds to the maximum activity of the catalyst, which indicates an increase in the binding energy of hydrogen with the surface. A similar conclusion was drawn from the calculation of the apparent activation energy of the butyl aldehyde hydrogenation reactions in the presence of promoted catalysts, which were 3–4 kcal/mol less than when the process was carried out on relatively non-promoted catalysts.To assess the amount and form of adsorbed hydrogen, the thermal desorption method was applied. The data show that the release of H2 from the surface of the catalysts under study begins at (−30 °C) and proceeds up to ≥800 °C; however, it proceeds at different rates. An increase in the rate of hydrogen evolution is noted in the temperature range: the first at (−20 °C); the second main desorption peak is located in the range 130–140 °C, and is characterized as a surface-adsorbed form of hydrogen. The third peak of desorption in the temperature range 220–230 °C characterizes the structural form of hydrogen.The introduction of Ru or Rh into the skeletal nickel catalyst is accompanied by an increase in the second main desorption peak and its shift to higher temperatures up to 170–180 °C, indicating a relative increase in the binding energy of adsorbed hydrogen with the catalytic surface. It confirms the conductometric measurements. The total volume of adsorbed hydrogen varies slightly from 45 to 47 ml/g.The results obtained are presented in Table 2 and indicate that the butyl aldehyde hydrogenation process on skeletal nickel is nonselective. The addition of Ru and Rh causes an increase in the selectivity index from 0.59 to 0.94. Comparison of the obtained data on thermal desorption measurements allows to observe a tendency to an increase in the selectivity index with the strengthening of the bond of adsorbed hydrogen with the catalyst surface.Analysis of the reaction products shown in Table 2 reveals that under the hydrogenation conditions, the butyl aldehyde not only turns into butyl alcohol but also undergoes destructive endothermic decomposition [23–25]. It should be noted that in the first minute, both butyl alcohol and propane are formed. In the beginning, active surface-adsorbed hydrogen takes part in the formation of butyl alcohol, after a slowdown in the reaction rate, strongly adsorbed hydrogen arriving at the catalyst surface from its depths. Propane is formed during the destructive endothermic decomposition of butyl aldehyde at the time of adsorption. Table 3 shows that the volume of H2 adsorbed by skeletal nickel is 17.6 ml/g, and the Н2 surface is 63.8 m2/g; for promoted skeletal nickel catalysts, these values increase: in the case of Ru – 26.4 ml/g and 96.0 m2/g; in Rh – 30.0 ml/g and 110.0 m2/g, respectively.This increase is due to an increase in the area on which energetically homogeneous hydrogen is adsorbed. This hydrogen may have a significant effect on the activity of catalysts in the hydrogenation of compounds, the rate of which is limited by the activation of hydrogen. Fig. 3a shows the obtained chromatogram of the butyl aldehyde hydrogenation products. Fig. 3b shows the simultaneous formation of butyl alcohol and propane, and the concentration of butyl alcohol reaches 87.1% (Table 2) by the time of the absorption of 0.75 mol of hydrogen. An intense increase in the concentration of butyl alcohol occurs after butyl aldehyde disappears from the solution. IR spectroscopic studies were carried out (Fig. 4 ) in order to establish the chemical structure of n-butyl alcohol [27].IR spectra were recorded on a Shimadzu IR Prestige-21 FT-IR spectrometer with a Miracle ATR attachment (Pike Technologies) in the frequency range 600–4000 cm−1.The method has been developed for the hydrogenation of butyl aldehydes under mild conditions, which helps to reduce energy costs. Novel efficient catalytic systems based on the alloyed nickel catalyst containing Ru or Rh additives have been created, which allow increasing the hydrogenation process rate by 1.7–1.9 times as compared to the skeletal nickel catalyst without additives. The hydrogenation of butyl aldehydes under mild conditions proceeds with high selectivity and stability ensures the quality of the target product and is of practical interest for improving the technology for the production of butyl alcohols.None.The study was conducted at the Chair of Oil Refining and Petrochemistry, M. Auezov South Kazakhstan State University, Shymkent, Kazakhstan.

The processes of heterogeneous catalytic hydrogenation of carbonyl-containing compounds into the corresponding alcohols mostly occur under rather harsh conditions; therefore, catalysts and optimal parameters that facilitate the process under mild conditions need to be selected. New ways of synthesizing catalytic systems and the use of harmless, environmentally friendly supercritical liquids as solvents are some of the key areas of green chemistry. At present, due to their high activity, two-component skeletal Ni catalysts are widely used in industry. This work attempts to develop highly efficient catalytic systems based on a three-component alloyed Ni catalyst containing metals (Ru, Rh) for the liquid-phase hydrogenation of aldehydes to butyl alcohols. The physicochemical and catalytic properties of modified alloyed Ni catalysts in the reaction of liquid-phase hydrogenation of butyl aldehyde to butyl alcohol have been investigated. Butyl aldehyde hydrogenation was carried out on a skeletal Ni catalyst promoted with ruthenium or rhodium in water at a temperature of 20–25 °C and the hydrogen pressure of 0.1 MPa. It was found that modified catalysts with high activity and selectivity reduce butyl aldehydes to butyl alcohols. The processes of catalyst preparation and butyl aldehyde hydrogenation are scalable and can be used in various processes of organic and petrochemical synthesis.

In recent years, CO2 hydrogenation is the most important process because it not only reduces global warming but also produces value-added products [1–5]. The hydrogen obtained from water electrolysis using renewable and sustainable energy sources such as solar, wind, geothermal, biomass etc. can be used to hydrogenate CO2 [6]. A variety of chemicals such as CO, CH4, CH3OH etc. can be obtained through catalytic hydrogenation of CO2 [7–9]. CO is a valuable precursor molecule that can be used as a raw material for the production of olefins, methyl alcohols and liquid hydrocarbons for the Fischer-Tropsch synthesis [10,11]. Various precious metals including Au, Pt, Pd, Rh and Ru [12–16] and non-precious metals Ni, Fe and Cu [17–19] loaded on various supports have been reported for the production of CO.Several metal catalyst have been reported to be active in CO2 methanation reaction including noble metal catalysts such as Pd, Ru and Pt [20–22] and non-noble metal catalysts such as Cu, Ni and Co [19,23,24] loaded on various supports such as Al2O3, SiO2, TiO2, CeO2 and ZrO2 [25–29]. NiO have been extensively investigated in various fields as catalysts [30,31], sensors [32–34] as well as batteries [35,36] due to their high catalytic activity, low cost and low toxicity. γ-Alumina is widely used as a catalyst [37–39], adsorbent [40] and catalyst support [41] due its thermal, chemical and mechanical stabilities. Compared with individual performances of NiO and Al2O3, NiO-Al2O3 composites are more efficient and have been used as catalysts [42–45], batteries [46,47], sensors [48,49] and adsorbents [50,51]. The non-metal phosphorus element has been proven to serve as a structure stabilizer of alumina [52–54]. Meanwhile, the incorporation of phosphorus into alumina brings about the decoration of hydroxyl groups on the surface of alumina accompanied by the formation of POH groups. The dehydroxylation between POH groups results in the formation of PO groups, which will rehydrate into POH groups in the presence of water and consequently reduce the adsorption of water on the adsorption sites of the catalysts, hence enabling the improvement of the hydrothermal stability for the NiO/Al2O3 catalysts. Moreover, the reducibility of metal-alumina have been found to be increased after doping phosphorus [55].In the present work, a series of 5 wt% NiO-xP-Al2O3 (x = 0, 5, 15, 20 wt%) catalysts have been prepared by a modified sol-gel method and investigated their performances in CO2 hydrogenation reaction. The results show that addition of phosphorus can significantly increase the catalytic activity. Compared with phosphorus unloaded 5 wt% NiO-Al2O3 catalyst, phosphorus loaded 5 wt% NiO-xP-Al2O3 catalysts exhibit superior performance.Nickel (II) nitrate hexahydrate (Ni(NO3)2.6H2O, ≥ 98.5 %) and pluronic P-123 (average Mn ∼5800) were purchased from Aldrich. Ammonium dihydrogen phosphate was purchased from Reanal. Aluminium isopropoxide (≥98 %) and acetic acid (99.7 %) were purchased from Sigma-Aldrich. Ethanol absolute (99.96 %) was purchased from VWR chemicals. Isopropanol (99.99 %) was purchased from molar chemicals. All chemicals were used as received without any further purification.P-loaded alumina was prepared via a modified sol-gel method. 1.50 g P123 (Mav = 5800) was dissolved in the mixed solution of 30 mL absolute ethanol, 10 mL isopropanol and 0.18 mL acetic acid at ambient temperature. Then, a required amount of ammonium dihydrogen phosphate and 0.015 mol of aluminium isopropoxide were introduced into the above solution under vigorous stirring for 4 h. The product was dried overnight at 80 °C and calcined in air at 500 °C for 4 h and then at 1000 °C for 1 h.The as-obtained P-loaded Al2O3 support was loaded with 5 wt% NiO by the incipient wetness impregnation method using Ni(NO3)2.6H2O aqueous solution as metal precursors. The product was dried overnight at 80 °C and calcined in air at 500 °C for 1 h. The final sample was marked as 5 wt% NiO-xP-Al2O3 (x = 0, 5, 15, 20 wt%).The specific surface area (BET method), the pore size distribution and the total pore volume were determined by the BJH method using a Quantachrome NOVA 2200 gas sorption analyser by N2 gas adsorption/desorption at −196 °C. Before the measurements, the samples were pre-treated in a vacuum at 200 °C for 2 h.XRD studies of all samples were performed on a Rigaku MiniFlex II instrument with a Ni-filtered Cu-Kα source in the range of 2θ = 20−80°.Imaging of the all the samples were carried out using a FEI TECNAI G2 20 X-Twin high-resolution transmission electron microscope (equipped with electron diffraction) operating at an accelerating voltage of 200 kV. The samples were drop-cast onto carbon film coated copper grids from ethanol suspension.The temperature-programmed reduction (TPR) was carried out in a BELCAT-A analyser using a reactor (quartz tube with 9 mm outer diameter) that was externally heated. Before the measurements, the 50 mg of catalyst was pre-treated in oxygen at 400 °C for 30 min and in N2 at 400 °C for 15 min. Thereafter, the sample was cooled in flowing N2 to 50 °C. The oxidized sample was flushed with N2 containing 10 % H2, the reactor was heated linearly at a rate of 5 °C/min from 50 °C to 500 °C and the H2 consumption was detected by a thermal conductivity detector (TCD).The chemical states (and the atomic ratio of the elements) were investigated by X-ray photoelectron spectroscopy (XPS). The SPECS instrument was equipped with a Phoibos150 MCD-9 analyser. The Al Kα x-ray source was operated at 14 kV and 10.8 mA (150 W) and the analyser was used in FAT mode with a pass energy of 20 eV in the case of high-resolution spectra. CasaXPS software was used for data evaluation. The binding energy was set to the adventitious carbon C1 s peak is at 284.8 eV. The peaks of the P2p were fitted with single due to closely spaced spin-orbit coupling. Specs FG 15/40 flood gun was operated during the accumulation at 0.6 V and 0.3 mA to the prevent the charging of the surface of the sample.Before the catalytic experiments in a continuous-flow reactor the as-received catalysts were oxidized in the O2 atmosphere at 300 °C for 30 min to remove the surface contaminants and thereafter were reduced in H2 at 300 °C for 60 min. Catalytic reactions were carried out at atmospheric pressure in a fixed-bed continuous-flow reactor (200 mm long with 8 mm i.d.), which was heated externally. The dead volume of the reactor was filled with quartz beads. The operating temperature was controlled by a thermocouple placed inside the oven close to the reactor wall, to assure precise temperature measurement. For catalytic studies, small fragments (about 1 mm) of slightly compressed pellets were used. Typically, the reactor filling contained 150 mg of catalyst. In the reacting gas mixture, the CO2: H2 molar ratio was 1:4, if not denoted otherwise. The CO2: H2 mixture was fed with the help of mass flow controllers (Aalborg), the total flow rate was 50 mL/min. The reacting gas mixture flow entered and left the reactor through an externally heated tube in order to avoid condensation. The analysis of the products and reactants was performed with an Agilent 6890 N gas chromatograph using HP-PLOTQ column. The gases were detected simultaneously by thermal conductivity (TC) and flame ionization (FI) detectors. The CO2 was transformed by a methanizer to methane and it was also analyzed by FID. CO2 conversion was calculated on a carbon atom basis, i.e. C O 2   c o n v e r s i o n % = C O 2   i n - C O 2   o u t C O 2   i n × 100 % CH4 selectivity and CO selectivity were calculated as following C H 4   s e l e c t i v i t y % = C H 4   o u t C O 2   i n - C O 2   o u t × 100 % C O   s e l e c t i v i t y % = C O   o u t C O 2   i n - C O 2   o u t × 100 % where C O 2   i n and C O 2   o u t represent the C O 2 concentration in the inlet and outlet respectively, a n d   C H 4   o u t   a n d   C O   o u t   r e p r e s e n t   t h e   a m o u n t   o f   f o r m e d   C H 4   and CO, respectively. Fig. 1 presented the XRD pattern of all the catalysts. The diffraction peaks observed at 2θ = 21.9°, 28.2°, 31.1° and 35.8° are assigned to diffraction planes (111), (021), (112) and (220) of AlPO4-tridymite crystal structure (JCPDS No. 11-0500) with no impurity phases [56]. The small peak at 20.6° may be due to the presence of structural defects [57]. The peak intensity increased with phosphorous content, suggesting the growth of AlPO4-tridymite crystal phase. All the phosphorus loaded samples exhibited no characteristic diffraction peaks related with α-Al2O3 and NiO phases. 5 wt% NiO-Al2O3 catalyst exhibited main peaks at 2θ of 19.8°, 32.7°, 37.3°, 39.5°, 45.6°, 60.6°, 66.9° are assigned to diffraction planes (111), (220), (311), (222), (400), (511) and (440) of cubic γ-Al2O3 crystal phase (JCPDS No. 29-0063) [58]. There is no obvious signal of NiO phase detected in the XRD patterns of 5 wt% NiO-Al2O3 catalyst except a low intense peak at 2θ = 43.4° as a result of overlapping with the γ-Al2O3 peaks and low NiO content.The BET surface area, pore volume and pore size of the catalysts were summarized in Table 1 . All the catalysts showed type H3 hysteresis loop indicating the presence of slit shaped pores. The pore size distribution reveals the existence of mesopores in the range 2–9 nm. It is clear that the introduction of phosphorus into the 5 wt% NiO-Al2O3 results in considerable decrease in the surface area and pore volume with 22.67 m2/g and 0.06 cm3/g respectively for 5 wt% NiO-20P-Al2O3 in comparison with 100.52 m2/g and 0.26 cm3/g for 5 wt% NiO-Al2O3. This can be attributed to the blocking of pores of 5 wt% NiO-Al2O3 after phosphorus loading.The morphology and particle size of some of the catalytically active catalysts were examined by TEM measurements and shown in Fig. 2 . All the catalysts show nanoobjectives with porous structures with a building blocks of spherical shaped morphology with the size of ∼2−15 nm.The reducibility of the catalysts was studied by the H2-TPR technique and the results are shown in Fig. 3 . 5 wt% NiO-Al2O3 catalyst exhibited two reduction peaks. The low temperature peak at 406 °C was attributed to reduction of NiO not bound with Al2O3 and high temperature peak at 594 °C to the reduction of NiO that weakly interacted with Al2O3 [59–62].When phosphorus was added into the NiO-Al2O3 catalyst, there is a shift in reduction peaks of NiO towards lower temperature indicates that the addition of phosphorus facilitates the reduction of NiO. This observation corroborated well according to the previous work reported over Pd/xP-OMA catalyst [63]. The total H2 consumptions have been summarized in Table 2 .The catalytic performances of all the 5 wt% NiO-xP-Al2O3 (x = 0, 5, 15, 20) catalysts were evaluated at atmospheric pressure in the temperature range from 200 °C to 600 °C. The CO2 conversion is shown in Fig. 4 a. CO2 conversion increased with temperature raising from 200−600 °C. All the catalysts reach the highest conversion at 600 °C with 61.54 %, 62.89 %, 63.88 % and 66.13 % respectively for 5 wt% NiO-Al2O3, 5 wt% NiO-5P-Al2O3, 5 wt% NiO-15P-Al2O3 and 5 wt% NiO-20P-Al2O3. The CO2 conversion decreases in the order 5 wt% NiO-20P-Al2O3 > 5 wt% NiO-15P-Al2O3 > 5 wt% NiO-5P-Al2O3 > 5 wt% NiO-Al2O3.The main products identified during the catalytic reaction were CO and CH4. The CH4 and CO selectivity results are shown in Fig. 4b and c respectively. CH4 selectivity also depends on phosphorus loading, with 5 wt% NiO-xP-Al2O3 (x = 5, 15, 20) catalysts forming high CH4 than 5 wt% NiO-Al2O3 catalyst. The CH4 selectivity is high at low temperature and decreases to 64.2 %, 58.8 % and 56.7 % for 5 wt% NiO-20P-Al2O3, 5 wt% NiO-15P-Al2O3 and 5 wt% NiO-5P-Al2O3 catalysts respectively at 450 °C and then increased and decreased. The CO selectivity first increases then decreases and then again increases as the temperature increases. 5 wt% NiO-Al2O3 catalyst displayed higher CO selectivity. Gao et al. reported that, above 450 °C the formation of CO increases as a result of reverse water gas shift reaction while the CH4 formation decreases as a result of exothermic nature of CO2 methanation [64]. Fig. 5a shows the Arrhenius plot for CO2 hydrogenation over all the 5 wt% NiO-xP-Al2O3 (x = 0, 5, 15, 20) catalysts. The apparent activation energies were calculated from the slopes in the temperature range of 350−450 °C resulting in values of 56.3 kJ mol−1 for 5 wt% NiO-Al2O3, 68.3 kJ mol−1 for 5 wt% NiO-5P-Al2O3, 49.1 kJ mol−1 for 5 wt%NiO-15P-Al2O3 and 45.5 kJ mol−1 for 5 wt% NiO-20P-Al2O3. The apparent activation energy decreases with increase in phosphorus loading. 5 wt% NiO-20P-Al2O3 catalyst had the lowest apparent activation energy and this is in line with the catalytic activity. Fig. 5b shows the time on stream results of all the catalysts for CO2 hydrogenation at 600 °C. It can be seen that all the catalysts exhibit almost similar stability indicates that phosphorus addition does not improve the stability of the catalysts.The used catalysts have been characterized by XRD in order to identify possible structural changes (Fig. 6 ). Besides NiO and AlPO4 form, the XRD patterns revealed new peaks at 2θ = 44.5°, 51.8° and 76.3° for metallic nickel (JCPDS No. 65-2865) [65]. This metallic nickel is supposed to be formed during the pre-treatment with H2 prior to the catalytic reaction and is responsible for H2 dissociation during CO2 hydrogenation.For better understanding of the outstanding activity of phosphorous loaded catalysts, XPS studies were performed on the spent catalysts. The XPS was measured to study the chemical composition and valence states of the surface of the used 5 wt% NiO-xP-Al2O3 (x = 5, 15, 20) catalysts. The core level Ni 2p spectrum is shown in Fig. 7 a. The peak at binding energy of 856.5 eV is ascribed to Ni 2p3/2 and another peak at binding energy of 874.4 eV is ascribed to Ni 2p1/2. The two relatively weak peaks at 861.9 eV and 880.7 eV belong to shakeup satellites indicating the presence of NiO [66]. The small shoulder peak at 852.7 eV is ascribed to the presence of metallic nickel [67]. The peak at 74.4 eV is ascribed to the presence of either Al2O3 or AlPO4 [68]. However, based on the XRD results of the used catalyst, it could be expected as AlPO4. Fig. 7b. shows P 2p XPS spectra. The peak at 134.4 eV is ascribed to the presence of AlPO4 [69]. The small peak at 130.3 eV can be assigned to the Ni2P/Ni5P12 arising from the interaction of nickel with phosphorus [70]. This indicates the existence of Ni/NiO/Ni2P/Ni5P12/AlPO4 interfacial species in the catalyst during the reaction and this interfacial species increased with phosphorus loading and can be correlated with the higher catalytic activity.The formation ratio of the metal-phosphide is relatively low ∼3−5 % (Table 3 ) and this atomic concentration is decreasing with the rising of the phosphate content. However, the nickel enrichment in the surface layer presumable in Ni2P/Ni5P12 form is very likely according to the P 2p spectra. The authors assume that could be responsible for the enhanced catalytic activity. In the C 1s XPS spectra, three peaks were identified. The first peak at 284.78 eV is assigned to the CC and CH hydrocarbon and the second peak at 286.48 eV is assigned to COH arising from surface contamination. The third peak at 288.84 eV are assigned to CO and OCO species, which can be attributed to a surface contamination component or a solvent degradation product. In the O 1s XPS spectra, two peaks were identified. The first peak at 531.9 eV is assigned to CO and the second peak at 530.63 eV is assigned to the lattice oxygen, OCO [71]. The corresponding binding energies and atomic percentages as well as the phosphate and phosphide ratio according to the deconvoluted XPS spectra are reported in Table 3.5 wt% NiO-xP-Al2O3 catalysts were prepared by a modified sol-gel method. The effect of phosphorus addition on NiO-Al2O3 catalyst in CO2 hydrogenation was investigated. The BET surface area decreased with increasing phosphorus doping. The phosphorus loaded NiO-Al2O3 exhibited enhanced performance towards CO2 hydrogenation compared with pure NiO-Al2O3. Ni/NiO/Ni2P/Ni5P12/AlPO4 interfacial species were detected as active species on the used catalysts by X-ray photoelectron spectroscopy. The formation ratio of the metal-phosphide is relatively low ∼3−5 %, and this atomic concentration is decreasing with the rising of the phosphate content. However, the nickel enrichment in the surface layer presumable in Ni2P/Ni5P12 form is very likely according to the P 2p spectra and the authors assume that could be responsible for the enhanced catalytic activity. This work will not only help in understanding the role phosphorus in CO2 hydrogenation reaction but also provide insights for future design and development of high performance phosphorus loaded catalysts.The authors have nothing to declare about Credit of Statement.The authors declare no competing financial interest.This paper was supported by the Hungarian Research Development and Innovation Office through grants NKFIH OTKA PD 120877 of AS. ÁK, and KZ is grateful for the fund of NKFIH (OTKA) K112531 & NN110676 and K120115, respectively. The financial support of the Hungarian National Research, Development and Innovation Office through the GINOP-2.3.2-15-2016-00013 project "Intelligent materials based on functional surfaces - from syntheses to applications" and the Ministry of Human Capacities through the EFOP-3.6.1-16-2016-00014 project and the 20391-3/2018/FEKUSTRAT are acknowledged.

A series of 5 wt% NiO-xP-Al2O3 with different phosphorus loading contents (x = 0, 5, 15 and 20 wt%) were prepared by a modified sol-gel method. A significant promotional effect of phosphorus on NiO-Al2O3 in CO2 hydrogenation is observed. All the catalysts reach the highest conversion at 600 °C with 61.54 %, 62.89 %, 63.88 % and 66.13 % respectively for 5 wt% NiO-Al2O3, 5 wt% NiO-5P-Al2O3, 5 wt% NiO-15P-Al2O3 and 5 wt% NiO-20P-Al2O3 catalysts. Ni/NiO/Ni2P/Ni5P12/AlPO4 interfacial species were detected on the surface as active species on the used catalysts by X-ray photoelectron spectroscopy. The formation ratio of the metal-phosphide is relatively low ∼3−5 %, and this atomic concentration is decreasing with the rising of the phosphate content. However, the nickel enrichment in the surface layer presumable in Ni2P/Ni5P12 form is very likely according to the P 2p spectra and the authors assume that could be responsible for the enhanced catalytic activity.

With the increasingly serious global environmental degradation and energy crisis, renewable energy techniques have attracted considerable attention of scientific researchers [1–4]. Among numerous technologies for energy storage and conversion, over water splitting, which can achieve by green photocatalysis or electrocatalysis, has become a star because of its advantages of friendly environment, rich resources, and strong sustainability. However, the water splitting mode still faces many unavoidable barriers on the path to the large-scale practical application [5–7]. Some of the most difficult aspects consist in its oxygen evolution reaction (OER) as a half reaction, which is coupled with hydrogen evolution reaction (HER) for water splitting [8,9]. Since the OER process has an innate four-electron transmission mechanism different from the two-electron one in HER, the complex reaction paths slow down the electrocatalytic kinetics and the inferiority of energy barriers is more prominent for over water splitting [10]. The most straightforward approach to surmounting energy penalty is to optimize the design of catalyst. However, although noble metal catalysts (e. g., Ir- and Ru-based catalysts) have appropriate theoretical free energy, the well-known scarcity and high cost limit their large-scale application, so the research trend is biased towards the non-noble substitutes [11]. As reported, extensive research focuses on various transition metal-based oxides, hydroxides, sulfides, phosphates and selenides, etc [12–15]. These semiconductors can be used as excellent OER catalysts without exception since their following characteristics: outstanding inherent activity, high density of active sites, excellent electrical conductivity, and rapid transport of reaction species.Among various transition metal-based catalysts, nickel-containing catalysts undoubtedly garner focused attention because of their earth-abundant superiority and potential in the field of water oxidation [16,17]. Meanwhile, the Ni(OH)2 as a common member has been constantly optimized to improved OER catalytic performance in wide researches, including different preparation methods (hydrothermal [13], solvothermal [18], electrodeposition [19], etc. methods), multifarious modification for distinctive morphologies (hollow sphere [18], nanoplate [20], nanoparticle [18], etc.) and particular phase-dependent superiority (α or β phase) [21,22]. Although these studies give prominence to the advantages of Ni(OH)2 in their increasing inherent activity, there is still a need to design novel Ni(OH)2 catalysts with more excellent electronic conductivity and long-term stability through structural optimization.To our knowledge, building the one-dimensional, porous, and hollow characteristics is ideal optimized modification to serve high-performance electrocatalyst candidates. Because they can endow catalysts with high surface roughness, excellent permeability, large surface area (active site density) and plenty channels to meet the deep-seated chemical accessibility between electrolyte and electrode, as well as the rapid charge and mass transport. Moreover, the resulting low density will also save resource consumption [23–26]. Therefore, preparing tubular Ni(OH)2 with one-dimensional hollow structure will further improve its catalytic performance.The template-assisted method is a reliable and available method to prepare one-dimensional hollow materials, and when considering the formation of Ni(OH)2 with these unique morphologies, Cu2O is a promising choice as sacrificial template [27–30]. As we know, the Cu2O are perfect to be used as hard template for the preparation of hollow structures, because of its low cost, abundant reserves, reachable control of morphology and size. So various hollow or core-shell structures have been widely reported, such as copper sulfide [31], mental alloy [32], metallic oxide/hydroxide cages [33–36]. Among the Cu2O pre-shaped templates with various morphologies, Cu2O wires with the simple preparation are especially ideal to serve the hollow tubular construction. Here, Pearson's hard and soft acid–base (HSAB) principle, namely soft Lewis acids tend to combine with soft bases while hard acids prefer hard bases, provides a new inspiration for Cu2O-templated strategy to form hollow transition metal hydroxides/oxides, which always act as superior electrocatalysts and exhibit enhanced ionic conductivity and robust stability [33–36]. In addition, theoretically, choosing an appropriate etchant (like Na2S2O3) to react with Cu2O can realize the synthesis of hydroxide and the etching of Cu2O in one step, which will be a simple and feasible liquid-phase preparation strategy of hydroxide with hollow structure. Therefore, it is of great significance to develop a liquid-phase synthesis strategy with Cu2O wires as templates to form Ni(OH)2.Herein, we firstly propose a synthesis strategy of Ni(OH)2 hollow tubes (Ni(OH)2 HTs) with the above-mentioned hollow, porous and one-dimensional structural features, in which the etching of as-prepared Cu2O wires sacrificial template and the formation of Ni(OH)2 occurred simultaneously. The obtained Ni(OH)2 HTs were confirmed with immensely enlarged specific surface area to create accessible active sites, numerous charge transfer channels for fast kinetics, and a stable structural foundation inherited from Cu2O wires for long-term stability. Therefore, it displayed excellent electrochemical performance when used as OER catalyst, with a low overpotential of 207 ​mV to reach 10 ​mA ​cm−2 current density, an ideal Tafel slope of 79.8 ​mV dec−1, and undiminished sustainability of current density for 24 ​h. The innovative avenue for preparing hollow one-dimensional transition metal hydroxide in this work will open a new approach for the design of electrocatalysts with distinctive structure-activity relationship.All the reagents in this paper were utilized as received without further purification and had analytic purity grade. Polyvinyl pyrrolidone (PVP), pyrrole and Cu(Ac)2 were purchased from Macklin Biochemical Co., Ltd. NiCl2·6H2O, Na2S2O3·5H2O and ethanol were purchased from Sinopharm Chemical Reagent Co., Ltd. KOH was purchased from Aladdin Reagent Co. Ltd.According to the hydrothermal method previously reported [37], the preparation process was further specifically clarified as follows. Typically, a 30 ​mL deionized (DI) aqueous solution of 100 ​mg Cu(Ac)2 was prepared. Under 40 ​°C water bath, 67 ​mg pyrrole was dissolved in 10 ​mL DI water (0.1 ​M), and then the pyrrole solution was added to the Cu(Ac)2 solution with continuous stirring for 30 ​min. Afterward, the mixture was transferred into a 50-mL Teflon-lined stainless steel autoclave and maintained at 180 ​°C for 10 ​h. When cooled down to ambient temperature, the precipitate was collected by centrifugation and washed several times with DI water and ethanol. Then, the product was dried at 50 ​°C in vacuum for 24 ​h.Firstly, 333 ​mg PVP was dispersed in 10 ​mL DI water/ethanol solvent with volume ratio of 1:1. Then 5 ​mg as-prepared Cu2O wires and 6.66 ​mg NiCl2·6H2O were dissolved into the PVP solution, following by ultrasonication for 20 ​min and stirring for 40 ​min. Afterward, 4 ​mL of 1 ​M Na2S2O3·5H2O aqueous solution was prepared under ice bath and added dropwise into the above mixture solution with vigorous stirring. After dropping and a continuous stirring for 30 ​min, the final product was collected by centrifugation, washed several times with DI water and ethanol, and finally dried at 50 ​°C in vacuum for 24 ​h.The compositions, structural features and chemical states of all samples were identified using X-ray diffraction of X'Pert Pro MPD with Cu Kα radiation and λ as 0.154 ​nm (XRD, Nalytical, the Netherlands), X-ray photoelectron spectroscopy of PHI 5000 VersaProbe III (XPS, Ulvac-Phi, Japan), transmission electron microscopy of JEM-2200FS with a 200 ​kV accelerating voltage energy (TEM, JEOL Japan) equipped with energy dispersive X-ray spectroscopy (EDX), scanning electron microscope of SUPRA 55 (SEM, ZEISS, Germany), and nitrogen adsorption/desorption isotherm of Autosorb iQ (Quantachrome, USA).The electrochemical measurements were carried on a CorrTest electrochemical workstation and in a standard three-electrode configuration (about 25 ​°C room temperature). The Ag/AgCl, carbon rod and sample-modified nickel foam were used as reference, counter and working electrode, respectively. The slurry was prepared by mixing the catalyst and acetylene black in a 4:1 ​wt ratio in 500 ​μL ethanol and 20 ​μL Nafion. Then the slurry was pasted uniformly onto a Ni foam (1 ​cm ​× ​1 ​cm) and dried in vacuum to obtain a working electrode with a catalyst mass of 4 ​mg. The tested potentials were converted into reversible hydrogen electrode (RHE) scale via Nernst equation: ERHE ​= ​EAg/AgCl ​+ ​0.0591 ​pH ​+ ​0.1989. The electrochemical impedance spectra (EIS) were recorded in frequency of 0.1–105 ​Hz with an alternating current potential amplitude of 5 ​mV. The 1 ​M KOH was used as electrolyte. A constant 1.484 ​V (vs RHE) potential was exerted for chronoamperometric measurement. All electrochemical test results were processed without IR correction.In Fig. 1 , the scheme shows the fabrication process of Ni(OH)2 HTs, which was inspired by Pearson's HSAB principle. At the beginning, the as-prepared Cu2O wires were successfully synthesized through a previously reported hydrothermal method [37], which played the role of sacrificial templates for the further formation of Ni(OH)2 HTs. According to HSAB principle, since Cu+ possesses soft acid feature, the soft base ligand, S2O3 2− was chosen as the coordinating etchant to facilitate a soft-soft interaction, which is more stable than a soft-hard one between Cu+ and O2− existing in Cu2O. Thus, in the step (A) in Fig. 1, the following chemical reactions occurred [36]: Cu 2 O + x S 2 O 3 2 - + H 2 O → [ Cu 2 ( S 2 O 3 ) x ] 2 - 2 x + 2 OH - S 2 O 3 2 - +  ​ H 2 O →  ​ HS 2 O 3 - +  ​ OH - Ni 2 + +  ​ 2 OH -  ​ →  ​Ni ( OH ) 2 In the above process, while the surface of Cu2O wires were etched, the OH− derived from hydrolysis and etching effect of S2O3 2− was combined with Ni2+ to form Ni(OH)2 clinging to the surface of Cu2O. Then, due to the complete depletion of Ni2+, only the etching of Cu2O took place in the next step (B), that was, the formation of Ni(OH)2 hollow structure.The composition and morphology of Cu2O were characterized in detail. As shown in Fig. 2 a, the XRD pattern of Cu2O wires templet indicates that the main diffraction peaks could be assigned to cubic Cu2O (PDF #78–2076), but there were also several weak peaks corresponding to Cu phase (PDF #85–1326), which may be caused by the excessive reduction of pyrrole as reducing agent. The SEM image (Fig. 2b) suggests that the length and diameter of Cu2O wires were not identical from each other (about dozens of microns in length and 90–600 ​nm in diameter), but these wires had identical smooth surface, solid interior and single crystal property which reflected in the TEM image and corresponding selected area electron diffraction (SAED) pattern (Fig. 2c and its inset). Then, the distinct lattice fringe in HRTEM of Fig. 2d exhibited an interplanar distance of about 0.21 ​nm in accordance with the (2 0 0) plane of cubic Cu2O. In addition, the homogeneous distribution of Cu and O elements on EDX mapping (insets in Fig. 2d) and the appropriate atomic ratio of Cu/O (Figs. S1 and 61.69/38.31) all indicated the successful preparation of Cu2O wires.Similarly, the crystallographic and morphologic structure of Ni(OH)2 HTs transformed from Cu2O wires was characterized. In the XRD spectrum (Fig. S2a), there were several sharp diffraction peaks corresponding to Cu2O that had not been completely etched, and the characteristic peak positions of the prepared sample located at the rhombohedral α-Ni(OH)2·0.75H2O (PDF #38–0715), indicating the successful synthesis of the final product as Ni(OH)2·0.75H2O. The TEM and SEM images (Fig. 3 a and Fig. S2b) show that Ni(OH)2 tubes retained the wire-like appearance of Cu2O with nonuniform sizes and had the diameter range of 90–620 ​nm. Furthermore, the more detailed TEM image of a broken Ni(OH)2 HT further proves the hollow structure (about 17 ​nm in thickness of tube wall) and significantly rougher surface of the etched Ni(OH)2 tubes when compared with the Cu2O wires template, which both are essential conditions to improve catalytic performance. In Fig. 3b, the interplanar distances of approximate 0.20 ​nm were observed from HRTEM image corresponding to (0 1 8) plane of Ni(OH)2·0.75H2O, and the polycrystalline nature could be also testified by the SAED pattern as the inset. Besides, in EDX mapping (Fig. 3c), a Cu2O wire that was not completely etched (red wireframe 1) and a final as-prepared Ni(OH)2 hollow tube (red wireframe 2) were observed at the same time, which could be proved by elements distribution (Cu, Ni, O) and their atomic ratio (Fig. S3, 5.02/32.58/62.40). Concurrently, the above results were also consistent with the XRD result in Fig. S2a.The most obvious benefit of hollow structure and rough surface for catalysis is a substantial increase of its specific surface area. Therefore, the correlative analyses were carried out to the Cu2O wires template and the final Ni(OH)2 HTs. The N2 adsorption–desorption isotherm of the Ni(OH)2 HTs (Fig. 4 a) confirms a classic type Ⅳ isotherm, which has an apparent hysteresis loop at 0.1-1.0 ​P/P0 and a part of decreasing slope at initial relative pressure range. That is, the as-prepared Ni(OH)2 HTs possessed mesoporous microstructure. And expectedly, an outstanding specific surface area of 63.1 ​m2 ​g−1 for Ni(OH)2 HTs was calculated by Brunauer–Emmett–Teller (BET) method. As for the pore size distribution (Fig. S4a), the Ni(OH)2 HTs exhibited a mostly centralized distribution that located at 3.05 ​nm, meaning the major existence of mesopores in Ni(OH)2 HTs. As a comparison, the Cu2O wires template was also tested by the same ways (Figs. S4b and c), which gave a calculated BET specific surface area of only 6.2 ​m2 ​g−1. Evidently, the tenfold enlargement of high specific surface area and the transformation to porous structure are sufficient to illustrate the importance of template method and the advantages of structural modification, which make more approaches for reactants to enter the active material and provide more active sites for the next electrochemical reaction.Then, the XPS was employed to further understand the surface information about elemental composition and the electronic structure of Ni(OH)2 HTs sample. As shown in survey spectrum (Fig. 4b), the observation of Cu 2p, Ni 2p and O 1s peaks with the respective atomic ratio (table in Fig. 4b) corroborated component analysis referred above, meaning the successful synthesis of Ni(OH)2 HTs with few residual elemental Cu. In the high-resolution XPS spectrum of Ni 2p (Fig. 4c), accompanied by two shakeup satellite peaks (denoted as “Sat.”) at 880.0 and 861.7 ​eV, two intensive peaks that located at 873.8 and 856.1 ​eV were conformed to Ni 2p1/2 and Ni 2p3/2, which were good consistent with Ni2+ and further proved expected synthetic Ni(OH)2. The O 1s XPS spectrum can be deconvolved into three peaks (Fig. 4d). The peaks at 532.3 and 531.6 ​eV were ascribed to the absorbed water in surface [38] and the metal-oxygen bond resulting from residual Cu2O, respectively. And the one owning a binding energy of 530.9 ​eV was related to a hydrated phase of nickel (Ni–O–H) [38,39].To reflect the electrocatalytic activity of as-made catalysts, the OER measurements were performed. Firstly, the linear sweep voltammetry (LSV) curves were measured using a relatively low scan rate of 2 ​mV ​s−1 in order to minimize the capacitive current. And if no additional instructions, all LSV overpotentials mentioned below were chosen uniformly at the benchmark of 10 ​mA ​cm−2 for easy comparison. As comparing all samples in Fig. 5 a, Cu2O was not suitable for OER catalysis because of the high overpotential (358 ​mV) of Cu2O wires. At the same time, introducing Ni foam as substrate did not provide more activity for various catalytic materials, which could be demonstrated by the extremely high overpotential of Ni foam (363 ​mV). In sharp contrast, the Ni(OH)2 HTs catalyst prepared by template method had the most negative overpotential (207 ​mV), even better than commercial RuO2 catalyst (226 ​mV). The comparison of overpotentials intuitively stressed the active advantage of Ni(OH)2 HTs as OER catalyst. In addition, the catalytic performance of Ni(OH)2 HTs outperformed that of other newly reported typical Ni-based catalysts (Fig. 5b and Table S1) in terms of overpotential at 10 ​mA ​cm−2 [40–46]. It is worth mentioning that the designs of pure phase Ni(OH)2 through different synthetic methods and its representations in OER have been reported in previous reports. For example, Gao and coworkers successfully synthesized nanosheet-assembled α-Ni(OH)2 hollow spheres, β-Ni(OH)2 hexagonal nanoplates and β-Ni(OH)2 nanoparticles through a simple solvothermal strategy, where α-Ni(OH)2 hollow spheres showed the most effective OER performances with an overpotential of 331 ​mV [18]. Luan' group prepared α-Ni(OH)2 with various morphologies by lamellar reverse micelles method, including three kinds of layer-stacking Ni(OH)2 (bud-like, flower-like and petal-like) and ultra-large sheet-like Ni(OH)2. Due to the dominant structural effects, the fewer stacked layer petal-like Ni(OH)2 owning more active boundary sites exhibited the best OER activity with an overpotential of 260 ​mV [47]. The above studies show that the structure effects of the catalyst have a great impact on its activity. In comparison, Ni(OH)2 HT in this work has the lowest overpotential among these excellent Ni(OH)2 catalysts.Then, the kinetics advantage of Ni(OH)2 HTs could be evaluated by the Tafel slope which was calculated through LSV date at an extremely slow scan rate of 0.1 ​mV ​s−1. As displayed in Fig. 5c, the Tafel slope of 79.8 ​mV dec−1 for Ni(OH)2 HTs was significantly smaller than that of Cu2O wires template (189.5 ​mV dec−1) and commercial RuO2 (139.3 ​mV dec−1), indicating an excellent OER kinetics in Ni(OH)2 HTs catalyst. Besides the Tafel slop, the EIS Nyquist plot is another important characterization to investigate kinetic behavior defining the interfacial charge transfer of the catalyst [48–50]. Fig. 5d shows the Nyquist curve of Ni(OH)2 HTs and its fitting results corresponding to total frequencies (dotted line) and high-frequency region (inset in Fig. 5d) respectively. According to the obtained fitting data, Ni(OH)2 HTs possessed a expectedly low values of solution resistance (Rs, 1.71 ​Ω) and charge transfer resistance (Rct, 5.12 ​Ω), which were greatly improved when compared with Cu2O wires with the Rs and Rct fitted to 1.99 and 19.21 ​Ω (Fig. S5). Both Tafel slopes and EIS results suggested that a rapid charge transfer kinetics existed in Ni(OH)2 HTs to prompt more active catalytic reaction, largely due to the construction of numerous one-dimensional channels. Moreover, the estimated electrochemical double-layer capacitance (Cdl), a slope deduced from a functional relationship between capacitive current and scan rate which all extracted from the cyclic voltammetry (CV) tests (Figs. S6a and b), was also used to evaluate the electrochemically active surface area (ECSA) which is proportional to Cdl. As shown in Fig. 5e, the values of Cdl were calculated to be 8.68 and 1.42 ​mF ​cm−2 for Ni(OH)2 HTs and Cu2O wires respectively, illustrating more accessible active sites in Ni(OH)2 HTs with the assistance of template method. Moreover, the ECSA results exactly corresponded to the trend originating from above calculated BET specific surface area.In order to guarantee the application of Ni(OH)2 HTs in practice, the chronoamperometry method was applied to test the long-term durability of the catalyst. With a polarization for a time period of 24 ​h, no significant decay appeared (Fig. 5f), indicating a strong stability in Ni(OH)2 HTs. The robust structural foundation of Ni(OH)2 HTs comes from the stable Cu2O wires frame. Next, the XRD analysis (Fig. S7a) exhibited a transformation to amorphous state in Ni(OH)2 HTs after the OER measurement while the XPS curves unveil that the binding energies of Ni 2p and O 1s (Figs. S7b and c) had only a weak positive shift, meaning a persistent stability for the catalyst and that the real catalyst for OER might still be transition metal hydroxide with Ni2+ species as the active sites.With the structure-activity relationship revealed by all the above physical and electrochemical characterizations, the advantages of Ni(OH)2 HTs as OER catalyst can be summarized as follows. Firstly, the template-assisted method using the wire-like Cu2O as sacrificial template brought the hollow tubular interior and rough surface to Ni(OH)2 HTs, which greatly increased the specific surface area of the catalyst with more active sites. Thus, more contact opportunities between OH− in the electrolyte and the active sites could be achieved to improve the catalytic efficiency. Secondly, the tubular configuration of Ni(OH)2 HTs provided numerous channels for the rapid charge/mass transport. Coupled with the relatively good inherent conductivity of Ni(OH)2·0.75H2O, Ni(OH)2 HTs obtained excellent electrochemical kinetics. Finally, the hollow structure evolved from stable Cu2O wires frame was extremely firm, which provided a structural basis for the dynamic OER long-term process.In summary, Ni(OH)2 HTs were synthesized according to HSAB principle through template-assisted method with Cu2O wires as sacrificial template. The stable hollow one-dimensional structure and rough surface from this strategy resulted in greatly increased specific surface area (raising ECSA) for abundant active sites, numerous transmission paths for fast electrochemical kinetics, and sturdy build for OER long-term stability. Benefiting from the structure-activity relationship, the Ni(OH)2 HTs as OER catalyst showed outstanding catalytic activity and kinetics with only overpotential of 207 ​mV to drive the current density of 10 ​mA ​cm−2 and a Tafel slope of 79.8 ​mV dec−1. And there was almost no decay to maintain a current density within 24 ​h. The novel preparation pattern in this article provides a new strategy to design transition metal-based hydroxide materials with hollow porous tubular structure, and helps to further dig into the research on advanced catalysts for water splitting.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the National Key Research and Development Program of China (No. 2018YFA0703700), the National Natural Science Foundation of China (Nos. 12004031, 12034002 and 51971025), Beijing Natural Science Foundation (Grant No. 2212034), Foshan Talents Special Foundation (BKBS202003), Scientific and Technological Innovation Foundation of Foshan (No. BK22BE005) and Foshan Science and Technology Innovation Project (No. 2018/T100363).The following is the Supplementary data to this article: Multimedia component 1 Multimedia component 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.pnsc.2022.09.002.

Cu2O is an ideal template material for the preparation of transition metal hydroxide/oxyhydroxides with oxygen evolution reaction (OER) enhanced catalytic performance. Here, inspired by Pearson's principle, Cu2O wires were prepared and used as a sacrificial template to prepare Ni(OH)2·0.75H2O hollow tubes (Ni(OH)2 HTs) with highly improved surface roughness. Benefiting from unique structural advantages, the Ni(OH)2 HTs showed excellent catalytic activity, rapid kinetics and a long-term stability as the OER catalyst, where an overpotential of only 207 ​mV was required to drive a current density of 10 ​mA ​cm−2, an ideal kinetics with a Tafel slope as 79.8 ​mV dec−1 was calculated, and no obvious attenuation in chronoamperometry was discovered after operation for 24 ​h. This paper provides a novel template-assisted strategy to prepare high-performance transition metal-based OER catalysts possessing hollow and tubular structures.

The metal nanoparticles (MNPs) take over the growing interest in the catalytic processes. MNPs are characterized by a high surface-to-volume ratio providing a large number of active sites (surface atoms) per unit area [1]. The presence of large number of reactive surface atoms enhances the efficiency of the catalytic process as a whole [2,3]. However, the high surface energy of MNPs makes them thermodynamically unstable and more likely to aggregate during the catalytic process, suppressing the catalytic activity. Also, the use of MNPs catalysts is limited due to the difficulty of their separation and recycling.To address these obstacles in homogeneous catalysis, researchers pay their attention in developing novel, green, and efficient heterogonous catalytic systems [4,5]. Loading MNPs onto a solid matrix (metal oxide, inorganic silica, glass, polymers, etc.) has been emerged as an efficient pathway for designing eco-friendly catalytic systems that enable the recyclability and reusability of the MNPs catalysts [6]. However, the heterogeneous catalysts usually show lower activity and selectivity as compared with their homogeneous counterparts, due to the lower surface area available for the interaction between the substrates and the catalyst surface [7]. This is the critical issue that restricts the development of heterogeneous catalytic systems. In this regard, researchers continue to innovate novel heterogeneous catalysts that combines the advantages of both homogenous catalysts (high activity and selectivity, etc.) and heterogeneous catalysts (ease/safe synthetic pathway, efficient separation of MNPs, possibility of recycling the MNPs catalysts, easy catalyst separation, and thermal stability) [8].Polymeric matrices as catalyst supports have attracted a considerable interest in modern catalytic reactions due to their functionality, ease of preparation, excellent mechanical properties, and the ability of surface modifications. The preparation of polymer/MNPs nanocomposites as free-standing film enables the efficient removal and recycling of the catalyst from a catalytic reaction. This is the most required feature from economic and environmental points of view. In recent years, several approaches have been developed to fabricate polymer/MNPs nanocatalyst such as Zn/chitosan/Fe3O4 [9], Ag and Au/cellulose nanocrystals [10], TiO2/polyethersulfone [11], Ni/Hyperbranched polyaromatic polymer [12], Cu/poly(aminobenzoic acid) [13], Fe/coordination polymers [14], and (Fe/Co)/Bis(pentamethylene)pyridyl [15]. Most of these approaches faced with technical and/or economical hardships i.e., they have high capital and operational costs, unacceptable sensitivities to operational conditions, and high energy consumption as well as high sludge generation. An ideal heterogeneous catalyst is expected to overcome the above limitations as well as has a strong activity for a wide range of catalyzed reactions and enables a facile contact between the active sites and chemical substrates.The Graphene oxide (GO) shows outstanding properties such as abundant oxygenated functional groups, large surface area, strong mechanical properties, and chemical stability. These properties make GO as a strong candidate in heavy metal removal from solutions. Recently, it has been found that the dispersion of GO sheets inside polymeric chains suppress the agglomeration/aggregation of GO sheets in the adsorption of heavy metal ions such as Pb2+. The immobilization of GO sheets within a polymeric matrix facilitates the bonding between GO and metal ions leading to an increase in the Pb2+ adsorption capacity from 800 to 1730 mg g−1 [16,17].Herein, it was considered worthwhile to develop an eco-friendly nanocomposite of poly(vinyl alcohol) (PVA), reduced graphene oxide (rGO), and nickel (Ni) nanoparticles, as it could be used in catalytic oxidation/reduction of water pollutants. The synthetic route satisfied the cost aspects and avoided the usage of hazardous materials. The physical shape of Ni@rGO/PVA nanocomposite (film-like structure) enabled the efficient separation and reuse of the catalyst for several cycles.An aqueous PVA solution (3% wt/V) was prepared by dissolving PVA powder (PVA, 98–99% hydrolyzed, molecular weight from 75000-80000, LOBA CHEMIE) in distilled water at 80 °C for 10 h until a clear solution was obtained. The polymeric solution was cooled naturally to room temperature. GO was prepared using Modified Hummers’ method described elsewhere [18–20]. Then, 1 mL of GO (1 mg/mL) was added to 10 mL of PVA solution under magnetic stirrer for 6 h to attain a mixture of higher homogeneity. The GO/PVA mixture was casted on a polyethylene dish and allowed to dry at room temperature to get the desired film. A 10 mL of PVA was casted on a polyethylene dish and allowed to dry at room temperature to prepare pure PVA film for comparison.The free-standing film of GO/PVA was immersed into a solution of 1 M NiCl2 for 24 h. The polymeric film Ni2+/GO/PVA was removed and left to dry. The polymeric film was soaked in cold solution of NaBH4 (4 °C) to reduce Ni2+ to Ni nanoparticles. The Ni@rGO/PVA nanocomposite film was left to dry for the subsequent characterization and stored for future use. The whole preparation steps are summarized in the schematic description shown in Fig. 1 .The PVA, GO/PVA, and Ni@rGO/PVA nanocomposite films were prepared as (0.5 × 0.5 cm) film and weighed accurately (W0). The samples were soaked in distilled water from 0.0 to 2 h at room temperature (≈25 °C). The swollen films were removed and the excess water which present on the surface of the PVA films was wiped using a soft tissue and weighed again (W1). The swelling% of the polymeric films was calculated using the following equation: (1) Swelling % = 100 × ( W 1 − W 0 / W 0 ) Thereafter, the films were dried (at 40 °C in an oven) until constant weight was reached (W2). The gel% was calculated according the following relation: (2) Gel % = 100 × ( W 2 / W 0 ) The crystalline phase of the prepared Ni@rGO/PVA nanocomposite film was characterized using X-ray powder diffractometer (XRD) of the type Schimadzu 6000 (Japan). A scanning electron microscope (SEM) of the type PHILIPS/FEI QUANTA 250 quipped with Energy dispersive X-ray analysis (EDX) was used to study the morphology of the prepared samples. The UV/visible (UV/vis) spectra were collected using a Jasco V-770 spectrophotometer. A WITec alpha 300 R confocal Raman microscope was used to record Raman spectra. The samples were excited by 532 nm laser line. The surface chemical structure of the prepared samples was studied using X-Ray photoelectron spectrometer (XPS, Thermo Scientific K-ALPHA instrument).The catalytic reduction of Cr6+ using Ni@rGO/PVA as a catalyst and formic acid (FA) as a reducing agent was performed in a 15 mL glass beaker at ambient conditions. In short, a 10 mL aqueous solution of Cr(VI) (0.7 mM), 1 mL FA, and 0.02 g of Ni@rGO/PVA film were placed in a glass beaker under stirring. The performance of the Ni@rGO/PVA was assessed by following the decay of the 350 nm-peak of K2Cr2O7 as a function of time. The data were collected in triplicate.A 0.03 g of Ni@rGO/PVA film was soaked in a volume of 3 mL aqueous solution of 0.08 mM 4-NP and 0.5 mM of NaBH4. The catalytic reduction was followed by observing the decay of the 400 nm absorption peak using UV/vis spectrophotometer.A 0.03 g of Ni@rGO/PVA film was dipped into 3 mL solution containing 0.05 mM MB and 0.5 M H2O2. The catalytic oxidation cycle was observed by following the diminishing of the absorption peak at 664 nm during the reaction. Fig. 2 displays the XRD patterns of PVA, GO/PVA, and Ni@rGO/PVA films. The diffraction peak at 2θ = 19° represented the semi-crystalline phase of the PVA matrix which is consistent with the previous report [21]. In GO/PVA nanocomposite, the main peak of PVA got weaker and broader. This was attributed to the decreasing crystallinity of PVA due to the probable interactions between GO and PVA segments [22]. The absence of the characteristic peak of GO at 2θ = 11° [23] might be due to the masking effect of PVA matrix and/or the peak is too weak to be detected. The XRD pattern of Ni@rGO/PVA film showed three new peaks at 2θ = 26°, 44°, 52°, and 76°. The XRD peak at 2θ = 26° indicated the reduction of GO to rGO sheets under the action of NaBH4. The reduction of GO into rGO by NaBH4 was reported by Muda et al. [24]. It is well known that the bulk graphite with interlayer distance of ≈0.34 nm has a distinct peak at 2θ = 26.7° [25]. Accordingly, the peak at 2θ = 26° could be attributed to the formation of restacked rGO layers i.e., confirming the removal of oxygen functional groups during the reduction. The XRD peaks at 2θ = 44°,52°, and 76° are corresponding to the crystallographic planes (111) (200), and (220) of fcc Ni nanoparticles (JCPDS #04–0850). The broadening and weakness of the XRD peaks of Ni nanoparticles is in accordance with their small grain size and low degree of crystallinity.SEM imaging was used to characterize the surface morphology of pure PVA, GO/PVA, and Ni@rGO/PVA films. Fig. 3 (a) shows the SEM image of the pure PVA film where a smooth surface without any distinct morphological features were observed. The GO/PVA showed densely packed distribution of GO with a flat plates shape (see Fig. 3(b)). Some cracks were observed which were attributed to the sensitivity of the PVA film to the electron beam. Fig. 3(c) shows the morphology of Ni@rGO/PVA film. The film is characterized with the appearance of bright dots which could be attributed to the formation of Ni nanoparticles. The SEM results clearly demonstrated the successful formation of Ni nanoparticles onto the GO/PVA film, and no free Ni nanoparticles were present in the image. The EDAX spectrum (Fig. 3(d)) revealed that Ni@rGO/PVA was composed of 44.7% C, 43.89% O, and 11.41% Ni elements without any impurities from precursors. Fig. 4 (a) depicted the Raman spectra of the pure PVA film. The Raman peak centered at ≈ 2913 cm−1 is due to the CH2 stretching vibrations mode. The stretching vibration of –CH was appeared at ≈ 1444 cm−1. The fine structure between 1145 and 1100 cm−1 was attributed to the C–O stretching modes, while the C–C stretching modes are observed at 920 and 853 cm−1 [26,27]. The Raman spectrum of GO was shown in Fig. 4(b). The two distinguished bands at 1344 and 1593 cm−1 were due the defective (D) and graphite (G) modes of graphitic materials, respectively. The large intensity of the D-mode was in a good indication for the formation of highly oxidized graphene sheets. The peaks appeared at 2650, 2930, and 3170 cm−1 could be assigned to G′, and D + G, and 2G′ modes, respectively. The appearance of these peaks is a direct evidence of oxidization of graphene sheets edges. The addition of GO to the PVA matrix did not affect the PVA structure as shown in Fig. 4(c). The D and G modes of the GO at 1344 and 1593 cm−1, respectively, were clearly seen in addition to the main characteristic peaks of the PVA chains at 853, 920, and 2913 cm−1. The PVA and GO peaks did not show any significant shift indicating that GO sheets were mixed with the PVA chain through the weak van der Waals forces. In the case of Ni@GO/PVA film (Fig. 4(d)), in addition to the appearance of the main characteristic peaks of PVA and GO, one can clearly observed the appearance of a low intensity Raman peak at 490 cm−1. It is well known that pure Ni metal is Raman in-active. Accordingly, this peak could be assigned to the formation of few Ni–O bonds on the surface of Ni nanoparticles [28]. Fig. 5 (a) depicted XPS survey scan spectra of Ni2+ doped GO/PVA before and after reduction with NaBH4. The two strong peaks corresponding to the binding energies at about 284 and 532 eV are assigned as C1s and O1s orbitals, respectively. The XPS spectra showed an electronic signal at 858 eV which was attributed to the presence of Ni species within the nanocomposite matrices. The peaks in Fig. 5(b) and (e) depict high-resolution spectra of Ni2p orbitals before and after reduction of Ni2+ ions, respectively. The peaks at 855.8 and 873.4 eV correspond to binding energies of Ni2p3/2 and Ni2p1/2. Four electronic signals were seen from the high-resolution spectrum of Ni2p before reduction at 856.9, 838.3, 864.4, and 867.9 eV. These peaks were assigned as the satellites peaks about Ni2p orbital [29]. After reduction of Ni2+ with NaBH4, the high resolution XPS spectrum showed two main peaks at 854.1 and 873.4 eV. These two peaks were attributed to Ni2p3/2 and Ni2p1/2 of the zero-valent Ni species [30,31].The high-resolution spectra of C1s of Ni2+-doped GO/PVA before and after reduction with NaBH4 were presented in Fig. 5(c) and (f), respectively. The C1s spectrum acquired before reduction (Fig. 5(c)) was resolved into three peaks centered at 284.3, 286.3, 288.2 eV, arising from C=C/C–C in aromatic rings, C–O of epoxy and alkoxy, and C=O groups [32]. The peak intensity of the oxygenated functionalities was reduced markedly after reduction (see Fig. 5(f)).For the O1s spectrum before reduction (Fig. 5(d)), the 532 eV peak was resolved into three peaks. The peak from (COOH/C=O/R–O–R) was observed at 532.4 eV. The peak at 533.5 eV can be related to the (O–H) groups, and the signal at 536.1 eV can be assigned as an epoxy group [33]. After reduction (see Fig. 5(g)), the peak area of O1s was decreased markedly as well as the epoxy peak (536 eV) totally disappeared. The atomic% of C/O from XPS analysis was increased from 1.23% before reduction to 3.14% after reduction, suggesting that some oxygen-containing groups were removed due to the reduction of GO to rGO with NaBH4.The swelling and gel% of PVA, GO/PVA, and Ni@rGO/PVA films were determined and the results were displayed in Fig. 6 . After ≈15–20 min, pure PVA film adsorbed large quantity of water and a gel-like swollen PVA film was formed at room temperature. It is well known that pure PVA films of moderated molecular weight are characterized by their high degree of water absorption and film completely dissolves at ≈ 37–40 °C [34,35]. Here, at room temperature, the swollen film lost its mechanical flexibility and teared down to small pieces i.e., no data could be collected. For GO/PVA and Ni@rGO/PVA films, the selling rate was fast within the first 40 min reaching 750% and 600%, respectively. Thereafter, it gradually slowed down reaching its equilibrium value after 90 min. The GO/PVA has the highest maximum swelling ratio (≈960%) as compared with that of Ni@rGO/PVA (≈880%). The gel% of the two samples seemed to not change markedly after soaking in distilled water after 130 min (see inset of Fig. 6).GO is a two-dimensional (2D) carbon material with one-atom thickness rich with hydrophilic groups i.e. (–OH) (–C=O), and (–COOH) groups [36,37]. These hydrophilic groups improve the miscibility of GO sheets with water-soluble PVA through the formation of hydrogen bonds with the PVA chains. Due to the 2D nature GO sheets, the formed hydrogen bonds between them will result in an increase in both tensile modulus and tensile strength of PVA matrix. On the other hand, the GO nanosheets serve as crosslinking agent that restrain the solubility of PVA film yielding highly hydrophilic and insoluble polymeric matrix [36–38]. These findings are consistent with previous data which reported GO as a perfect candidate for fabricating novel superabsorbent hydrogels [39].The investigated catalyst, Ni@rGO/PVA film, has been designed taking into account the environmental and industrial requirements. PVA was chosen as a supporting matrix due to its bio-compatibility, hydrophilicity, and film forming ability that shows good anti fouling properties [40]. However, PVA swells rapidly in aqueous solution until complete dissolution. To overcome this disadvantage, PVA must be crosslinked before it can be used in aqueous media. PVA can be crosslinked chemically (using glutaraldehyde, acetaldehyde, formaldehyde, etc.), and physically (electron beam, γ-irradiation, freeze-thawing cycles, etc.) [41]. These methods require the use of toxic chemicals and laboratory equipment. Also, the crosslinking decreases the equilibrium degree of swelling i.e., the ability of water uptake of the PVA is reduced markedly.GO is enriched with oxygen-containing functional groups, such as (–OH) (–COOH), (R–O–R), (–C=O) and epoxy groups as deduced from the results of XPS. Blending GO with PVA will introduce these polar groups to the polymeric matrix to achieve three tasks. First, the polar groups crosslink the PVA matrix through the formation of hydrogen bonds with PVA chains. This produces the three-dimensional network structures preventing the dissolution of PVA. Second, the presence of the polar groups imbibes water within the polymeric networks and increasing the equilibrium degree of swelling of the PVA matrix. Third, the oxygenated functional groups on the GO surface and π-electron system provide plentiful binding sites for the adsorption and tight fixing of the Ni2+ ions [42]. In catalytic reactions, when the Ni@rGO/PVA film was soaked within aqueous solution, the polymeric film swells (up to ≈ 900 times of their dry weight) and the dissolved reactants enter the polymeric chains. This will allow the effective adsorption of the reactants onto the catalytic centers (Ni NPs in the present case) where the catalytic process takes place. After reduction, there is a metal–support interaction, which is due to the overlap between the π-orbitals of graphene with the d orbitals of transition metal atoms. The bonding between Ni particles and graphene sheets secures them against falling off from the composite film [43].The presence of Cr6+ in the industrial wastes induces harmful issues to the humans, animals, and plants. Thus, the reduction of Cr6+ into Cr3+ (which has a low mobility and is not toxic to the most living organisms) is environmentally required to minimize chromium contamination in the environment [44]. The catalytic performance of the as-prepared Ni@rGO/PVA had been studied for the reduction of Cr6+ to Cr3+ in aqueous solution using formic acid. The UV/vis spectrum of K2Cr2O7 and formic acid solution displayed two peaks at 437 and 350 nm due to ligand-to-metal charge transfer absorption [45]. The mixture (K2Cr2O7 and formic acid) maintained its spectral profile without considerable change for several hours i.e., the addition of formic acid could not reduce Cr6+ in absence of catalyst. In contrast the dipping of Ni@rGO/PVA film in the reacting solution markedly accelerated the reduction of Cr6+ into Cr3+ in the presence of excess formic acid (Fig. 7 (a)). The progress of the reduction was followed by recording the decay of the 350 nm-peak of K2Cr2O7 as a function of time [46]. It was observed that the 350 nm-peak gradually decayed with increasing reaction time. The 350 nm-peak vanished completely with decolorization of the solution (from yellow to colorless) after 35 min indicating the reduction of Cr6+. The reduction of Cr6+ to Cr3+ was confirmed by the appearance of the green color after treating the resulting solution with excess NaOH solution [47]. It was suggested that formic acid was adsorbed onto the Ni surface to produce a Ni-formate intermediate, which was reactive for the reduction of Cr6+ [46].Due to the high concentration of formic acid (in comparison with that of Cr6+), the kinetics follows a pseudo-first-order model as follow: (3) ln ( C t / C 0 ) = ln ( A t / A 0 ) = − k a p p t where C0, Ct, A0, and At are the concentrations and their corresponding absorption of K2Cr2O7 at the beginning and after time t of the reaction. k app is the apparent rate constant of the catalytic reaction which is equal to the slop of the straight line of ln(A t /A0) vs t.The effect of [Cr6+] on the k app of the catalytic reduction was evaluated at the following conditions (0.02 g of Ni@rGO/PVA, 1 mL of formic acid, and pH = 6) as presented in Fig. 7(b) and Table 1 . The k app of the catalytic reduction decreased from 1.45 × 10−3 to 0.54 × 10−3 s−1 by increasing the [Cr6+] from 0.5 to 1 mM. The reduction was achieved just in 26 min for the lowest concentration studied (0.5 mM) whereas it exhausted about 44 min for completion when [Cr6+] is increased to 1 mM. The presence of a large number of Cr6+ at higher concentrations) might block the active centers of the catalyst thereby adsorption onto and/or shielding the catalyst surface. This in turn prevents/minimizes the decomposition of the formic acid at the catalyst surface slowing down the catalytic reduction as a whole. The catalyst dosage plays a vital role in optimizing the operational parameters of the catalytic reaction and to avoid the excess use of the catalyst. The impact of catalyst dosage on the catalytic reduction of Cr6+ was investigated by varying the weight of Ni@rGO/PVA from 0.02 to 0.1 g, at the following conditions; ([Cr6+] = 0.7 mM, 1 mL of formic acid, and pH = 6). The k app of catalytic reduction increased from 0.95 × 10−3 to 1.94 × 10−3 s−1 by increasing the catalyst from 0.02 to 0.1 g (Fig. 7(c) and Table 1). The addition of more doses of the catalyst necessarily means adding more active centers i.e., more reactive cites are available for adsorption of Cr6+ as well as for the decomposition of formic acid. This in turn can speed up the reaction rate to attain equilibrium. It could be seen that the increase of the catalyst dosage from 0.07 to 0.1 g did not yield a significant increase in the reaction rate. Hence, the 0.07 g catalyst dosage was chosen for the subsequent study.The dosage of formic acid has a key role in the catalytic reduction of Cr6+ as it is the hydrogen donor. The study was conducted to evaluate the effect of formic acid dosage on the k app of the Cr6+ reduction at the following conditions (0.07 g of Ni@rGO/PVA, [Cr6+] = 0.7 mM, and pH = 6). The results were displayed in Fig. 7(d) and Table 1. The results indicated that the reduction rate was accelerated with the increase of formic dosage. The k app was increased from 1.72 × 10−3 to 2.67 × 10−3 s−1 as the dosage of formic acid increased from 1 to 3 mL. The use of more formic acid means the more hydrogen was produced at the catalyst surface which in turn accelerates the reduction of Cr6+ [48].The catalytic reduction of Cr6+ is strongly pH-dependent reaction and catalyzed by the dissolved and surface-bound metals [48]. The rate constant (k app ) was calculated as function of the pH value of the reaction at the following conditions (0.07 g of Ni@rGO/PVA, [Cr6+] = 0.7 mM, and 3 mL of formic acid) and the results were given in Fig. 7(e) and Table 1. The catalytic reduction of Cr6+ to Cr3+ was accelerated markedly as the acidity of the reacting solution was increased (at lower pH values). The k app was increased from 2.67 × 10−3 to 7.29 × 10−3 s−1 as the pH value decreased from 6 to 2.The pH value controls the surface charge, substrate ionization and the hydrogen donor dissociation, when it participates in the hydrogen transfer process [49]. The lower pH values are favored for the adsorption of negatively charged formate (HCOO−) and dichromate (Cr2O7)2− ions onto the catalyst surface [50]. As a result, the adsorption of the reactants onto the catalyst surface is accelerated and higher catalytic rates are promoted at lower pH values. Hence, the catalytic reduction of Cr6+ was highly pH dependent.It is nonetheless worth conducting recycling trials to evaluate the stability and reusability of Ni@rGO/PVA film. This was achieved by using thoroughly rinsed Ni@rGO/PVA film for the next catalytic cycle, followed by distilled water washing after each cycle. Fig. 7(f) showed the catalytic performance of Ni@rGO/PVA film for ten catalytic cycles. It could be seen that the reduction activity decreased less than 12% of the initial activity whereas the Cr6+ conversion% lost about 7% of its initial value after the tenth catalytic cycle. The content of the segregated Ni within reacting medium was determined by atomic absorption. The collected data depicted that no Ni species were determined, suggesting that Ni was fixed in the rGO/PVA film composites. The film shape of Ni@rGO/PVA catalyst could realize instantaneous and efficient separation of the catalyst from the reacting medium. This eliminates the need for industrially inapplicable separation methods such as centrifugation, filtration, precipitation by pH shift, etc. [10]. This type of nanocomposite films could be built in fixed bed reactors under flow of reacting molecules on the way of real applications of such catalysts.The catalytic reduction of 4-NP to 4-AP by NaBH4 is considered as the foremost model catalytic reactions. This is due to it the straightforward assessment of the catalyst performance using the real-time spectroscopic analysis of an aqueous solution. The nitrophenol derivatives are essential intermediates for variety of industries such as pesticides, petrochemical, production of paper, pharmaceuticals, etc. [51]. In addition, the nitrophenol compounds are classified by U.S. Environmental Protection Agency (EPA) as non-biodegradable pollutants. The catalytic reduction of 4-NP to 4-AP is environmentally and industrially of great interest. The visual inspection of the reaction between 4-NP and NaBH4 (i.e., without catalyst) indicated the stability of the yellow color of the solution for several days without considerable change. This is attributed to the potential barrier between the standard reduction potentials (E o ) of the reactants; E o (4-NP/4-AP) = −0.76 V; E o (H3BO3/BH4 −) = −1.33 V [52]. After the dipping of the Ni@rGO/PVA film in the reacting medium, the yellowish color of the solution vanished gradually until complete decolorization. Fig. 8 (a) showed the time-dependent UV/vis spectra of the reduction of 4-NP with NaBH4 in presence of Ni@rGO/PVA film as a catalyst. The spectra is dominated with strong peak at 400 nm which is attributed to the nitrophenolate ions [53,54]. As the reaction progressed, the 400 nm-peak faded gradually while a new peak at 298 nm appeared which is a sign of 4-AP formation [55]. The disappearance of the 400 nm-peak was taken as sign for the completion of the reaction. It could be concluded that Ni nanoparticles provided the reactive sites for the reactant adsorption and reducing the kinetic energy barrier of the reduction.As the reduction of 4-NP occurred in excess of NaBH4 ([NaBH4]/[4-NP] = 85), the catalytic reaction was described by the first-order model i.e., the rate constant of the reaction depends only on the 4-NP [56]. To evaluate the impact on k app , the catalytic experiments were performed by varying a single operational parameter of the 4-NP concentration, catalyst dose, NaBH4 concentration, and pH value. Fig. 8(b) displayed the plots of (ln(A t /A0) vs. t) at different initial 4-NP concentrations at the following conditions; 0.03 g of Ni@rGO/PVA [NaBH4] = 7 mM, and pH = 7.6. The value of k app decreased from 4.18 × 10−3 to 1.36 × 10−3 s−1 as the [4-NP] increased from 0.08 to 0.16 mM concentration as listed in Table 2 . On the other hand, the k app was found to increase from 2.72 × 10−3 to 10.2 × 10−3 s−1 as the catalyst dose increased from 0.02 to 0.05 g at the following conditions [4-NP] = 0.08 mM, [NaBH4] = 7 mM, and pH = 7.6 (see Fig. 8(C) and Table 2). According to Langmuir–Hinshelwood (L-H) model, the co-adsorption of 4-NP and NaBH4 (to the catalyst surface) is essential for the efficient catalysis. Therefore, the increased concentration of 4-NP occupied more active sites of the catalyst and minority sites were available for active hydrogen species. This resulted in suppressing the catalytic process and decreasing k app value. Also, the more the catalyst dose, the more is the number of the active sites available for the adsorption of the reactants. This accelerates the catalytic reaction [57].The effect of [NaBH4] on the rate of the catalytic degradation of 4-NP was also evaluated at the following conditions: 0.05 g of Ni@rGO/PVA, [4-NP] = 0.08 mM, and pH = 7.6. According to Fig. 8(d) and Table 2, one can observe that in the concentration range (5–10 mM) of NaBH4, the k app increased from 5.82 × 10−3 to 13.3 × 10−3 s−1. However, increasing the [NaBH4] to higher than 10 mM did not show significant effect on the k app i.e., the rate remains almost constant. The advantages of higher [NaBH4] are limited and the most would be wasted because of the limited substrate available. Beyond the optimal level, the increasing [NaBH4] would not effectively enhance the efficiency of the process.The effect of pH on the k app in the representative experiment of the reduction of 4-NP was performed at the following conditions: 0.05 g of Ni@rGO/PVA [4-NP] = 0.08 mM, and [NaBH4] = 10 mM. Fig. 8(e) displayed the first-order kinetic curves of 4-NP over Ni@rGO/PVA at different pH values. It is quite evident that the k app was speeded from 5.57 × 10−3 to 23.6 × 10−3 s−1 with raising the pH value from 6 to 9 (Table 2). With further increase in the pH value, no remarkable changes were recorded on the k app values. The aqueous NaBH4 solution is stable in the highly alkaline solution and the deprotonation of –OH moiety of 4-NP is decreased at pH > 10 [58].The reusability of the Ni@rGO/PVA catalyst was also studied with the catalytic reduction of 4-NP. As depicted in Fig. 8(f), the catalytic activity of 4-NP to 4-AP was 83% after 15 cycles. The Ni@rGO/PVA showed good catalytic performance after 15 cycles without significant loss of active sites (Ni species), due to the supporting effect of rGO/PVA matrix.The oxidative degradation of MB dye over Ni@rGO/PVA film was represented in Fig. 9 (a). The catalytic oxidation of MB was detected by following the diminishing of the absorption peak at 664 nm [59]. This experiment was performed with 0.03 g Ni@rGO/PVA [MB] = 0.05 mM, [H2O2] = 0.5 M, pH = 6 and stirring the reaction mixture at 300 rpm. It was noted that the 664 nm-peak decayed completely within 32 min. This happened in conjunction with the disappearance of the distinctive blue color of the dye. For blank reaction, separate experiments were carried out under similar conditions, once without catalyst and another without H2O2. It was noted that there was no oxidation occurred and the dye solution maintained its color for several days without considerable changes.The effect of the initial [MB] on the k app of the catalytic processes was evaluated with 0.03 g Ni@rGO/PVA [H2O2] = 0.5 M, pH = 6 and stirring the reaction mixture at 300 rpm. It was observed that as the initial MB concentration increased the decolorization of MB was markedly slowed down (Fig. 9(b) and Table 3 ). At higher MB concentrations, the co-adsorption of the reactants (MB and H2O2) was significantly disrupted i.e., more MB molecules were adsorbed to the catalyst at the expense of H2O2. This in turn inhibited and/or reduced the production of hydroxyl radicals necessary for the efficient oxidation [21]. The effect of Ni@rGO/PVA dosage was evaluated through a series of experiments carried out at [MB] = 0.05 mM [H2O2] = 0.5 M, pH = 6 and stirring the reaction mixture at 300 rpm. The calculated data were presented in Fig. 9(c) and Table 3. It could be seen that the decolorization efficiency of MB increased with increasing the catalyst dosage i.e., the k app increased from 0.68 × 10−3 to 2.91 × 10−3 s−1 with increasing Ni@rGO/PVA dosage from 0.1 to 0.07 g.The effect of [H2O2] was evaluated at the following experimental conditions: 0.07 g Ni@rGO/PVA, [MB] = 0.05 mM, pH = 6 and stirring the reaction mixture at 300 rpm. Fig. 9(d) and Table 3 showed the behavior of k app of the MB oxidation at different initial [H2O2]. It could be seen that the k app increased from 1.09 × 10−3 to 4.81 × 10−3 s−1 with increasing [H2O2] from 0.25 to 1 M. The further increase of [H2O2] to 1.25 M resulted in decreasing the k app value to 4.42 × 10−3 s−1. The excess H2O2 might serve as a scavenger of OH and forms a less reactive perhydroxyl radicals, resulting in reduction of the catalytic rate [60].The pH effect was followed by changing the pH value of the reacting solution from 5 to 2.5 with 0.07 g Ni@rGO/PVA [MB] = 0.05 mM, [H2O2] = 1 M, and stirring the reaction mixture at 300 rpm. It is worthwhile to mention that there is no remarkable degradation of the MB dye in the pH range of 6.0–11. In the alkaline medium, H2O2 loses its oxidation ability due to the formation of oxygen and H2O [21]. Fig. 9(e) and Table 3 showed that the decrease in the pH value resulted in higher oxidation rates of MB dye. For the catalytic oxidation reaction, the k app  = 11.5 × 10−3 s−1 was achieved at a pH = 2.5. At low pH values, H2O2 yields a considerable amount of OH radicals which accelerate the oxidation process of MB dye. For pH values below 2.5, the k app was decreased markedly. This might be attributed to the presence of large number of H+ which in turn reacts with OH radicals to form H2O and slows down the reaction rate.In order to study the robustness and recyclability of Ni@rGO/PVA, the catalyst film was ejected manually by using a forceps and washed repeatedly with water and ethanol to get rid of any dye residuals before reuse. The reusability of Ni@rGO/PVA was carried out by using fresh film (1st run) or used ones (2nd–5th runs) in repeated optimum degradation cycles. Fig. 9(f) depicted that Ni@rGO/PVA showed a high catalytic performance in the five reaction runs. The conversion% and activity% of MB after five successive catalytic runs reached 83% and 78%, respectively. However, the conversion% and activity% of MB decreased sharply to about 41% and 33%, respectively, in the next five runs. This might be to the accumulation of the MB molecules at the active sites of the catalyst which in turn lead to a depression in the catalytic efficiency.Ni@rGO/PVA film was successfully prepared as recoverable/separable dip catalyst for oxidative and reductive catalytic reactions. The swelling properties of Ni@rGO/PVA film showed that the GO nanosheets serve as crosslinking agent that restrain the solubility of PVA film yielding highly hydrophilic and insoluble polymeric matrix. Also, GO shows strong affinity toward Ni2+ ions leading a tight fixing of the Ni NPs onto the polymeric matrix. Using GO replaces the usage of hazardous chemical crosslinkers. The XRD, XPS, and Raman results indicated the reduction of GO and Ni2+ to rGO sheets and Ni NPs, respectively, under the action of NaBH4. The SEM result clearly demonstrated the successful formation of Ni nanoparticles onto the GO/PVA film. Ni@rGO/PVA film showed efficient catalytic activity in both reductive and oxidative reactions. After, optimization of operational parameters (pollutant concentration, catalyst dose, reducing/oxidizing agent concentration, and pH value), the calculated k app for the degradation of Cr6+, 4-NP, and MB are 7.29 × 10−3, 23.6 × 10−3, 11.5 × 10−3 s−1, respectively. The recyclability of Ni@rGO/PVA film was investigated by several cycles without significant loss of activity. The prepared Ni@rGO/PVA film was shown to be simple and low-cost, eco-friendly, recyclable, and separable catalyst with potential catalytic activity for a variety of reactions.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This paper is based upon work supported by Science, Technology & Innovation Funding Authority (STDF) under grant number 45556. Also, it is acknowledged the financial support from the in–home project unit, National Research Centre, project No. 12020226.

Graphene oxide was blended with poly(vinyl alcohol) film as a solid support for nickel nanoparticles as a dip catalyst. Simple preparation steps were followed avoiding the use of hazardous materials and/or any surface treatment. Graphene oxide plays the key role in the proposed catalyst by cross-linking poly(vinyl alcohol) film, increasing the affinity of the substrate for adsorbing nickel ions, and increasing the hydrophilicity of the nanocomposite. The prepared dip-catalyst was employed in catalytic reduction and oxidation to get rid of Cr6+, 4-nitrophenol, and methylene blue as water pollutants. A reasonable control over the operational parameters of the catalytic reaction was achieved. Significantly, the free-standing film form of the catalyst enabled the facile separation of the catalyst without release of the nickel nanoparticles. Also, the recyclability of the catalyst was investigated for several runs without any considerable loss of catalyst efficiency. The proposed catalyst fits the environmental and industrial requirements because of its stability, low cost, and activity for wide range of catalytic reactions.

The rapid technological development in recent decades has led to an increase in the population in the world, which in turn has caused a noticeable increase in the consumption of fuels such as jet fuel, gas oil, and gasoline, which contain organic sulfur compounds (OSCs).) and are a major cause of air pollution (Choi et al., 2014). These OSCs are found in many forms like sulfides, thiol, and thiophene with their derivatives which may be described as highly harmful to health and the environment via sulfur oxides (SOx) emission during combustion. Moreover, the presence of sulfur in petroleum products also may be causing corrosion of internal combustion engines, poisoning catalytic converters, and causing air pollution, (Alwan et al., 2021). For all mentioned above the sulfur compounds must be removed or eliminated to allowed limits which pay attention to scientists working towards sulfur removal. There are various desulfurization technologies like hydrodesulfurization (HDS), oxidative desulfurization (ODS), extractive desulfurization, and biodesulfurization etc. for production of low-sulfur fuel. HDS is a classical technology used in large-scale processes which desulfurizes different fuels by using hydrogen at high pressure and Ni-Mo or Co-Mo catalyst under elevated temperature, but the HDS has low reactivity towards benzothiophene (BT), dibenzothiophene (DBT), and its alkylated derivatives (Alwan, 2022). There are many disadvantages to using HDS as follows; it requires severe operating conditions such as high reaction temperature (between 300 and 400 °C), hydrogen at high pressure (30–75 bars), the huge amount of catalyst, use of large reactors, and long residence time which causes high operation cost (Choi et al., 2016) .The ODS technology may be described as a promising technique because it does not need to work at extremely high temperatures, as well as atmospheric pressure is enough to achieve the reaction (Alwan 2021) .The ODS efficiency is a two-step process, first step the sulfur present is oxidized to sulfoxide or suldones in presence of oxidation agent such potassium ferrate, tetra‑butyl hydroperoxide, hydrogen peroxide ozone, molecular oxygen, etc., this oxidation reaction is a selective oxidation, it is oxidized OSCs to its sulfur forms without breaking Carbon-carbon bonds. Among these oxidant agents hydrogen peroxide H2O2 is preferred due to high oxidation reactivity and it maybe consider as green oxidant (environmentally-friendly) as well as its low cost safety, and high selectivity (Choi et al., 2014). The oxidation of sulfur caused increasing in polarity sulfur containing compounds. Thus, the sulfoxides and sulfones are easily removed from oil phase with polar solvent or adsorbents and this is the second step. (Zhu et al., 2012; Choi et al., 2022). The most common used solvents such as acetonitrile, methanol,dimethylsulfoxide (DMSO), acetone and, dimethylformamide (DMF). Using of solvent has many disadvantages; toxicity, disposal, reusability, cost and explosiveness, thus solvent selection may represent challenge for example recovery of DMSO challenge via similar boiling point, while acetonitrile characterized by its high polarity which extract a lots of aromatic, and methanol is a good solvent for sulfones extraction but has density closed to diesel density separation is difficult. The ease of OSCs oxidation depends on electrons densities on sulfur atom. High electron density sulfur atoms are easier to oxidize (Badoga et al., 2018).The ODS process require to use various transition metal oxides catalyst such as titanium, copper, cobalt, manganese, iron, tungsten, molybdenum, vanadium and so on .The metals oxides catalyst need support (carrier) like alumina .Using of synthesized CoMo/Alumina with different Co/Mo ratio for BT and DBT oxidation on fixed bed reactor was showed 30% removal for BT and 90% removal for DBT, and they reported about that using MoOx catalyst supported on alumni is very active but it has faster deactivation rate (Chica et al., 2006). Titanium oxide nanotubes and H2O2 exhibited good activity for DBT oxidation (Lorençon et al., 2014). Tian et al. conducted ODS reaction for removing BT and DBT with H2O2 and phosphomolybic acid supported on silica and they get removal efficiency about 99% (Tian et al., 2016). To promote classical molybdenum based catalyst for ODS reaction of DBT at mild operating conditions M. Yaseen et al. used 2 wt.% loading as promoter to classical molybdenum based catalyst in presence of oxidation system consists H2O2 and formic acid and (Muhammad et al., 2018). There are many workers interested to use carbon and its allotropes as catalyst support via its high chemical and thermal stability as well as its mechanical strength such as grapheme and carbon nanotubes (Alwan, 2022).In this study the molybdenum-based catalyst was synthesized by wet impregnation for activated carbon, the molybdenum oxide represented as active phase while nickel oxide is a promoter because the molybdenum base catalyst lost its activity during oxidation desulfurization reaction so the goal for this study is the effect of adding nickel as catalyst promoter as well as to analyze the effect of some other variables on DBT oxidation to remove sulfur from model fuel. The studied variables are catalyst amount, nickel (Ni wt.%) loading, and initial sulfur concentration while the response is the sulfur removal efficiency. The experiments were designed by applied Box-Bohenken experimental design combined with Response Surface Methodology (RSM).Activated carbon AC (568 m2/g and 0.0062 cm3/ g for specific surface area and pore volume respectively) purchased from the local market was used as catalyst support, ammonium heptamolybdate (NH4)6Mo7O24·6H2O (AHM) with purity 99% (HOPKIN & WILLIAMS), nickel nitrate Ni(NO3)2·6H2O with purity 99% (CHD Ltd.).The catalyst was prepared by wet impregnation AC with nickel and molybdenum oxide from their precursor as follows; AHM and Ni (NO3)2·6H2O are sources for molybdenum oxide and nickel oxide respectively. The molybdenum was loaded 15 wt.%, while the nickel loaded (2, 4, and 6 wt.%) to investigate the impact of nickel content as a catalyst promoter. For impregnation of 10 g from AC, two solutions were prepared as follows; first solution, 2.007 g of AHM salt (as molybdenum oxide source), second solution, 1.0297, 2.0594 and 3.0891 g of nickel nitrate hexahydrate (Ni (NO3)2·6H2O) salt (as nickel oxide source) dissolved in distilled water to get loaded nickel percentage 2%, 4%, and 6% where they are symbolized as 2%NiMo/Ac, 4%NiMo/Ac and 6%NiMo/Ac respectively. These two solutions were added, followed loaded on an AC surface by co-impregnation method to precipitate cobalt and molybdenum oxides. The impregnated AC was dried at 110 ºC for two hours and calcination was done at 400 ºC for four.RSM is a practical procedure used for evaluating the relation between actual experimental results (response) with studied variables (control variables), and this is usually done by combining RSM with factorial design techniques such as central-composite design CCD and Box-Bohenken design BBD. BBD technique can reduce the required number of experiments without decreasing the accuracy of the optimization in comparison with other factorial design methods (Alwan, 2021). The required experiments number to cover the studied variables system according to using BBD is: (1) N = 2 k ( k − 1 ) + r Where N is the number of experiments, k is the number of variables, and r is the replicate number of central points (3–6). BBD stated that the levels of the studied variables were adjusted to only three levels (-1.0, 1) with equal values for the interval between each level, thus for three variables with three levels, the number of experiments was 15−18 depending on the number of replicated experiments number (r in the equation). The catalyst dosage, Ni% loaded in catalyst, and initial sulfur concentration is chosen as studied (controlled) variables on DBT conversion (Table 1 ), the experimental design with using of design expert version 13 as shown in Table 2 .The experiments results for the effects of catalyst dosage (x1), Ni wt.% loaded on catalyst (x2) and, sulfur initial concentration (x3) on oxidative desulfurization were fitted as second-order polynomial, and it can be used to estimate predict values and optimization the system, for three variables where the second-order polynomial represents by equation (2) R % = β 0 + ∑ β i x i + ∑ β i i x i 2 + ∑ β i j x i x j + ε Where R% is predicated response, β0 is the intercept coefficient, β i is the linear effect (slope) of input variable xi , β ij is interaction effect of linear by linear between two input variables xi , and B ii is squared effect.The model fuel (DBT dissolved in n-heptane) was prepared by using three different DBT concentration (400, 600, and 800 ppm); the DBT concentration prepared according to Box-Behnken design BBD. ODS reaction for DBT was conducted under mixing of model fuel at 50 °C in presence of prepared catalyst and H2O2 –CH3COOH oxidation system, where the ODS reaction was examined under the effect of three independent variables; catalyst dosage, Ni% loaded on the catalyst and, initial sulfur concentration with the range for these studied variables which shown in Table 1. The total number of experiments required to cover the three-level for the three-variables system is 15 according to Eq. (1), all experiments were arranged according to Box–Behnken experimental design as shown in Table 2. The oxidation reaction starts by heating 100 ml of model fuel using the magnetic stirrer heater to reach the required temperature (50 °C), 10 ml of hydrogen peroxide, and 5 ml of acetic acid with the needed dosage of catalyst added to model fuel. The reaction stopped after 60 min. Subsequently OSCs were converted into the polar compounds such as sulfoxides and /or sulfones m which separated by using acetonitrile (with 1:1 volume ration) during extraction step. The separation done in separation funnel in which the upper phase was the low sulfur fuel while the below phase was the mixture of oxidative compounds and solvent (acetonitrile).The sulfur content in the final product was measured by X-ray fluorescence (Sulfur Meter model RX-620SA/Tanka Scientific). The DBT conversion (R%) is related with initial sulfur concentration (Si) and final sulfur concentration (So) as in the following equation: (3) D B T c o n v e r s i o n ( R % ) = S i − S o S i × 100 The XRD (Shimadzu Model XRD- 6000 –Japan) patterns for prepared catalysts are shown in Fig. 1 , which contains the pattern for 2% NiMo/Ac (blue curve) and 6% NiMo/Ac (red curve). As the result show the peaks around 2θ equal to 28.9 and 28.77 are attributed to graphite (carbon) at 2% NiMo/Ac and, 6% NiMo/Ac respectively (Wang et al., 2015). There are many peaks for molybdenum trioxide MoO3 at 27.38° at 2%NiMo/Ac, while peaks at 2θ equal to 32.72, and 39.26° at 6% NiMo/Ac (JCPDS No.05–0508), these peaks with sharp shapes indicate that MoO3 have good crystalline and Juehan noted the result closed to this work results (Alwan 2022), (Jegal et al., 2013) and (Dedual et al., 2014). NiMoO4 phase is diffracted at 2θ equal to 23.46 and 23.94 on 2% NiMo/Ac and,6% NiMo/Ac patterns respectively, according to the standard card (JCPDS No. 86–0361), and this good agreement with (Ghosh et al., 2013). Furthermore, the nickel oxide exhibited diffraction peaks at 2θ equal about 43.26 and 54.34 which agreed with Dong et al. (Jang and Park, 2012).The presences of dispersion active metallic oxides (nickel and molybdenum) were further confirmed by EDX (BRUKER Model X Flash 6l10 Germany) elemental mapping as shown in Fig. 2 . Table 3 , shows the DBT conversion for all experiments done according to Box –Behnken design BBD. The DBT conversion ranged between 23 and 71% whereas these results fitted with a second-order polynomial (quadratic model), this equation relate between R% as a function for a function of independent variables (catalyst dosage, Ni% loaded and sulfur initial concentration) and as with respect to actual value below: (4) R % = − 1.516 + 0.989 X 1 − 0.1557 X 2 + 0.00555 X 3 − 0.0121 X 1 X 2 − 0.0011 X 1 X 3 + 0.00132 X 2 X 3 − 0.043 X 1 2 + 0.01018 X 2 2 − 0.000003 X 3 2 The analysis variance ANOVA results for the predicated model as seen in Table 4 , ANOVA gained by Minitab software version 17. The predicted model shows good fitting for actual data due to the high value of correlation coefficient R2 (0.9719) and close value for adj. R2 (0.9213) indicates that the assumed model is reasonably well fitting with actual results. F-value for regression model is 16.77 is greater than tabulated value (F 95, 5,0,05 = 4.77). Based on F-value results, the initial sulfur concentration shows the highest effect on DBT conversion (sulfur removal efficiency) followed by catalyst dosage and Ni% loaded as predicated according to their F-values 89.45, 8.07, and 0.61 for initial sulfur concentration, catalyst dosage, and Ni% loaded respectively. The optimum DBT conversion is 75.74% at 0.5 g, 6% and 700 ppm for catalyst dosage, Ni loaded and initial sulfur concentration respectively.The impact of the studied variable individually and optimization of the studied system were shown in Fig. 3 , the DBT conversion increased with an increase in initial sulfur concentration from 200 ppm until reached near 700 ppm, with further increases in initial sulfur concentration the DBT conversion decreased and this may be because of the presence of the limited number of active sites in a fixed amount of catalyst, in which these limited active sites are insufficient for conversion of BDT (Subbaramaiah et al., 2018) . DBT conversion was raised via the increasing of dosage (amount) of catalyst, which increased the amount of catalytic intermediate produced by reaction with oxidant agent (H2O2), in another meaning when catalyst amount increased will provide more active sites (providing more chance of surface interaction between DBT molecules and catalyst active phase) that responsible on DBT conversion (Yu and Wang, 2013), (Cheng et al., 2015) and (Chu et al., 2010). The impact of nickel weight percent loaded on DBT conversion was decreased with increasing of nickel weight percentage because increasing of amount of nickel loaded leads to less-active surface species formation which maybe caused blockage of some active cites by Ni species (Kim et al., 1996). Zhang et al. (2008) reported that increasing nickel content led to lower nickel dispersion, Figs. 4–6 show the interaction effect for each pair from studied variables.For better understanding the ODS mechanism by H2O2/CH3COOH system in presence of NiO-MoO3/Ac catalyst, by assuming is the presence of NiO as a catalyst promoter, while the MoO3 as an active phase, the reaction initiated by MoO3, involving the hydrolytic cleavage of hydrogen peroxide to produce strong oxidation agents (active hydroxyl radical (OH•) (Ahmad et al., 2021), these active radicals were attack acetic acid to produce peracetic acid, which offers oxygen to DBT to produce DBTO (sulfoxide;contains S = O) and with further attack the DBTO2 (sulfones; contains O = S = O) was produced (Scheme 1 ) . (5) D B T → [ o ] D B T O → [ o ] D B T O 2 In this study, the oxidation reaction ODS for DBT dissolved in n-heptane is done using molybdenum oxide-nickel oxide supported on AC and an H2O2 – acetic acid system as an oxidant agent. The study consists of the investigation of the effect of three parameters which are arranged by combined RSM and Box-Behnken design. The studied variables were catalyst dosage, Ni% loaded, and initial sulfur concentration. Results show that DBT conversion (sulfur removal efficiency) ranged between 23 and 71%, and they were fitted with seconds–order polynomial (high correlation coefficient R2 = 0.9719). These results agreed with many previous studies but the most point considered is the use of nickel oxide for enhancement of the molybdenum-based catalysis activity. In contrast, the using nickel oxide caused decreasing in sulfur removal efficiency and which may mean that the deactivation of the catalyst was happen rapidly. The optimum DBT conversion is 75.74% at 0.5 g, 6% and 700 ppm for catalyst dosage, Ni loaded and initial sulfur concentration respectively.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to acknowledge to Mr. Riyadh Noaman manager of Chemical and Petrochemical Research Center / Corporation of Research and Industry Development /Ministry of Industry & Minerals and Mr. Quraish, Mr. Zuhair for their help in measuring sulfur con-tent to support in this research .

In this study, oxidative desulfurization of dibenzothiophene (DBT) with an H2O2-acetic acid system whereas the catalyst used is molybdenum oxide supported on activated carbon (AC). The effect of loading nickel oxide as a promoter as well as the impact of catalyst dosage and the initial sulfur concentration were investigated. The ranges for these parameters are catalyst dosage (0.5–1.5) g, nickel loading (2–6) wt.% and initial sulfur concentration (400–800) ppm. A Response Surface Methodology (RSM) combined with Box-Behnken design (BBD) was utilized to evaluate the impacts of studied variables; the evaluation consists of the level of order significance of each factor, the interaction effects of parameters was analyzed with Analysis of variance (ANOVA) and determine the optimum conditions for oxidative desulfurization (ODS). Results showed that sulfur removal efficiency from model fuel ranged between 23 and 71%, and these results were fitted with a second-order polynomial model with a high correlation coefficient R2 (0.9719). The optimal condition for DBT oxidation is 0.5 g. Ni wt. 6% and 700 ppm for catalyst dosage, nickel loading, and initial sulfur concentration respectively.

Fuel cell technology is a potent alternative for the production of clean energy. The main fuels which are widely used to power fuel cells are hydrogen, methanol, methane, formic acid or hydrazine. Of particular interest are low temperature fuel cells that are powered by clean hydrogen giving electricity, water and heat. This is because such cell systems powered by the hydrogen offer highly efficient and environmentally friendly energy production technology (Munjewar et al., 2017; Mahapatra et al., 2014; Lenarda et al., 2007). Specifically, the polymer electrolyte membrane fuel cell (PEM), also called proton exchange membrane fuel cells (PEMFC), is one of the most popular types of fuel cell. A drawing of a PEM is shown in Fig. S1 of the supporting information. Nowadays, the polymer electrolyte membrane fuel cell (PEMFC) is the one of the most advanced fuel cells; it can be used in portable electronics, electric vehicles or stationary power plants (Lamy et al., 2009; Wang et al., 2011). PEM-FCs uses a polymer membrane as an electrolyte, which is an ion conductor and contains two electrodes: an anode and a cathode. Catalysts are very important in these systems, platinum being the most common. However platinum catalysts are very sensitive to CO poisoning and if the hydrogen is supplied from a hydrocarbon fuel, it is necessary to eliminate CO from the feed gas (Lu et al., 2016).Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.arabjc.2018.10.002.Fuel cell technology is a potent alternative for the production of clean energy. The main fuels which are widely used to power fuel cells are hydrogen, methanol, methane, formic acid or hydrazine. Of particular interest are low temperature fuel cells that are powered by clean hydrogen giving electricity, water and heat. This is because such cell systems powered by the hydrogen offer highly efficient and environmentally friendly energy production technology (Munjewar et al., 2017; Mahapatra et al., 2014; Lenarda et al., 2007). Specifically, the polymer electrolyte membrane fuel cell (PEM), also called proton exchange membrane fuel cells (PEMFC), is one of the most popular types of fuel cell. A drawing of a PEM is shown in Fig. S1 of the supporting information . Nowadays, the polymer electrolyte membrane fuel cell (PEMFC) is the one of the most advanced fuel cells; it can be used in portable electronics, electric vehicles or stationary power plants (Lamy et al., 2009; Wang et al., 2011). PEM-FCs uses a polymer membrane as an electrolyte, which is an ion conductor and contains two electrodes: an anode and a cathode. Catalysts are very important in these systems, platinum being the most common. However platinum catalysts are very sensitive to CO poisoning and if the hydrogen is supplied from a hydrocarbon fuel, it is necessary to eliminate CO from the feed gas (Lu et al., 2016).The next highly advanced fuel cell is the Solid Oxide Fuel Cell (SOFC). This type of fuel cell can be directly powered by a hydrocarbon stream (e.g., natural gas) without the need for carbon monoxide removing from the feed stream. In this type of fuel cell, the material of the anode or cathode does not have to contain a platinum catalyst. This provides more fuel options. The most important disadvantage of this fuel cell is the high operating temperature. However, it shows high efficiency and stability (Papurello and Lanzini, 2018; Ramadhani et al., 2017; Papurello et al., 2016). Typically, anode materials used in this type of fuel cell contain nickel oxide and others metal oxides supported usually on zeolite such us: Ni-Gd0.1Ce0.9O1.95AFL, Ni-Gd0.1Ce0.9O1.95, NiO-YSZ, NiO-Fe2O3-Ce0.8Sm0.2O2-δ (Gao et al., 2016) (see Fig. 1 ).Several sources of hydrogen are well known to power a fuel cell, which may include: alcohol, hydrocarbons, ammonia etc. Methanol is one of the most promising source of hydrogen because it is the simplest alcohol without C C bond in the molecule and provides a high H:C ratio. These properties indicate that methanol can be easily decomposed to a hydrogen rich mixture. Basically, there are four methods available for hydrogen production from CH3OH:Steam reforming of methanol (SRM) (1) CH3OH + H2O → CO2 + 3H2 Decomposition of methanol (DM) (2) CH3OH → CO + 2H2 Partial oxidation of methanol (POM) (3) C H 3 OH + 1 2 O 2 → C O 2 + 2 H 2 Oxidative Steam Reforming of Methanol (OSRM – combination of SRM and POM) (4) C H 3 OH + 1 2 H 2 O + 1 4 O 2 → C O 2 + 5 2 H 2 It is worth emphasizing that a combination of steam reforming and partial oxidation of methanol is energetically favourable and the OSRM process can run in an auto – thermal manner, without the need to supply any external heat. Previously mentioned properties of methanol indicate that the OSRM reaction can be carried out in the temperature range 150–330 °C without the formation of carbon deposits (Mierczynski et al., 2016; Mierczynski et al., 2016). Typical catalysts used in the reforming of methanol processes are Cu, Ni, Fe, Co, Pd, supported on mono, and binary oxide systems (Mierczynski, 2016; Mierczynski et al., 2017; Mierczynski et al., 2016; Pojanavaraphan et al., 2015; Mierczynski et al., 2013; Abrokwah et al., 2016; Sá et al., 2010; Schuyten et al., 2009; Ahn et al., 2009). It is also well known that binary oxides exhibited the superior catalytic properties compared to the monometallic systems (Maniecki et al., 2009; Maniecki et al., 2009). In addition, promotion of monometallic copper or nickel catalyst by noble metals improves the catalytic activity and selectivity in methanol reforming processes (Lenarda et al., 2007; Mierczynski et al., 2016; Mierczynski, 2016; Mierczynski et al., 2017; Mierczynski et al., 2016). Modification of copper catalyst by ZrO2 cause increase of catalyst surface, stabilize crystallites size of copper, and in the same time protects crystallites against their aggregations. In addition, ZrO2 stabilizes the copper Cu+ ions on catalyst surface (Papavasiliou et al., 2007). Jeong et al. (2006) examined the influence of ZrO2 addition on yield of copper catalysts in reforming of methanol reaction and reported that system containing ZrO2 exhibited an increase of approximately 16% in methanol conversion and a CO molar fraction 7.3 times lower.Although much work has been focused on addition of noble metals or transition oxides to nickel and copper catalysts influence on their catalytic properties in OSRM reaction, there has been no study exploring the possibilities of improving the catalytic activity of nickel catalyst through an activation process carried out in a mixture of 5% H2–95% Ar at various temperatures or the promotion of nickel catalysts by noble metal. To fill these knowledge gaps, we prepared monometallic copper and nickel, and bimetallic Rh(Pd)-Cu(Ni) catalysts supported on various binary oxides in order to determine the most optimal catalytic composition for OSRM and to correlate their physicochemical properties with catalytic activity. In this work, we present how precious metals influence the catalytic and physicochemical properties of nickel and copper catalysts supported on selected binary oxide in OSRM process. The manuscript describes in detail the effect of partial reduction of nickel catalyst on its catalytic properties in the tested reaction. In addition, we studied how changes in the composition of binary oxide support influences on the catalytic and physicochemical properties of the copper catalysts obtained in the OSRM reaction.Monometallic copper and nickel catalysts supported on (ZrO2)x · (Al2O3)y binary oxide supports were prepared by a wet aqueous impregnation method. Binary oxide ZrO · Al2O3 (Zr:Al = 2:1, 1:1, 1:2) systems were prepared by a co-precipitation method. In order to prepare a working range of binary oxides, the following molar ratios of Zr:Al = 2:1, 1:1, 1:2 were used. Aqueous solutions of 1 mol/L zirconium (IV) nitrate and 1 mol/L aluminium nitrate were mixed in appreciate quantities under vigorous stirring at 80 °C. A concentrated ammonia solution was then added dropwise until the pH reached values of between 10 and 11, respectively. Then the mixtures were stirred for another 30 min. The resulting fine precipitates were washed two times in deionised water and then dried at 120 °C for 15 h and calcined for 4 h at 400 °C in air atmosphere. The metal phase (i.e., Cu or Ni) was introduced on the supports using aqueous solutions of copper nitrate (V) or nickel nitrate (V). Copper or nickel loading on the catalyst surface was 20 wt%. The supported catalysts were then dried for 2 h at 120 °C and calcined for 4 h in an air atmosphere at 400 °C. Bimetallic supported catalysts 1% Pd(Rh)–20% Cu/ZrO2·Al2O3 and 1% Pd(Rh)–20% Ni/ZrO2·Al2O3 (Zr:Al = 1:2) were prepared by an impregnation method on the surface of the previously prepared supported copper or nickel catalysts systems.The specific surface area and porosity of catalytic material were determined by the BET method based on low temperature (−196 °C) nitrogen adsorption in a Sorptomatic 1900 Carlo-Erba apparatus. The pore size distributions of the investigated material were defined based on the BJH method. Temperature programmed reduction TPR-H2 measurements of supported catalysts were performed in order to study their reducibility. TPR-H2 measurements were carried out in an automatic apparatus Altamira (AMI-1). The reduction behaviour of all supported copper catalyst systems was studied in the temperature range of 25–900 °C, with a linear heating rate of 10 °C min−1. In each investigation, a sample about 0.1 g was placed in a micro-reactor and was reduced in a mixture of hydrogen in argon stream (5% H2–95% Ar) with a volumetric flow rate of 40 cm3 min−1. The hydrogen consumption rate was monitored by a thermal conductivity detector (TCD). The TPD-NH3 system was used to study the acidity of the catalysts. The temperature programmed desorption of NH3 experiments were performed in the temperature measurements were carried out in a quartz flow micro-reactor using NH3 as a probe molecule. Before all experiments, the catalyst surface was purified in a flow of He at 600 °C for 60 min. After purification, the NH3 was adsorbed on the catalyst surface at 50 °C for 30 min. The temperature programmed desorption of NH3 were performed in the temperature range 100–600 °C using a linear temperature ramp (25 °C min−1). Before each TPD-NH3 experiment, physically adsorbed NH3 has been removed from the catalyst surface. All measurements were performed using IR Tracer-100 FTIR (Shimadzu) spectrometer equipped with a liquid nitrogen cooled MCT detector. Before each experiment, a catalyst was reduced at 300 °C in a mixture of 5% H2–95% Ar mixture. A resolution of 4 cm−1 was used in collecting all spectra. 128 scans were taken in order to achieve a satisfactory signal to noise ratio. The background spectrum was collected at 50 °C after the reduction process of each catalytic material. After the reduction process, a reducing mixture was shifted to a mixture of 1 vol% CH3OH in argon stream and at the same temperature spectra were collected. Powder X-ray diffraction patterns were recorded on a PANalytical X’PertPro MPD diffractometer in Bragg-Brentano reflecting geometry. Cu Kα radiation (λ = 154.05 pm) from a sealed tube was used in the 2Θ angle range 5–90°. The morphology and composition of the investigated catalyst systems were studied using S-4700 scanning electron microscope HITACHI, equipped with an energy dispersive spectrometer EDS. The XPS spectra were recorded for selected catalysts on a Specs SAGE XPS spectrometer using Mg Kα radiation source (hν = 1253.6 eV) operating at 10 kV and 20 mA. The elements present on the sample surface were identified from a survey spectrum recorded over the energy range 0–1000 eV at pass energy of 100 eV and a spectrum acquisition step of 0.5 eV. The areas under selected photoelectron peaks in the spectrum were used to calculate the percentage of atomic concentrations of each species. High-resolution (spectrum acquisition step of 0.1 eV) spectra were collected for pertinent photoelectron peaks at a pass energy of 20 eV to identify the chemical state of each element. All the binding energies (BEs) were referenced to the C1 s peak (285 eV) coming from adventitious carbon to compensate for the effect of surface charging. The analysis area had a diameter of 0.7 mm. Casa XPS software was used during analysis of the high-resolution spectra.OSRM reaction was performed out using a flow quartz micro-reactor under atmospheric pressure in the temperature range 160–300 °C. The following reaction mixture was used in each catalytic test: H2O/CH3OH/O2 = 1/1/0.4 (molar ratio) and the GHSV was 26700 h−1 (calculated at ambient temperature and under atmospheric pressure). The total flow of the reaction mixture was 31.5 cm3/min. Argon was used as a balance gas. The catalytic activity tests were done after two hours of stabilization process performed at each temperature. The mass of the catalysts used in each test was 0.2 g. Before each catalytic activity test copper containing catalysts and all bimetallic systems were activated for 1 h in a mixture of 5% H2–95% Ar at 300 °C. While, the monometallic supported nickel catalyst was activated under the same conditions as well as at a higher temperature of 500 °C using the same reduction conditions. The analysis of the obtained products in the investigated process was monitored using GC systems. Analysis of the organic products (methanol, methane, methyl formate, dimethylether (DME), and formaldehyde) were performed using chromatograph equipped with FID detector and 10% Carbowax 1500 on Graphpac column. While, CO and CO2 concentrations were monitored by GC system equipped with TCD detector (150 °C, 60 mA), and Carbosphere 60/80 (50 °C) column. The hydrogen concentration was monitored also by a GC chromatograph equipped with TCD detector (120 °C, 60 mA) and molecular sieve 5a (120 °C) column. Material balances on carbon were calculated for each run to verify the obtained results. The selectivity results towards hydrogen, carbon monoxide, carbon dioxide and DME formation in OSRM was calculated using Eqs. (5)–(8). While, the methanol conversion was calculated using Eq. (9): (5) S H 2 % = ( nH 2 - o u t ) ∑ p r o d u c t s o f t h e r e a c t i o n ∗ 100 (6) S CO % = ( nCO out ) ∑ p r o d u c t s o f t h e r e a c t i o n ∗ 100 (7) S CO 2 % = ( nCO 2 - o u t ) ∑ p r o d u c t s o f t h e r e a c t i o n ∗ 100 (8) S DME % = ( nDME out ) ∑ p r o d u c t s o f t h e r e a c t i o n ∗ 100 where n CH3OH and n H2 is the molar flow rate of CH3OH and H2, respectively. (9) Conv . CH 3 O H % = n 1 in CH 3 O H - n 2 out CH 3 O H n 1 in CH 3 O H ∗ 100 where nH2-out – molar flow rate of H2 feed out, nCO2-out – molar flow rate of CO2 feed out, nCOout – molar flow rate of CO feed out, n1 in CH3OH, n2 out CH3OH – molar flow rate of CH3OH feed in and feed out, respectively. Organic compounds such as: methane, formaldehyde and methyl formate formation were not detected in the obtained product. Only carbon monoxide, carbon dioxide, hydrogen and DME were formed as reaction products during the OSRM reaction. nH2-out – molar flow rate of H2 feed out,nCO2-out – molar flow rate of CO2 feed out,nCOout – molar flow rate of CO feed out,n1 in CH3OH, n2 out CH3OH – molar flow rate of CH3OH feed in and feed out, respectively. Organic compounds such as: methane, formaldehyde and methyl formate formation were not detected in the obtained product. Only carbon monoxide, carbon dioxide, hydrogen and DME were formed as reaction products during the OSRM reaction.The main goal of this paper was to optimize of the catalyst composition to suit the purpose of the OSRM process. Therefore, in the first step of our catalytic investigations we decided to carry out activity tests for copper catalysts supported on various (ZrO2)x · (Al2O3)y binary oxide systems in order to choose the best carrier. The results of the catalytic activity expressed as methanol conversion and selectivity towards hydrogen and other products are given in Table 1 . The methanol conversion results showed that the most active catalyst among all studied copper systems supported on Zr and Al was the one with the lowest content of Zr. This catalyst showed the highest methanol conversion at both studied temperatures (i.e., 160 and 200 °C) and also high selectivity towards hydrogen formation in the oxy-steam reforming of methanol process. Furthermore, the reactivity tests showed that increasing of the aluminium content caused an increase in methanol conversion and selectivity towards hydrogen formation. It is worth noting that carbon monoxide was not formed during the reaction which is very advantageous from an application point of view. In summary, it is clear that the activity and selectivity of copper catalysts is a function of catalyst composition according to Abrokwah et al. (2016). The authors studied various monometallic Cu, Co, Ni, Pd, Zn and Sn catalysts supported on MCM-41 in reforming of methanol. They reported that the methanol conversion values and selectivity towards main products depend mainly on the active phase of the catalyst used in the process. They also confirmed that supported copper catalysts exhibited the highest methanol conversion value ∼82% and high selectivity to hydrogen formation. Cu/MCM-41 system also showed the lowest selectivity towards carbon monoxide formation (Abrokwah et al., 2016). Based on the obtained activity results for supported copper catalysts we decided to prepared analogous nickel catalysts supported on ZrO2·Al2O3 (Zr:Al = 1:2) binary oxide and test them for the same reaction. The catalytic activity results obtained for Ni/ZrO2·Al2O3 system clearly showed that this catalyst exhibited lower activity in the low temperature range 160–250 °C. The methanol conversion value for this catalyst at low temperature (160, 200 and 250 °C) was below 15%. Increasing reaction temperature up to 300 °C resulted in a significant increase in the methanol conversion value to about 94% and high selectivity towards hydrogen production. In addition, the results of the catalytic activity in oxy-steam reforming of methanol obtained at 300 °C showed that the carbon monoxide was formed as one of the main products of the reaction (CO selectivity = 25%). On the other hand, in the case of the monometallic supported nickel catalyst, we investigated the effect of higher temperature (500 °C) of the activation process carried out in a mixture of 5% H2–95% Ar. The results of the catalytic activity showed that the use of a higher reduction temperature before the reactivity test did not improve the activity of the nickel catalyst. The catalysts exhibited lower methanol conversion at 300 °C and selectivity towards hydrogen formation. In addition, large amounts of the dimethyl ether formed at 250 and 300 °C were observed. Abrokwah et al. (2016) also claimed that Ni/MCM-41 exhibited lower activity in the SRM process compared to the monometallic supported copper catalyst. Further, we attempted to improve the catalytic activity of our systems by introducing metallic promoters. Therefore, in the next step of our reactivity studies we prepared and tested bimetallic Pd-Cu(Ni) and Rh-Cu(Ni) catalysts supported on a previously selected carrier ZrO2·Al2O3 (Zr:Al = 1:2). The results of the activity tests performed in oxy-steam reforming of methanol showed that promotion of monometallic nickel catalysts by Pd or Rh significantly improves the activity. Both bimetallic Pd-Ni and Rh-Ni supported catalysts exhibited higher methanol conversion compared to the monometallic systems at 160 and 200 °C. However, carbon monoxide was formed for both catalysts at high temperature. In contrast, DME formation was not observed during the OSRM process at 160 and 200 °C. The catalytic activity tests performed for bimetallic Pd-Cu and Rh-Cu supported catalysts showed that there was an improvement in the catalytic activity. Notably, carbon monoxide formation was not observed in Pd-Cu and Rh-Cu catalytic systems at the reaction temperatures. The comparison of the catalytic activity obtained for bimetallic Pd-Ni and Pd-Cu supported catalysts showed that both systems exhibited practically the same values of methanol conversion and selectivity towards hydrogen formation. On the other hand, bimetallic Rh-Cu catalysts showed the highest activity and selectivity towards hydrogen formation at 200 °C. The catalytic activity tests also showed that the only undesired products which were formed during the reforming process were carbon monoxide and DME. Based on the results of catalytic activity measurements we further optimized the content of rhodium in Rh-Cu bimetallic supported catalysts. Chang et al. (2012) investigated the catalytic properties of copper CuO/ZnO/Al2O3 (30/60/10) catalysts promoted by noble metals such as: Pt, Pd, Ru and Rh in oxidative steam reforming of methanol and they reported that addition of noble metals improves the methanol conversion during the reaction in all cases but also increases the formation of CO. The catalytic tests performed by authors in OSRM showed that only copper catalysts promoted by platinum prepared by co-precipitation method exhibited higher methanol conversion and low CO selectivity. We also investigated in this work the influence of the rhodium content on methanol conversion and selectivity results towards H2, CO, CO2 and DME. We prepared three bimetallic Rh-Cu catalysts supported on ZrO2·Al2O3 (Zr:Al = 1:2) by an impregnation method and tested for OSRM. The results obtained in oxy-steam reforming of methanol reaction are also given in the same Table 1. The reactivity results clearly indicate that the most active catalyst was the system containing the lowest content of Rh.The reactivity results clearly indicate that the most active catalyst was the system containing the lowest content of Rh. It is also worth mentioning that this catalyst also exhibited the highest selectivity to hydrogen production among of all bimetallic promoted by Rh catalysts at low temperature i.e. 160 °C. While, at 200 °C, the results of selectivity towards hydrogen formation showed that the 0.5% Rh–20% Cu/ZrO2·Al2O3 (Zr:Al = 1:2) catalyst exhibited also high selectivity towards hydrogen formation similar to the copper catalyst containing 1% wt. of Rh. Further increase of the Rh loading in the catalytic system resulted in decrease of the selectivity to hydrogen production.In the next step of our investigations, we determined the Specific Surface Area (SSA) and average pore size for binary oxides and all tested catalysts. The specific surface area results are presented in Table 2 . The results of the SSA measurements clearly show that the binary oxide (ZrO2·Al2O3 (Zr:Al = 1:2)) and catalyst with the highest content of the aluminium among all copper catalysts exhibited also the highest specific surface area. In addition, all investigated catalysts had average pore size below 3 nm.Notably, the 20% Cu/ZrO2·Al2O3 (Zr:Al = 1:2) system had specific surface area value about 20% higher compared to the rest of the monometallic copper catalysts supported on the same carrier. In contrast, the nickel catalysts supported on ZrO2·Al2O3 system showed significantly lower SSA value compared to the copper catalyst supported on the same support. On the other hand, the SSA measurements obtained for bimetallic supported catalysts showed that the promotions of copper or nickel catalyst by noble metals does not cause significant changes in the specific surface area. In the case of the monolayer capacity values the results indicate that the bimetallic systems exhibited slightly higher values. Whereas, the values of the pore radius obtained for all catalytic systems were below 3 nm in all cases.Next, we studied the reducibility of the monometallic and bimetallic supported catalysts. TPR-H2 measurements recorded for copper catalysts supported on various ZrO2-Al2O3 systems are presented in Fig. 2 . The reduction measurements obtained for 20% Cu/ZrO2·Al2O3 (Zr:Al = 1:2) catalyst showed two unresolved reduction effects in the temperature range of 150–320 °C (Mierczynski et al., 2013). The maxima of hydrogen consumption peaks at about 210 °C and 280 °C are associated with reduction of CuO species according to the following scheme (Mierczynski, 2016): Cu2+ → Cu+ → Cu0 The first reduction peak located at 210 °C is assigned to the reduction CuO to Cu2O species. The next peak with a maximum of hydrogen consumption at 280 °C is associated with the reduction of Cu2O species to metallic Cu. In the case of the rest of the supported copper catalysts the same reduction stages were visible on the TPR-H2 profiles as for 20% Cu/ZrO2·Al2O3 (Zr:Al = 1:2) system. Ren et al. (2015) investigated the reducibility of Cu/Al2O3 catalyst modified by ZnO, ZrO2 and MgO. They observed also two reduction stages in the case of all investigated Cu catalysts. TPR-H2 profile recorded for 20% Cu/Al2O3 catalyst modified by ZrO2 showed two unresolved reduction peaks with maxima of hydrogen consumption peaks at 220 and 270 °C, respectively. These authors reported that a first reduction effect was assigned to the reduction of the highly dispersed CuO phases. The second effect is attributed to the reduction of CuO species strongly interacted with the support. They strongly suggested that the modification of 20% Cu/Al2O3 catalyst by ZrO2 improves the dispersion of CuO species on the catalyst surface. This is due to decrease in the interaction between CuO and support surface which also prevents the migration of metallic copper species onto support surface. Zhu et al. (2015) studied also the reduction behaviour of Cu/ZrO2/Al2O3 catalyst calcined in the temperature range 350–650 °C. The TPR-H2 profiles recorded for all investigated catalysts showed, similar as in our case, two unresolved reduction effects located in the temperature range 150–300 °C. These reduction stages were assigned to the two steps of the reduction process described by the following scheme Cu2+ → Cu+ → Cu0. It is worth emphasizing that the Cu/ZrO2/Al2O3 catalyst calcined at 350 °C exhibited slightly higher reduction temperature than catalysts calcined at 450 °C. The TPR-H2 profiles recorded for Cu/ZrO2/Al2O3 catalysts calcined at 750 and 850 °C also showed the reduction effect with the maximum of the hydrogen consumption peak at 370 °C. This reduction peak was attributed to the reduction of the CuAl2O4 spinel structure. Fig. 3 presents a comparison of the reducibility of monometallic Cu and Ni and bimetallic Pd-Cu, Rh-Cu, Pd-Ni, Rh-Ni supported catalysts. The TPR-H2 profile recorded for 20%Ni/ZrO2 · Al2O3 catalyst showed a two-step reduction process (see Fig. 3). The first reduction effect located in the temperature range 350–450 °C is associated with the reduction of unbounded NiO species. The second hydrogen consumption peak located above 450 °C is assigned to the reduction of NiO species differently interacted with support. In the same Fig. 3 the reduction results of bimetallic supported catalysts are also given. The TPR-H2 profiles recorded for all bimetallic catalysts showed that addition of noble metals into monometallic supported copper and nickel catalysts facilitates their reduction. The TPR-H2 profiles of bimetallic systems showed the same reduction stages which were observed in the case of monometallic catalysts but shifted towards the lower temperature range. These shifts confirm the facilitated reduction of copper or nickel oxides after introduction of noble metal. In addition, in the case of the 1%Pd-20%Ni/ZrO2·Al2O3 catalyst a low temperature (maximum at about 90 °C) consumption peak was observed in the TPR profile and was assigned to PdO reduction step. Guo et al. (2014) investigated Ni/ZrO2/Al2O3 catalysts with the different ZrO2 content. The authors reported three various reduction steps assigned to the reduction of α, β and γ nickel oxide species differently interacted with the support present in the TPR-H2 profiles recorded for these catalytic systems. The α species represent the unbounded NiO which are reduced at low temperature (320–450 °C). The reduction effect located in temperature range 450–720 °C was attributed to β species correspond to NiO interacted with the support. The last high temperature effect is assigned to the reduction of NiAl2O4 spinel structure. Furthermore, they reported that increasing the ZrO2 content in Ni/ZrO2/Al2O3 catalysts leads to the growth of α species which are reduced in low temperature. The reduction properties of supported nickel catalysts were also studied by Richardson at al. (Richardson et al., 1994). The authors observed in the TPR-H2 profile high temperature reduction effects, also assigned to the reduction of NiAl2O4 spinel structure. They have reported that the incorporation of Al3+ to the NiO structure or mutual migration of ions leads to the formation of NiAl2O4 spinel structure. These processes take place on the support surface during the heat treatment. The reduction studies performed for all catalytic material clearly indicate that all copper catalysts reduced in two steps and are connected with the reduction of CuO and Cu2O species, respectively (Mierczynski et al., 2015; Mierczynski et al., 2014; Águila et al., 2008). Fig. 4 presents the influence of the Rh content on the reduction behaviour of supported copper catalysts. The observed TPR profiles recorded for all bimetallic supported catalysts indicate that for all investigated bimetallic catalysts two reduction peaks are visible on the TPR curves. These two steps are connected to the reduction of CuO through Cu2O intermediates. It is worth noting that in the case of bimetallic catalysts with low Rh loading, the reduction process took place in the temperature range of 100–300 °C. Additionally, the first reduction stage with the maximum of hydrogen consumption located at about 140 °C had the highest intensity compared to the rest of the bimetallic supported catalysts (see Table 3 ). The highest intensity of the first reduction peak recorded on the TPR-H2 curve recorded for 0.5% Rh-20% Cu/ZrO2·Al2O3 catalyst means that this system is the easiest reduced catalyst.To further explain the differences in activity we also studied the phase composition of monometallic 20% Cu/ZrO2·Al2O3 and 20% Ni/ZrO2·Al2O3 catalysts being after various treatments. We studied the phase composition of the catalysts calcined in an air atmosphere for 4 h at 400 °C and catalysts after reduction in a mixture of 5%H2–95%Ar at 300 °C for 1 h and reaction performed in oxy-steam reforming of methanol. The XRD results are given in Figs. 5 and 6 . The X-ray diffraction studies were used to determine the changes of the phase composition after various treatments and in order to indicate the interaction between an active phase component and the support. X-ray diffraction curve recorded for 20% Cu/ZrO2·Al2O3 catalyst being after calcination confirmed the amorphous nature of the ZrO2 (see Fig. 5). The XRD diffraction pattern shows diffraction peak positioned between 30 and 35 theta angles which was attributed to amorphous ZrO2. While, the diffraction peaks positioned at 36, 38, 48 and 62° were assigned to the CuO phase. In the same XRD curve, γ-Al2O3 phase was visible at 2θ angles = 46, 67 and 68°. Whereas, the XRD curve recorded for the same catalyst reduced at 300 °C showed the occurrence of the diffraction peaks assigned to metallic copper and wide diffraction peak positioned between 30 and 35 2θ angle assigned to amorphous zirconia. Fig. 6 present the phase composition studies of 20%Ni/ZrO2 · Al2O3 catalyst. The XRD curve recorded for 20%Ni/ZrO2 · Al2O3 catalyst calcined in an air atmosphere at 400 °C shows diffraction peaks positioned at 2θ angles = 36.43, 63, 75 and 79° which are attributed to nickel (II) oxide phase. The XRD pattern recorded for this catalyst also showed a wide XRD peak attributed to amorphous zirconia. However, in the case of the same nickel catalyst after being reduced the diffraction curve showed the presence of peaks assigned to metallic nickel (2 theta angles = 44.52°, 76°), NiO phase (2θ angles = 36.43, 63, 75 and 79°) and amorphous zirconia. Diffraction curve recorded for the reduced nickel catalyst confirmed its partial reduction. The existence of other phases in the diffraction curve was not confirmed by the XRD technique.In order to elucidate the differences in activity measurements in OSRM process for bimetallic Rh-Cu supported catalysts the XPS high-resolution spectra of the binding energies between 920 and 970 eV were recorded and the results are given in Fig. 7 and Table 4 . The performed surface analysis of the supported copper catalysts showed that in the investigated binding energy range several peaks were visible. Photoelectron peaks visible in the XPS spectrum can be assigned to metallic copper and copper in first and second oxidation states (Kulkarni and Rao, 2003). The presented on each spectra binding energies bands located at 936, 934.5 and 932.4 eV were assigned according to Ertl and co-workers (Ertl et al., 1980) to Cu2+, Cu0 and Cu+, respectively.The peaks located at about 943 and 963 eV are satellite peaks and are characteristic only of Cu2+ species. The detailed analysis of the presented data gave evidence that increasing of the Rh content in the investigated catalyst to 1 and 2% wt. of Rh leads to higher content of metallic copper species present on the catalyst surface (see Table 4).Cu0 and Cu+ are active centres in the reaction of methanol reforming, and their number and ratio affect on the catalytic activity. Oxidation reaction of CH3OH to CH3O− occurs on metallic copper, while the oxidation of CH3O− takes place on Cu+ in order to formate species creation causing an increase in the conversion of methanol. The presence of Cu+ forms on the catalyst surface leads to greater stability of the catalyst in the process of steam reforming of methanol compared to metallic copper which sinters much easier than in the case of Cu2O species due to its higher Tamman temperature (Mierczynski et al., 2013). In our case the concentration of the Cu+ species on the catalyst surface was the highest for the 2%Rh–20%Cu/ZrO2·Al2O3 catalyst. The role of Cu0 and Cu+ centres in the reforming of methanol process is still unclear but based on the results of the copper species concentration on the catalysts surface it can be assumed that system with the lowest ratio between Cu0 and Cu+ exhibited the highest activity in OSRM process. In addition, these results also confirmed that the occurrence of Cu0 and Cu+ species and their ratio is a critical parameter to achieve highly active systems in OSRM (Kulkarni and Rao, 2003). These results agree well with our previous work (Mierczynski et al., 2013) performed for copper catalysts supported on Zn-Al containing systems tested in steam reforming of methanol reaction. Similar results of the catalytic activity were also confirmed by other authors (Oguchi et al., 2005). The increase of metallic copper content in the catalyst surface after reduction agree well with the temperature programmed reduction results carried out for bimetallic catalysts. This also confirmed that increase of the Rh content in the bimetallic catalysts facilitates the reduction of a copper oxide species present on the catalyst surface.In order to explain and understand the difference in activity and selectivity results in the OSRM process, we carried out acidity measurements for all catalytic material. The acidity measurements performed for supports, monometallic and bimetallic supported catalysts are given in Table 5 . The results show that all investigated systems exhibited three kinds of the acidic centres on their surface namely weak, medium-strong and strong acid sites. The acidity measurements performed for supported copper catalysts confirmed that the highest total acidity, calculated based on the surface under the peaks, had the system with the highest Al content. This result also indicates that catalysts which showed the highest activity also exhibited the highest total acidity among of all monometallic supported catalyst. The results obtained in this work agree well with our previous investigations (Mierczynski et al., 2016; Mierczynski et al., 2017). The acidity measurements showed that nickel 20% Ni/ZrO2·Al2O3 (Zr:Al = 1:2) system exhibited lower total acidity compared to the monometallic copper catalyst supported on the same support.While, in the case of the bimetallic supported catalysts also their high total acidity was detected, what suggest also high activity. This tendency can be easily explained by the fact that acidic sites play crucial role during the OSRM process. These centres are indeed responsible for stabilizing of intermediates such as methoxy, monodenate and bidentate formate species and even carbonates, which are then transformed into the main products CO2 and H2 (Mierczynski et al., 2016; Hereijgers and Weckhuysen, 2009). These findings are in agreement of other published studies such as this conducted by Hereijgers and Weckhuysen (2009).To reinforce the observed activity results and hypothesis concerning the important role of the acidity centres during OSRM process, additional experiments were conducted by FTIR and shown in Fig. 8 . As we can easily distinguish on all presented spectra that during the sorption process carried out at 50 °C the formation of methoxy (peaks at 2995, 2936, 2919, 2825, 1470, 1443, 1350, 1200 and 1020–1100 cm−1), formate (peaks at 2925, 2850, 1620, 1364, and 1350 cm−1) and carbonate species (peaks at 1620, 1570–1440, and 1220 cm−1) are formed on the catalyst surface. The presented results showed that in the case of the most active systems (·20% Cu/ZrO2Al2O3 (Zr:Al = 1:2)) the highest intensity of the IR bands assigned to methoxy, formate and carbonate species were detected. These results agree well with the hypothesis presented above that catalysts which contain the largest number of acidic centres on its surface has the highest sorption properties with respect to methanol at the studied temperature range. The sorption of methanol is one of the important stages during the oxy-steam reforming process. It is also worth mentioning that catalytic systems which showed the highest intensity of bands assigned to methoxy and formate species during the adsorption process had also the highest activity in the oxy-steam reforming of methanol process.SEM-EDS measurements were also performed for monometallic and bimetallic copper and supported nickel catalysts. This useful technique allows determining the morphology and composition of the catalyst surface. SEM images and EDS spectra collected for the investigated mono- and bimetallic catalysts supported on the selected support ZrO2·Al2O3 (Zr:Al = 1:2) were present on Figs. 9–11 . Fig. 9 and Fig. 10 presented the images and EDS spectra collected for monometallic supported copper and nickel catalysts calcined in an air atmosphere at 400 °C for 4 h, respectively. The presented data clearly confirms the composition of the investigated catalysts. In both spectra the occurrence of the same elements such as Zr, Al, O were confirmed on the catalyst surface. In addition, Cu and Ni were also detected for monometallic copper and nickel catalysts, respectively. Analogical measurements were also performed for bimetallic supported catalysts and the results are given on Fig. 11. The analysis of the bimetallic catalyst confirmed the presence of the same elements which were found on the surface of monometallic systems. The only difference in the case of the bimetallic supported catalyst was the presence of rhodium and palladium for the appropriate bimetallic system.In summary, we prepared monometallic and bimetallic copper and nickel catalysts supported on binary oxides by an impregnation method and tested in oxy-steam reforming of methanol in order to determine the optimal composition of the catalyst. The physicochemical properties of the catalysts were investigated by TPR, BET, XRD, FTIR, SEM-EDS and XPS techniques and the obtained results were correlated with the reactivity results obtained in OSRM process. We found that the activity and selectivity of the tested systems are strongly dependent on their acidity and sorption properties in relation to methanol. The reactivity results confirmed that the highest active systems were copper catalysts supported on ZrO2-Al2O3 (Zr:Al = 0.5) binary oxide promoted by noble metals such as Pd or Rh. These catalysts showed the highest specific surface area, the highest number of acidic centres on their surfaces. The reactivity results obtained for bimetallic copper containing systems also confirmed that system with the lowest ratio between Cu0 and Cu+ exhibited the highest activity in OSRM process. Furthermore, these results also confirmed that the occurrence of Cu0 and Cu+ species and their ratio is a critical parameter to achieve highly active systems in OSRM. In addition, copper catalysts showed higher activity and selectivity towards hydrogen formation in oxy-steam reforming of methanol compared to the monometallic supported nickel catalysts. Furthermore, the reactivity measurements carried out for monometallic supported nickel catalysts confirmed that the pre-treatment process before the activity tests has great influence on the reactivity results in oxy-steam reforming of methanol. The supported copper or nickel catalysts described in this work have an application potential in fuel cell technology, especially in Solid Oxide Fuel Cell technology owing to their high efficiency towards hydrogen generation.This work was partially funded by Polish Ministry of Science and Higher Education within the “Iuventus Plus” Programme (2015–2017) (project no.0305/IP2/2015/73). Magdalena Mosinska thanks the Lodz University of Technology for a scholarship (Własny Fundusz Stypendialny PŁ programme, W-3D/FMN/10G 2018).I would like to thank Mr A. Kedziora for help in the research (BET measurements) carried out in the framework of the work.

Monometallic copper, nickel and bimetallic Pd(Rh)-Cu(Ni) catalysts supported on a binary oxide containing various content of ZrO2 and Al2O3 were prepared by impregnation method. Their physicochemical and catalytic properties in oxy-steam reforming of methanol reaction (OSRM) were extensively investigated. Selecting an optimal composition of the catalyst for the OSRM process was the main goal of this work. The influence of zirconia content on the reactivity and physicochemical properties of supported copper catalysts in OSRM was also studied. The reactivity measurements showed that the supported copper catalyst was more active than the nickel catalyst. The catalytic measurements showed that the catalyst properties depend on their surface composition, acidity and adsorption properties. High selectivity of supported copper catalyst with composition 20%Cu/ZrO2·Al2O3 (Zr:Al = 1:2) towards carbon dioxide and hydrogen was confirmed. In addition, the promotion effect of palladium and rhodium on the activity of monometallic supported copper and nickel catalysts in OSRM was confirmed. The most active system in the OSRM process was 0.5%Rh-20%Cu/ZrO2·Al2O3.

Due to increasing evidence of global warming in the present century, scientists at the UN Intergovernmental Panel on Climate Change have reached a consensus for reduction of greenhouse gas emissions, especially carbon dioxide, to the atmosphere [1–8]. This has also prompted steering committees of industrialised countries to assess their energy strategies based on mitigation of greenhouse gas emissions [9–14] (see Table 1 ).Extensive literature has covered on the various alternatives for cleaner energy sources [15,16,227–229], smart drilling techniques [17–20], efficient fracturing technologies [21–25,230–232], usage of nanoparticles [26–31,233], their economic aspects and advantages to mitigate CO2 emissions and reduce environmental pollution [234–241]. Among the list of proposed alternative energy sources, hydrogen appears to be the most promising large-scale fuel due to its efficient storage over time and clean combustion [2,244]. Recent, there has been enormous interest in hydrogen and it's been increasing rapidly due to its potential applications in fuel cells. They also serve as an excellent replacement to batteries in the field of portable electronics, internal combustion engines as well as power plants. The demand for hydrogen in the most important sector of road transport is depicted in Fig. 1 , which illustrates that the annual hydrogen demand is projected to surge from 25 tons in 2020 to 945.5 thousand tons in 2045.Furthermore, the soaring demand for hydrogen in Japan [33] can be illustrated in Fig. 2 . This shows a gradual increase in hydrogen demand from 2015, which will increase rapidly to 21 million tons in 2035.Additionally, electric engines in any vehicle are energised by electricity from fuel cells, which is generated by conversion of clean and environmentally friendly hydrogen (and oxygen from the air) in fuel cells. A schematic diagram representing hydrogen supply from various sources, and its applications, are illustrated in Fig. 3 .It has been established that CO2 emission levels [35] to the atmosphere can be significantly decreased by substitution of traditional fuels such as diesel, gasoline and carbon with higher (H/C) ratio fuels such as natural gas or biomass, as shown in Fig. 3. Therefore, production of hydrogen from hydrocarbons is regarded as the most economic and efficient way of achieving a significant degree of reduction in the emissions of greenhouse gases. Natural gas is a non-renewable energy source; essentially a blend of lighter hydrocarbons existing in the basement of gas accumulations present in porous rock which might or might not be associated with oil. It is mainly constituted of saturated hydrocarbons, mainly methane, with butane and propane in insignificant quantities, and other compounds composed of inorganic gases. Production of synthesis gas comprising of a mixture of CO along with purified H2 being obtained from natural gas by using various catalyst and is currently the most preferred choice (Table 1). With the advent of the hydrogen economy, there has been an increased focus on the transformation of petroleum gas into more ecologically friendly hydrogen fuel.The Global carbon dioxide emissions from various industrial processes and fossil fuel combustion have been estimated to be around 35.7 billion tons [36] annually, which has contributed to increased global warming [242,243]. Therefore, it is imperative to develop clean technologies for the utilisation of fossil fuels [245–247] and to introduce alternative greener fuels for inhibiting the adverse effects of greenhouse gas emissions and subsequent climatic changes. Among various alternative fuels, hydrogen [37] is considered to be a sustainable energy carrier and offers near zero end-use emissions of greenhouse gases and pollutants [38].For the creation of clean fuel, like hydrogen, natural gas needs to undergo a catalytic process described as natural gas reforming. Reforming is the most common technique used in industries for production of synthesis gas via through one of three reforming processes i.e., partial oxidation of methane (POM) [39], steam reforming of methane (SRM) [40–48] and CO2 reforming of methane (DRM) [49–55]. SRM is a fully developed generation technique which utilises steam at high temperature (700–1000 C) for the production of H2 from natural gas. During SRM, CH4 interacts with steam with pressures ranging from 3 to 25 bar using catalyst to produce H2, CO and a moderate quantity of CO2. Eventually during the water-gas shifting reaction, steam and CO2 interact to generate CO and more H2 using an efficient catalyst. Steam reforming of methane requires rigorous energy input because of its endo-thermicity and higher H2O/CH4 ratio which results in better yields of H2, thereby making the SRM process energetically unfavourable [56–62] and accelerates the catalyst deactivation process [51]. On the other hand, DRM [63–69]can be utilized for the generation of syngas from methane and is valuable for the immediate transformation of CO2 into different compounds, Equations 1 to 8 (1) SRM: CH4 + H2O → CO + 3H2 = ΔH° 298, − 206 kJ/mol (2) POM: CH4 + 1/2O2 → CO + 2H2 = ΔH° 298, −38 kJ/mol (3) DRM: CH4 + CO2 → 2CO + 2H2 = ΔH° 298, − 248 kJ/mol POM can deliver syn gas with a H2/CO proportion of 2.0. Nonetheless, controlling this process is an arduous task due to the danger associated with explosions [70] and the presence of hot spots Also, Partial oxidation of methane needs an air separation unit (ASU), which markedly impacts the expenses associated with the reforming plant. Because of these disadvantages associated with POM, combined steam and dry-reforming of methane (CSDRM), where H2O is used in conjunction with CO2, has been considered as a worthwhile strategy for the mass production of syn gas with a H2/CO proportion of 2.0 [71,72]. The CSDRM can generate syn gas with flexible H2/CO proportions, which can be effectively controlled by modifying the feed gas (H2O, CO2 and CH4) composition. The utilized procedure is alternatively known as bi-reforming (BRM) where a 3/2/1 proportion of CH4 along with CO2 and steam produces a gas blend with basically a 2/1 proportion of H2 to CO. This formed gas is also called ‘met gas’ to underline its distinction from broadly utilized syngas blends of different H2/CO proportions. The formation of syngas with this ratio has potential applications in Fischer-Tropsch operations for the preparation of long hydrocarbon chains [73–75]. as well as in the production of methanol [76–78].Furthermore, bi-reforming [79–94] of CH4 has captivated massive interest from both environmental and industrial perspectives. CO2 and CH4 are the most abundant carbon-containing, ozone-depleting substances from an environmental perspective, which can be used successfully in this reaction and can undergo conversion to useful chemical products. In reality, the combination of both steam and dry reforming provides a more pragmatic route for enhancing the H2/CO ratio compared to the introduction of CH4 [95–97]. Additionally, this method possesses the merit of producing synthesis gas by using methane and carbon dioxide which are coined as greenhouse gases.It has been reported [98] that at lower temperatures higher conversion of methane can be achieved in the bi-reforming process. In addition to the operating conditions, catalysts also play a crucial role in bi-reforming reactions. One of the most important advantages of bi-reforming [85,98–104] is that the consumption of major greenhouse gases occurs, thereby creating a significant environmental impact. These gases are water vapour, which accounts for 36–70% of the feed gas, CO2 at 9–26%, CH4 at 4–9% and ozone (O3) for the rest (3–7%) [105]. Hence, there has been a renewed interest in the application of these gases via bi-reforming of methane towards the production of value-added chemicals that are useful for both scientific and industrial communities.Additionally, bi-reforming technology can be regarded as a method for enhancing the caloric value of biogas, which is composed of CO2, H2 and CH4 through the solar reforming process [106,107]. One of the biggest stumbling blocks for the methane reforming process is related to the sudden catalyst deactivation, which might be due to sintering and coke formation on the active sites [108,109]. CH4 decomposition (Eq. (3)), CO disproportionation (Eq. (4)) and CO reduction (Eq. (6)) are the primary processes that lead to coke formation. The reaction in Equation (3) shows an endothermic reaction that is highly favourable at higher temperatures and lower pressures, whereas Equations (4) and (5) are exothermic in nature and favoured at lower temperatures [110] and higher pressures through the reverse water gas shift reaction (Eq. (6)). (4) CH4 ↔ C + 2H2 = ΔH, − 74 kJ/mol (5) 2CO↔ C + CO2 = ΔH, −172 kJ/mol (6) CO + H2 ↔ C + H2O = ΔH, −131 KJ/ mol (7) CO + H2O ↔ CO2 + H2 = ΔH, − 41 kJ/mol (8) 3CH4 + CO2 + 2H2O ↔ 4CO + 8H2 = ΔH, +220 kJ/mol Since catalysts deactivation is caused by formation of coke from the above reactions (4), (5) and 6, hence, it is desirable to establish promising catalysts that demonstrate greater selectivity, excellent stability and activity during the production of syngas. Several investigations [49,50,111], have been reported for assessing the most suitable catalyst for syngas production employing different technologies. Common catalysts that have been used in reforming reactions include catalysts such as copper, nickel supported by transition metals and other supported noble metal catalysts such as ruthenium, platinum, rhenium. Several noteworthy reviews reporting on various innovations recorded in catalyst development for DRM reactions have mainly focussed on catalysts configurations [112], the influence of process parameters [50], noble metal catalysts [49], coke deposition and management [111], development of oxygen carriers in chemical looping [113], Ni and Ni-based catalysts [51], low temperature dry reforming [114], and advances in synthesis of catalysts with mesoporous SBA-15 support [115]. Fidalgo et al. [36], conducted a review on carbon black catalysts and activated carbon which have the unique characteristic of operating without being deactivated by carbon deposition. The catalysts role on methane decomposition and carbon dioxide reforming of CH4 was assessed, and the characteristics of carbon deposits during CO2 reforming of CH4 were listed. The influence of nanocatalysts [38] on the oxidative coupling, steam reforming and CO2 reforming of CH4 has been previously reported, which suggested that methane conversion over a nanocatalyst occurred significantly than the ordinary catalyst and there existed no interdependence between the average particle size of nanoparticles and the conversion of methane.Pakhare et al. [49], reviewed DRM for catalysts based on metals such as palladium, platinum, rhenium and ruthenium, which involved the role of these elements on the mechanism, deactivation, kinetic behaviour of these catalysts. Abdullah et al. [51], conducted a comprehensive review on the potential of nickel based catalysts employed during syngas production using dry reforming process. Their result suggested that strong metal support interactions were dependent on the catalyst supports and these factors were responsible for highest coke resistance, high thermal resistance and greater stability. The authors also examined the synthesis of catalyst supports from cellulosic materials and stressed the enhanced catalytic activity of the cellulose in the DRM reaction due to its superior mechanical strength and distinct structure.However, to the author's knowledge there is a lack of comprehensive literature on the synthesis, characterisation and the role of catalysts and their promoters in the generation of synthesis gas during bi - reforming of methane (BRM). Therefore, the present review encompasses in details the role of various catalysts: Ni-based, Co-based, Ru-based, mesoporous and La-based, on BRM process. Additionally, the review describes the recent progress relating to the most relevant topics on catalysts used in bi-reforming technology.Though Ni-based catalysts [116] are inexpensive, they display exhibit superior performance in comparison to precious metal catalysts. Nevertheless, sintering and formation of carbon affects the sudden deactivation of catalysts. Since bi-reforming employs low S/C ratios for adjusting the H2/CO ratio, hence catalysts involving Ni undergo deactivation by carbon deposition [76]. Two main methods have been documented to diminish the deactivation of catalyst due to formation of coke. One method described the effect of promoters such as lanthanum [117], cerium, magnesium and calcium [118,119], on the characteristics of the catalysts during the reforming process and other was aimed at controlling the particle size at the nano-level in the active metals.It has been established that during bi-reforming [120–122] and DRM, deactivation of Ni supported catalysts occur due to coke formation. Hence, it is highly imperative for the development of most active and stable catalysts in bi-reforming. Several authors [120,121,123], have developed nickel catalyst of high activity and stability supported by Ce–ZrO2, ZrO2 and MgO during the DRM process. Several literature studies on the effect of nickel catalysts and their promoters during bi-reforming of methane have been documented in the present review. For example, Roh et al. [121] employed numerous supported Ni catalysts during bi-reforming reactions for production of syngas having H2/CO = 2. The supported Ni catalysts were prepared by incipient wetness method with Ni(NO3)2. The Ni catalysts were supported by small nanoparticles of ZrO2 or MgO which were highly active and stable for BRM. Fig. 4 illustrates scanning electron microscopic (SEM) images of catalysts used in the reaction and the coke formation observed when subjected to 800°C. The authors observed that the degree of carbon formation and shape varied with different catalysts. Ni catalyst with MgO–Al2O3 as support generated filamentous coke but of insignificant intensity (Fig. 4a). However, Ni catalyst with MgO as support generated a lot of coke from the filaments (Fig. 4b) during occurrence of BRM reaction while Ni/ZrO2 exhibited a worm-like coke feature (Fig. 4c). Similar shape of coke (Fig. 4d) was observed for Ni/CeO2 catalyst. Nevertheless, Ni/α–Al2O3 generated rod shaped-like coke (Fig. 4e).The authors [121] also made a comparative study between the Ni/MgO–Al2O3 catalyst and the commercial Ni catalyst supported with MgAl2O4 (Fig. 5 ) at various temperatures. The methane conversion was 83% and the conversion of carbon dioxide was 71% in presence Ni catalyst supported by MgO–Al2O3 at 700°C, while CO2 and CH4 conversions with Ni/MgAl2O4 were both found to be 20% lower during BRM of methane. Commercial Ni/MgAl2O4 was used as a reference catalyst. Also, coke formation was more severe with the commercial Ni/MgAl2O4 catalyst than with Ni/MgO–Al2O3, which was attributed to the efficient dispersion of Ni [124] supported MgO–Al2O3. The high activity and stability of Ni/MgO Al2O3 catalyst was attributed to the beneficial role of MgO which resulted from basic property, fine dispersion of nano-sized Ni and strong interaction of Ni to the support.From Fig. 6 it was clearly observed that the CH4 conversion was highest for Ni catalyst supported by MgO Al2O3 and approximately 90% methane underwent conversion which continued for 2 h. However, rapid catalyst deactivation occurred in the case of Ni α-alumina catalyst with changes in time attributed to carbon formation. Moreover, Ni/MgO catalyst exhibited around 60% CH4 conversion and Ni/ZrO2 demonstrated around 70% CH4 conversion. Both the catalysts were found to be highly stable during the reforming process. Nevertheless, the conversion of methane for Ni catalyst supported by CeO2 was initially 57% which decreased to 50% followed by its saturation. Results revealed that Ni/MgO–Al2O3 possessing lowest nickel oxide crystallite size exhibited highest stability along with higher CH4 conversion with time on stream.Ryi et al. [125] conducted tests over a catalytic nickel membrane during bi-reforming of methane for a shorter residence time of 120 ms for various CO2/H2O ratios in the range of 0–1.0, along with (H2O + CO2)/CH4 ratio of around 3.0 in the reactant feed for temperatures ranging from 923 to 1023 K.The purpose of this study was to examine the performance of bi-reforming of methane over a catalytic nickel membrane for the GTL (gas to liquid) process. GTL process possess two advantages. One is that carbon formation is reduced due to the oxidation of carbon precursor species and a desirable H2/CO can be achieved by adjusting CH4/H2O/CO2 ratio in the feed stream. Porous wall of catalytic nickel membrane was chosen for reforming studies since hydrogen passed through the catalytic nickel membrane was faster than the other gases because of viscous and Knudsen flow. Generally, the catalyst that contained relatively small size pore was more affected by internal diffusion than the one which has large sized pores.The results revealed that the change in the feed ratio of CO2/H2O strongly affected the conversion of methane and furthermore an increase in the feed ratio of CO2/H2O at a temperature of 923 K (Fig. 7 ) decreased the methane conversion. The authors noted a very high conversion of methane in the range of 92.7–96% above 973 K, when the CO2/H2O feed ratios were in the range of 0–1.0 during bi-reforming of methane. The authors ascertained that with change in the CO2/H2O ratio during the reforming reaction, a change in H2/CO also occurred. Hence H2/CO molar ratio obtained at 973 k were 8.1, 5.7, 3.7 & 2 when the molar ratio of CO2/H2O were 0, 0.11, 0.33 and 1.0. However, increase in temperature to 1023 K changed the H2/CO molar ratio to 7.5, 5.3, 3.4 and 1.8 respectively for similar values of H2O/CO2. Additionally, the CO2 registered an increase with increasing temperature attributed to CO2 reforming of methane occurring at higher temperatures, which remained almost constant at ≥ 973 K attributed to the limitations of CH4 as the reacting species [125].Al-Nakoua and El-Naas [126] experimented with different molar proportions of H2O/CH4 and CO2/CH4 in a detachable reactor covered by catalyst B Nickel (33%)−Chromium (5.6%)−Barium (11%)/La2O3 (19%) and catalyst A which represents Ni (49%)/Al2O3 (51%). The author observed that rapid carbon deposition was observed at 700°C, 1 atm during dry reforming of methane. However, when CO2 reforming was performed in conjunction with steam reforming reaction in thinner channels deposited with a thinner layer of catalyst, a reduction in carbon deposition was noticed on the surface. The authors determined the equilibrium compositions of the CH4 and CO2 reactants, which are shown in Figs. 9 and 10 at pressures of 1, 2, 3, 4, 5, 10, and 20 bar respectively. Fig. 8 shows that CH4 conversion was highest (85–90%) at pressures ranging from 1 to 3 bar and at temperatures in the range of 810−900°C.Similarly, CO2 conversion (Fig. 9) was found to be above 80% in the pressure range of 1–3 bar and temperatures varying from 840 to 900°C. SEM studies (Fig. 10) revealed that cracks formed in catalyst A possessed a length of 250 μm and width of about 10 μm, whereas for catalyst B the cracks were formed with a width of 20 μm and was spread up-to a certain length where the cracks were interconnected. Furthermore, the results [126] established that there was a significant improvement in the catalyst stability when the H2/CO was around 2.2 during bi-reforming of methane. These conditions were appropriate for Fischer-Tropsch applications and synthesis of methanol. Additionally, the authors observed a five-fold increase in the resistance of coke formation displayed by Ni/Al2O3 catalyst An upon addition of Cr, Ba, and La2O3 during a continuous reaction time of 140 h. The SEM results were also supported by EDX analysis.The results of catalyst activity test on Ni/Al2O3 in the ratio of 1:1 for Catalyst A is represented in Fig. 11 . The catalyst film exhibited 50% conversion of CH4 and 15–10% conversion of CO2 when the reaction was continuously operated for 24 h at 630°C. Furthermore, with increase in inlet pressure from 1 to 23 psig, carbon deposition was noticed. The flow rate of CO2 was 0.2 mol/h and the flow rate of CH4 was 0.8 mol/h and the steam: carbon ratio was 0.51. However, Ni–Cr–Ba/La2O3–Al2O3 (Catalyst B) displayed CH4 conversion in the range of 50–75% and conversion of CO2 increased from 20% to 60%.The authors found a reduction in the conversion percentage (Fig. 12 ) of methane and carbon dioxide when the reactor pressure of steam was increased up to 42 psig during continuous operation from 25 to 90 h. However, the conversion percentage of methane and carbon dioxide underwent an increase with further increase in temperature. Since catalyst deactivation has been caused by carbon deposition, hence suppression of coke formation was important. This was only achieved by optimization of the H2O/CH4 and CO2/CH4 and feed ratios.Formation of coke is usually attributed to the following reactions 9 and 10: (9) 2CO ↔ C + CO2 (10) CH4 ↔ C + 2H2 Major amount of coke formed in the temperature range of 850 to 900°C [127] resulted from disproportionation reaction involving carbon monoxide (reaction 9) and pyrolysis of methane (reaction 10).Son et al. [128], observed that Ni/γ-Al2O3 catalyst was rendered stable by pre-treatment with steam at a temperature operated at 850°C. Ni/Al2O3 based catalysts are relatively cheap because precious metals are not used and these catalysts can operate stably with high activity under excess steam. Ni/γ-Al2O3 catalyst used in this study was prepared by incipient wetness method. Thermodynamically, the catalyst promoted very high conversion of CH4 (98.3%) and CO2 (82.4%) when subjected to bi-reforming of methane for 200 h and resulting in H2/CO ratio of 2.01. Furthermore, the results revealed that the conventional catalyst system produced 15.4% coke after 200 h while the mass of carbon deposited was around 3.6% for catalysts exposed to steam. This novel steam pre-treatment technique significantly increased the resistance towards carbon formation in the presence of catalysts, thereby improving both long-term stability and activity.Transmission electron images of fresh untreated Nickel Aluminium and fresh steam-treated NiAl (WNiAl) catalysts shown in Fig. 13 (a) and (c) revealed that it was difficult for distinguishing nickel nanoparticles dispersed in the Nickel Aluminium catalyst, owing to their small size. However, the WNickel Aluminium catalyst showed the presence of distinct Ni nanoparticles in the range greater than 10 nm. Severe carbon deposition (Fig. 13b) was noticed for Ni/Al catalysts treated with steam for 200 h. The shape resembled to wire-type resembled carbon and the size of the nanoparticles were enhanced from 4.2 nm to 23.5 nm from Hydrogen chemisorption measurements. Nevertheless, carbon coke with wire typed shape did not appear in the WNickel Aluminium catalyst (Fig. 13d).Two alumina supported Ni catalysts with pore sizes of 5.4 nm and 9 nm were synthesized and tested in the bi-reforming process [129] for the production of hydrogen rich gases. Structural and functional characterisation of catalysts showed that Ni/Al2O3 with the largest pore size exhibited better characteristics i.e. higher capacity to adsorb CO2, higher surface area, higher proportion of stronger catalytic sites for hydrogen adsorption and lower Ni crystallite sizes. At all the investigated temperatures, for a CH4: CO2: H2O molar ratio of 1:0.48:1.2, a (H2+CO) mixture with H2:CO ratio around 2.5 was obtained. The optimum conditions for the production of hydrogen rich gases, were CH4: CO2: H2O = 1:0.48:6.1 and 600 °C.Dan et al. [130] have investigated the role of Ni/Al2O3, Ni/MgO–Al2O3 and Ni/La2O3–Al2O3 with bimodal pore structure in the bi-reforming process. The authors observed that La2O3 and MgO promoted catalysts presented better functional and structural properties. Among all the catalysts, Ni/La2O3–Al2O3 was found to be the catalyst with best stability and activity. The presence of both lanthanum and magnesium oxides contributed to excellent dispersion and stabilization of Ni nanoparticles on the catalyst surface. The catalytic activity for the bi-reforming process increased in the order Ni/Al2O3(r) < Ni/Al2O3 < Ni/La2O3–Al2O3 ≈ Ni/MgO–Al2O3.Lanthanide group metals (La, Ce) have been reported [131–133], to be efficient promoters for Ni-based catalysts. Recently, literature reports have suggested [120,121], that during the bi-reforming of methane smaller nanoparticles of Ce–ZrO2,ZrO2 and MgO supported by Ni catalysts were found to be highly stable and active. Koo et al. [134], used a stable and extremely active magnesium oxide promoted Nickel/Al2O3 catalyst to investigate catalytic activity and coke formation during bi-reforming for potential applications in gas to liquid (GTL) processes. In their study, the incipient wetness technique was employed to synthesise Ni/Al2O3 catalysts with different concentrations of MgO. The authors used H2-chemisorption, CO2-temperature programmed desorption (TPD), BET analysis, and X-ray diffraction (XRD) to examine the characteristics of the prepared catalysts. Furthermore, the authors established that by changing the feed ratio of H2O/CO2, a H2/CO ratio of 2 was obtained during the bi-reforming reaction. Additionally, catalysts containing 20 wt % magnesium oxide (MgO) showed high coke resistance and excellent catalytic performance during the bi-reforming reaction. MgO addition to the catalyst formed a stable MgAl2O4 spinel phase at high temperatures and was quite effective in eliminating formation of coke by enhancing the adsorption of CO2 because of higher base strength on the surface of the catalyst. SEM images of reduced Ni/MgO/Al2O3 catalysts with changing concentrations of MgO content are illustrated in Fig. 14 .In particular, Cerium oxide has been widely recognized as an efficient promoter for Ni-based catalysts. This is because the redox properties of Ce4+/Ce3+ results in easier gasification of the settled coke on the surface of the catalyst and also helps in storage and delivering of active oxygen thereby enhancing the dispersion of Ni. In another study [132], Ce-promoted Ni/MgAl2O4 catalysts synthesized by co-impregnation showed higher metal dispersion than Ni/MgAl2O4 catalyst alone and demonstrated outstanding reducibility properties at lower temperatures of around 550°C, as established by XPS. The authors found that the catalytic activities of Ni–Ce/MgAl2O4 catalyst were the highest and it generated enormous coke resistance during the bi-reforming reaction performed at lower temperatures with Ce/Ni ratio of 0.25. These were due to stronger metal-support interactions and powerful dynamic oxygen movement through close contact with Ni–Ce. Furthermore, when no Ce was present, the NiO crystallite size in the Ni/MgAl2O4 catalyst was observed to be enormous at a value of 11.0 nm and indicated a lower metal scattering of 3.49%. The authors used Brunauer–Emmett–Teller, (BET) adsorption H2-chemisroption, CO2-TPD and TPR to ascertain the crystallite size of NiO, basicity and reduction temperature of the catalysts. Results revealed that the nickel oxide (NiO) crystallite size, reduction degree and dispersion of the metal were significantly affected by cerium addition to the Ni/MgAl2O4 catalyst.The authors [132] also employed Raman spectroscopy in the range of 1200–1800 cm-1 to investigate coke formation in the presence of Nickel Cerium/MgAl2O4 catalysts with varying Cerium/Nickel ratios. The spectra in Fig. 15 revealed two peaks in the vicinity of 1600 cm−1 and 1350 cm−1 which corresponds to G band and D band. The role of the G band is to provide useful information related to the electronic characteristics of filamentous carbon [135] while the D band arose from imperfect and polycrystalline graphite. Additionally, Ce-promoted Ni/MgAl2O4 showed a decrease in peak intensity with a Cerium/Nickel ratio of 0.25 due to minimal coke formation on the surface of the catalyst in comparison to Ni/MgAl2O4 catalyst without addition of cerium. These results were also in accordance with results obtained from TGA studies: quantification of coke deposition by TGA established a rise in graphitic and amorphous carbon with an increase in Ce/Ni ratio to 1, which was further confirmed by the increase in D band peak intensity (Fig. 15) for amorphous carbon. The advantages of using a Ce-promoted Ni/MgAl2O4 catalyst in the bi-reforming reaction relate to its inherent ability to eliminate formation of amorphous coke in comparison to the Ni/MgAl2O4 catalyst.Recently, there has been renewed interest in developing Ce1-x−ZrxO2 catalytic systems [136]. It has been established that addition of zirconium oxide (ZrO2) to cerium oxide results in significant improvement in the oxygen storage capacity of cerium oxide, its thermal stability, metal dispersion and its redox properties. These improvements were attributed to the preferential replacement of Ce4+ with Zr4+ ion existing in the structure of the lattice surrounding cerium oxide (CeO2). [136–138], The Ce1-x–ZrxO2 catalytic unit has also been regarded as an outstanding material for support in Ni-based catalyst systems [139–141]. CeO2–ZrO2 has been reported to be an effective promoter for the Ni/θ-Al2O3 catalytic system and helps in significant suppression of coke formation with a high catalytic stability [142] Bae et al. [143], investigated the catalytic activity of Ni/MgAl2O4 catalyst in presence of cerium oxide-zirconium oxide (CeO2–ZrO2) during combined steam and CO2 reforming of methane. The synthesis of the catalysts were performed by employing an impregnation technique followed by co-precipitation process of CeO2–ZrO2 components.Furthermore, the basic supports such as MgO or ex-hydrotalcite MgAl2O4 employed in this study possessed beneficial effects such as minimising coke formation due to the reduced acidic site density [144,145]. The Cerium oxide-zirconium oxide (CeO2–ZrO2) component demonstrated a key role in the conversion of CO2 by increasing CO2 activation when contacted with crystallites of nickel. The catalysts synthesized by co-precipitation technique showed higher catalytic characteristics in comparison to catalysts synthesized by successive impregnation of Ni on support of MgAl2O4 with cerium zirconium oxide (CeO2–ZrO2).Addition of lanthanum to Ni/Al2O3 catalysts inhibited the agglomeration of Ni particles due to the enhancement of strong metal to support interaction (SMSI). SMSI of catalysts was reported to enhance thermal stability [145–147]. Park et al. [148], synthesized 10 wt % Nickel–xLanthanum/MgAl2O4 catalysts where x ranges from 0 to 5% by the co-impregnation technique during the bi-reforming of coke oven gas (COG). They conducted aging treatment with a H2: H2O: N2 ratio of 1:10:1.25 with temperature around 900°C run for 50 h. The results revealed an increase in the Ni crystallite size for all the investigated catalysts subjected to ageing. Furthermore, the lanthanum promoted catalysts exhibited greater nickel dispersion than Ni/MgAl2O4 catalyst due to their enhanced interactions between the metal and support. Results from catalytic tests performed at 900°C and at a pressure of 5 atmospheric pressure for 40 h with a CH4: H2O: CO2:H2:CO: N2 ratio of 1:1.2:0.4:2:0.3:0.3 also revealed that aged Ni–2.5La/MgAl2O4 catalyst showed maximum sinter stability and activity due to its enhanced nickel dispersion and surface area.Furthermore, the role of Ce/Zr ratio on the catalytic activity of Ni–Cex Zr1−xO2 catalyst and coke formation was demonstrated by Roh et al. [131], during the bi-reforming reaction. The authors used co-precipitation method to synthesise Ni–Ce–ZrO2 catalysts having different ratios of CeO2/ZrO2 for syngas production having potential applications in gas to liquid (GTL) processes. 15% Ni–Ce0.8Zr0.2O2 demonstrated excellent stability and highest activity during BRM which was attributed to the dispersion of nickel oxide having higher oxygen storage capacity and intimate contact with the support. Recently, it has been reported [131] that Cerium content along with nickel-cerium loading technique has an significant effect on the transfer of O2 occurring between nickel (Ni) and cerium (Ce). Studies have also shown an improvement in coke resistance of 12 wt % Nickel/α-Al2O3 catalyst in the bi-reforming reaction. [133,149], However, the characteristics of supports play a marked effect on coke formation [132]. For example, α-Al2O3 support caused carbon deposition due to its acidity [150]; active nickel metal and Al2O3 supports underwent interaction to form inactive NiAl2O4, resulting in deactivation of the catalyst [134]. Additionally, Baek et al. [151], observed higher coke resistance and enhanced catalytic stability of Ni–Ce/MgAl2O4 (MgO/Al2O3 = 3/7) in comparison to Ni–Ce/θ-Al2O3 in bi-reforming process.Gao et al. [152] develop an ideal Ni–Ce/ZSM-5 catalyst by the impregnation method for the bi-reforming process. The authors noted that by adjusting the parameters properly, highest conversion i.e. 99% and 94% of CH4 and CO2 to syngas was achieved in presence of Ni–Ce/ZSM-5 catalyst. Furthermore, the catalyst did not show any deactivation and maintained high activity for 40 h. SEM, XRD and H2-TPR analysis further established the structure as well as composition of the catalysts and provided better understanding of the catalytic performance.Chen et al. [153] synthesized highly dispersed mNi/xL/Si catalysts by one-pot sol-gel process and applied to the bi-reforming process for syngas production. Results revealed that the addition of lanthanum improved the stability, catalytic activity as well as the coke resistance of these catalysts. The17.5Ni/3.0LaeSi catalyst prepared using ethylene glycol and poly (ethylene glycol) displayed the best catalytic activity, coke resistance and stability. Additionally, the H2/CO ratios in the product gas were tuned by varying the C/S ratios in the feed.Jabbour et al. [154], employed a one-pot method followed by evaporation-induced self-assembly (EISA) to synthesise two types of catalyst namely Ni5%M5% where M represents Ca or Mg and Nix% where X corresponds to 5–10 wt % along with packing of mesoporous Al2O3.Low cost and widely available Mg2+ and Ca2+ containing salts were used as the additives based on their potential to yield basic properties (in their oxide form) and their positive impact on bi-reforming process [155]. Temperature programmed reduction (TPR) of calcined Ni-loaded samples displayed a strong reduction peak at higher temperatures ranging from 550 to 800°C (Fig. 16 b–f), which was ascribed to the reduction of oxidised nickel (Ni) undergoing stronger interaction with the support present in the mixed spinel phase [156]. The author observed that there was no peak signifying weakly-bounded Ni species below a reduction temperature of around 500°C, which was similar to the finding for non-porous impregnated alumina samples [82,157]. The authors in their studies noted that after the reduction process, the catalysts demonstrated higher dispersion of Ni within the arranged oxide cavity, possessed elevated activities and also showed long-lasting stability in bi-reforming of CH4 performed at 800°C. They also observed that a relationship existed between carbon deposition and reactivity level in the presence of Mg free catalysts.SEM (Fig. 17 A, B) and TEM (Fig. 17C, D) images for spent Ni10%Al2O3 clearly identified long carbon filaments on the exterior area of the alumina containing grains, with some grains which were found to more protected than others. These images also showed some Ni nanoparticles containing coke were situated either at the boundary between the support and the filament embedded into it. These images also resembled carbon nanotubes nucleation with a ‘closed end’ consisting a nanoparticle at either their closure or located inside the tip [158–160]. For both Ni5% Mg5% Al2O3 and Ni5% Ca5% Al2O3, a peak was observed at very high temperatures of above 800°C, as shown in Fig. 18 A (e and f); this was a possible indication of free metallic Ni existing under stronger interactions or may have been related to the reduction of Mg- or Ca-derived species. Furthermore, a linear correlation was established between H2 uptake and Ni content, which confirmed that all Ni used in the synthesis was completely reclaimed in the solid after preparation. The authors also reported that, due to the endothermic nature of the bi-reforming reaction, the conversion of both CH4 and CO2 decreased at lower temperatures, and CO2 conversion was more significant below 700°C.Results also revealed that there was a beneficial effect when 5 wt% magnesium or calcium was used for the conversion of both CH4 and CO2 represented by d and e (Fig. 18A, B), compared to catalyst containing 5 wt % Nickel and without additive. The enhancement in the reactivity in presence of 5 wt (wt%) magnesium or calcium was higher than with Ni7.5% Al2O3 despite the lower Ni content reported in previous literature [161–163]. Jabbour et al. [154], established that in addition to high activity levels, doping of the Ni catalysts with Mg or Ca additive resulted in excellent catalytic stability for high temperature bi-reforming operations. The authors found that mesoporous catalysts synthesized by one-pot method served as an ideal candidate for catalysing met gas production from biomass-related natural resources. The beneficial effect of nickel confinement in the pores was twofold, one in protecting the metal nanoparticles against sintering phenomenon and second against coking due to steric constraints.Kang et al. [164], synthesized core shell structured Ni catalysts Ni/MgO–Al2O3 and Ni/Al2O3 and via technique coined multi-bubble sono-luminescence and conducted tests using these catalysts for the bi-reforming process. The authors observed that Ni catalysts constituting of 10% Ni loaded on Aluminium oxide or MgO–Al2O3 exhibited exemplary performance during the steam reforming of methane, achieving 97% conversion of CH4 at a temperature of 750°C. Additionally, methane conversion was 96% at 850°C during dry reforming of CH4 and demonstrated greater thermal stability for the initial duration of 50–150 h. The results also established that supported Ni catalysts demonstrate excellent performance in both mixed and auto-thermal reforming of CH4, where satisfactory thermal stability was noted for the first 50 h. An interesting observation was that no significant carbon formation was obtained on surface of the investigated catalysts after the reforming reaction. Very recently, Koo et al. [118], synthesized nickel catalysts in the nanoscale by employing a mixture of magnesium oxide−aluminium oxides (MgO–Al2O3) obtained from a structure resembling hydrotalcite. Their results revealed an enhancement in the coke resistance with various mixed ratios of Mg/Al for the generation of syngas during bi-reforming for applications in GTL processes.Mesoporous SBA-15 has aroused enormous interest among researchers in steam reforming [165,166], and CH4 dry reforming [68,167], process due to its high surface area, high silanol group density, uniformity of pores and enhancement in active metal dispersion with smaller crystallite size [168]. A group of Nickel/SBA-15 catalysts with Ni content ranging from 5 to 15 wt % were synthesized by Huang et al. [169], along with 10% Ni/MgO/SBA-15 catalysts with MgO content ranging from 1 to 7 wt (wt %) during combined steam and dry reforming reaction in a continuous micro-reactor. XRD, H2-TPR and CO2-TPD techniques were used to investigate the structure of catalysts. The authors observed that selectivity of carbon monoxide (CO) for these reactions was almost 100% and they also noticed that with the change in the molar ratio of H2O/CO2, there can be effective control of the H2/CO ratio. After reaction at 850̊ C for more than 120 h with 10 wt % of Ni/SBA-15 catalyst, the conversion of methane underwent a decrease from 98% to 85% while the conversion of CO2 reduced from 86% to 53%, respectively. Additionally, the catalyst containing 3% MgO/SBA-15 loaded with Ni demonstrated excellent catalytic activity after a reaction for 620 h and the CO2 conversion over this catalyst underwent a decrease from 92% to 77%, while no change in CH4 conversion was observed. Furthermore, certain changes in the MgO promoter enhanced the Ni0 species dispersion and resulted in an increase in the adsorption affinity of CO2, thereby inhibiting coke deposition and retarding the deactivation phenomenon.Mg–Al mixed oxides derived from hydrotalcite-like materials are reported [118,154], to exhibit higher activity and stability in bi-reforming process due to its basic property, enhanced steam and CO2 adsorption, strong Ni to support interaction and fine dispersion. These catalysts were synthesized using various preparation methods such as impregnation of pre-calcined carriers or simultaneous co-precipitation of the mostly nitrate-based solution of all the constituents. Roohollahi et al. [170], synthesized numerous Ni-based catalysts supported on mesoporous MgO–Al2O3 resembling a Mg–Al hydrotalcite structure with Mg/Al ratio of 1. Mg–Al hydrotalcite-like components, represented by the formula [MgII 1-xAlIII x(OH)2]x+(CO3 2−)x/2⋅mH2O, have been regarded as the best candidates for precursors employed in the synthesis of mesoporous MgO–Al2O3 carriers possessing high surface area [171]. The synthesis of hydrotalcite-like components was performed at an optimized pH of 10, which were then calcined at various calcination temperatures from 500 to 800°C to obtain a homogenous texture. Results from the bi-reforming reaction conducted on the catalysts at 800°C for 36 h with feed stock constitution of CH4: CO2: H2O = 1.0:0.4:0.8 at GHSV = 150,000 mL·gcat−1 h−1 revealed that the sample derived from the carrier calcined at 700°C exhibited the lowest nickel crystallite size (2.68 nm) and largest nickel surface area (25.01 m2/g). The results also established excellent conversion efficiencies for CH4 (93.7%) and CO2 (75.2%) and higher resistance to coke formation. The high resistance to carbon formation was due to the enhanced strength of basic sites formed in the catalyst carrier during the calcination of Mg–Al hydrotalcite-like components at 700°C.He et al. [172] investigated the role of nickel nanoparticles supported on the binary Mg–Al metal oxide catalysts during bi-reforming of CH4. The successful synthesis of Ni/MgO, Ni/Mg x Al y O, and Ni/Al2O3 catalysts were also supported by XRD, TEM, and FT-IR results. The TPR profile revealed that the reduction temperature of Ni species underwent a slight decrease upon addition of Al due to the formation of the NiAl2O4 phase. Furthermore, the XPS spectra demonstrated that Ni/MgO and Ni/Al2O3 produced higher amounts of Ni0 after H2 reduction.NiO–CaO catalyst has been reported [173] to exhibit high selectivity, activity and productivity in the oxidative conversion of methane to synthesis gas. Choudhary et al. [174], have reported the role of NiO–CaO catalysts during SRM, DRM and combined steam and CO2 reforming of methane to produce CO and H2 at varying temperatures ranging from 700 to 850°C and gas hour space velocities (5000 to 70,000 cm3 g−1·h−1) They characterised the catalysts using various techniques including XRD, XPS and TPR. Their results revealed that the catalysts demonstrated high activity/selectivity during all of the reforming processes tested. When CO2 reforming was performed in conjunction with steam reforming process a drastic reduction in carbon deposition from 25.96% to 1.08% was observed for a feed composition of CH4:CO2:H2O = 1.0:0.55:0.55 [174]. Furthermore, the authors noted that when the feed composition was maintained for CH4: H2O = 1:1 during the steam reforming reaction, the reaction characteristics were outside the coke formation control. Nevertheless, for the dry reforming reaction with a reactant feed composition of CH4: CO2 in the ratio of 1:1, the coke formation was obtained from a gas mixture formed at equilibrium.The authors [174] also noticed that complete conversion of methane to syngas with 100% selectivity consisting of both CO2 and H2 and during bi-reforming reaction at 800°C with GHSV ranging from 20,000 to 30,000 cm3 g−1 h−1. The authors observed that by changing the carbon dioxide/steam (CO2/H2O) ratio in the reactant feed, a significant improvement in the bi-reforming process occurred and also a desirable H2/CO ranging from 1.5 to 2.5 was seen. TPR studies were performed to measure changes in the concentration of H2 owing to reduction of nickel oxide in the catalyst. The TPR curves (Fig. 19 ) revealed maximum value in the range 400 and 450°C, in accordance with the maximum peak temperature observed around 418°C attributed to reduction of bulk nickel oxide.Chen et al. [153] synthesized mNi/xLa/Si catalysts with efficient dispersion characteristics and comprising of various weight contents of nickel and lanthanum by using sol-gel method, and tested these catalysts for bi-reforming of CH4 to generate syngas. The authors noticed an increase in the stability, catalytic characteristics and an enhancement in the resistance of carbon deposited during bi-reforming in presence of mNi/xLa/Si catalysts upon addition of lanthanum. The 17.5Ni/3.0La/Si catalyst prepared using ethylene glycol and poly (ethylene glycol) demonstrated excellent coke resistance and catalytic activity. Additionally, modification of the carbon/sulphur (C/S ratios) in the reactant caused tuning of the H2/CO ratios in the gas generated as products. Furthermore, when the bi-reforming reaction was performed in presence of 17.5Ni/3.0La/Si catalyst produced a H2/CO ratio of about 2 for the C/S ratio of 0.5.Ni-phyllosilicate (PS) intermediates were used to synthesise [175] Nickel–SiO2–MgO materials for its application in bi-reforming of methane., and the role of reaction temperature as well as steam on the reforming process were also investigated. The results revealed that catalytic performance was excellent and resulted in 80% conversion of CH4 and 60% CO2 conversion respectively, at 750°C for 140 h in presence of a Ni–30 wt % SiO2–55 wt % MgO catalyst. Furthermore, carbon deposition was found to be stable when the H2/CO ratio was maintained at 2. The catalytic behaviour of the investigated catalyst was ascribed to its structural stability, acidic strength and enhanced basicity for the reforming reaction conducted at high temperatures. The presence of nickel-magnesium comprising phyllosilicates in the reduced catalysts were established by TEM and XRD technique. Furthermore, a TPR profile of around 750°C substantiated the presence of strong interlinkage between nickel and Silicon dioxide–Magnesium support species. A representative schematic diagram of this is illustrated in Fig. 20 .Jabbour et al. [82], used an one pot method (Fig. 21 ) for synthesis of mesoporous nickel–alumina catalyst containing 5 wt % Nickel and possessing an ordered structure. From their observations, the ordered Ni–alumina sample exhibited excellent stability in comparison to non-porous and impregnated catalyst during the bi-reforming process at 800°C over 40 h. The conversion percentage of methane was consistent with the thermodynamically expected variants. The authors also noted that nickel catalyst loaded with SBA-15 demonstrated enhanced catalytic activity than Ni/celites, however both these catalysts underwent rapid deactivation on stream which was attributed to the partial re-oxidation of the Ni active phase under the investigated conditions (see Fig. 22).SBA-15 support has been employed for suppressing carbon formation in steam reforming reactions [166,167], and has aroused significant interest due to its high surface area, high silanol group density, pore uniformity. An incipient wetness method was employed by Singh et al. [176], to synthesise SBA-15−packed Ni catalyst by impregnating nickel nitrate onto the SBA-15 support. They found that the surface area decreased from 669.5 m2 g−1 to 538.6 m2 g−1 with the change in catalyst support from SBA-15 to 10 wt % Nickel/SBA-15 catalyst was confirmed by BET surface area analysis. Analysis by H2-TPR demonstrated the complete reduction of NiO nanoparticles beyond 576.85°C where the temperature of reduction from nickel oxide to metallic nickel was completely dependent on metal-support interactions which was correlated to the location, confinement effect and crystallite size of nickel oxide. CO2 and H2O had a significant role in controlling formation of carbon during bi-reforming of methane due to their unique capability in converting the partially dehydrogenated CxH1-x to a mixture of CO and H2. The authors observed that carbon dioxide conversion and methane conversion was 58.9%, and 61.6% respectively. Furthermore, the resulting H2/CO ratio was found to be 2.14 during the combined CO2 and steam reforming of CH4 under stoichiometric conditions. A steep increase in the H2 and CO yield was noticed while increasing the CO2/(CH4 + H2O) ratio, and a considerable decrease in the ratio of both hydrogen and carbon monoxide ratio ranging from 2.14 to 1.83 was observed with a decrease in the H2O/(CH4 + CO2) ratio. Furthermore, Ni/SBA-15 exhibited higher resistance towards both coking and sintering which was related to the efficient distribution of nickel particles and steric effects caused by SBA-15.The synthesis, catalytic activity and characterisation studies on Ni/SBA-15 catalysts during BRM has been reported [177]. The authors observed that 25 wt % Nickel/SBA-15 SGM catalyst showed the maximum conversion of CH4 (23%, 548̊ C), which was followed Ni/SBA-15 HTM (CH4 ≈ 20% at 548°C) and 10% CH4 conversion was achieved in presence of 25 wt % Ni/SBA-15 CG catalyst. CO2 and CH4 conversion were found to be 82% and 23% at 548°C, respectively. These differences in the catalytic activity were related to the degree of availability of active metal for the reaction. Due to excellent catalyst activity of these catalysts, these catalysts were employed for the formation of membrane reactor with hollow fibres and catalytic hollow fibres.The authors employed commercially available SBA-15 for comparison. SEM micrographs (Fig. 23 ) revealed a needle shaped particle having a grain size of around 0.6 μm (A1 and A2). The SBA-15 particles synthesized by the sol-gel method did not display a homogeneous shape and consisted of a hard shell covering smaller particles whose grain size was approximately around 0.1 μm.Mesoporous siliceous SBA-15 material has been used as support for preparing active metal catalysts in several reforming processes [66,178]. The mesoporous SBA-15 support possessed uniform mesopores with thick framework walls, high thermal stability and wide specific surface area [179], [. Additionally, the ordered hexagonal mesostructure of SBA-15 support provided a confinement effect to anchor the nanoparticles inside its channels and also prevented deposition of carbonaceous species metal sintering [115]. Siang et al. [180], used the incipient wetness impregnation technique to synthesise stable and active boron (B) aided catalyst for bi-reforming of methane. Results revealed that B2O3 and nickel oxide particles were scattered on the outer area of SBA-15 support possessing higher surface area. Additionally, the authors observed an enhancement in catalytic activity that underwent a linear increase with temperature due to the endothermic behaviour of the catalysed process. They obtained H2/CO molar ratio of 2.7 and 67.3% of CH4 conversion at 799.8°C which was highly significant for downstream Fischer-Tropsch (FT) applications. Furthermore, XPS measurements revealed that B facilitated the adsorption of CO2 through the electron transfer to the Ni cluster at the neighbourhood, thereby improving its catalytic activity. More importantly, analysis by XRD and Raman showed that boron doped catalyst was completely free from graphitic and amorphous carbon deposition. This was due to the incorporation of B into the octahedral sites occupied by NiO, resulting in inhibition of carbonaceous deposits.Encapsulation of Ni particles in a suitable support material has been reported [181] to enhance the sintering resistance and coke resistance of Ni catalysts. The introduction of promoters namely rare-earth metals; metal oxides, alkaline earth and alkali metals is also one of the effective strategy to prevent the sintering of active sites/supports and enhance the coke-resistant ability of catalysts [182]. Chen et al. [153], synthesized highly dispersed mNi/xLa-Si catalysts by employing one pot sol-gel process by varying the weight percentages of nickel and lanthanum. These catalysts were subsequently applied to generate syngas during bi-reforming of CH4. The authors observed that La addition enhanced the stability, coke resistance and catalytic activity of mNi/x Lanthanum–Silicon catalysts. The 17.5Ni/3.0LaSi catalyst prepared by employing poly (ethylene glycol) and ethylene glycol displayed the coke resistance, maximum selectivity and catalytic activity. One notable observation was that a H2/CO ratio of about 2 was obtained when the carbon to sulphur ratio was maintained at 0.5, for the 17.5Ni/3.0LaSi catalyst, suitable for potential applications in Fischer-Tropsch synthesis.Literature reports [183] have established that CeO2–Al2O3 combinations are potential supports for reforming reactions. Furthermore, the redox properties of CeO2 resulted in a significant improvement in the oxidation of deposits thereby enhancing the lifetime of the catalysts [184,185]. Furthermore, second metal addition promoted the formation of an active phase along with modification of the support. Bimetallic systems [186] have been known to display superior catalytic activity and increased the resistance of carbon formation in comparison to their own counterparts. The bimetallic combination of Ni–Sn has proved to be of considerable interest in reforming reactions. Additionally, the dispersion of nickel over the catalyst surface has been shown to be enhanced in the presence of Sn [187].Straud et al. [188], synthesized a set of multicomponent advanced catalysts composed of Sn, CeO2 and Ni/Al2O3. A schematic diagram representing the production of syngas in the presence of the investigated catalysts are shown in Fig. 23. The authors observed that addition of minute amounts of the investigated dopants improved the performance of methane reforming using CO2. From their results it was noticed that a multicomponent Sn 0.02 Nickel/Cerium-Al catalyst showed excellent catalytic characteristic and remained active over a long period of 92 h. The catalyst also demonstrated an exceptional level of stability and conversion during BRM. Comparison of dry reforming and BRM reactions over the Sn0.02Ni/Ce–Al catalyst at 700°C revealed that H2/CO ratio remained above 1.6 for 24 h. This suggested that the catalyst could generate high quality synthesis gas by introduction of water into the reforming mixture.Therefore, the addition of water established the suitability of the Sn0.02Ni/Ce–Al catalyst for bi-reforming of CH4. Furthermore, the results also revealed that presence of ceria created high storage capacities for oxygen and changed both the acidic and basic characteristics of support thereby enhancing the catalyst performance. The multicomponent catalyst Sn0.02Ni/Ce–Al proved to be active over period of 92 h and fared well over a range of space velocities and temperatures.The remarkable level of stability and excellent conversions seen in the bi-reforming process has proved the versatility of Sn0.02Ni/Ce–Al catalyst which can be upgraded to variety of CO2 containing feed stocks.Literature reports [189,190]have established the role of ZrO2 as an excellent support for reforming reactions because of its higher oxygen mobilisation, excellent thermal stability as well as its unique basic and acidic properties. The reinforced interaction between nickel and zirconium oxide makes zirconium oxide an effective support for a nickel based catalyst.Agli et al. [191], reported that basic mineralisers affect the nucleation, rearrangement and crystallisation of gel made of zirconia during the synthesis of zirconium oxide. Hence, Zhao et al. [190], used the hydrothermal method [192,193] with various mineralisers along with l-arginine ligand-using wetness impregnation technique [194] to synthesise Ni/ZrO2 supports for bi-reforming of methane. Results from their studies revealed that the catalysts performance depended on texture of the zirconium oxide support and its morphology was also highly affected by the mineraliser amount. In this study, the authors synthesized Zirconium oxide support with a mole ratio sodium acetate/Zr4+ as 0.5 denoted by (SAc0.5). ZrO2−supported Ni catalyst was synthesized by employing sodium acetate where the mole ratio was NSAc/Zr = 0 and also showed increased catalytic activity in comparison to catalyst i.e zirconium oxide synthesized using (SC) where the ratio was Nsc/Zr = 0.5. The authors established from the results that sodium acetate would serve as a suitable mineraliser for making an excellent ZrO2 support and also in terms of its stability and activity. Furthermore, the authors [190] observed that, in general, the addition of different amounts of mineralisers to ZrO2 supports had a significant effect on textural properties, which in turn affected the behaviour of the Ni-supported catalysts on zirconium oxide and also influenced the catalytic activity of the Nickel/Zirconium oxide catalysts during bi-reforming of methane.The TEM micrographs in Fig. 24 show that all the investigated zirconium oxide supports resembled cobblestone like structure, with dimension of mesopores. Interestingly, reduction in pore volume and pore diameter along with expansion in the surface area was observed when the SAc/Zr molar ratio rose from 0.5 to 2.0. This provided the Ni/ZrO2 catalyst with a bigger crystallite size but also caused lower dispersion compared to Ni/ZrO2 (SAc0.5). From these studies, the authors noted lowering in the sintering resistance of nickel in Ni/ZrO2 (SAc2.0) catalyst than Ni/ZrO2 (SAc0.5), which was attributed to its imperfect interaction between nickel and zirconium oxide as established by H2-Temperature programmed reduction. Fig. 25 displays the catalytic characteristics of the prepared Ni/ZrO2 catalysts with varying amount of mineralizers present in the support.The figure above revealed the initial activity in the order of Ni/ZrO2 (SAc0.5) ≈ Ni/ZrO2 (Non) > Ni/ZrO2 (SC0.5) > Ni/ZrO2 (SAc2.0) respectively. Nevertheless, the Ni/ZrO2 catalyst without any acetate in the figure showed least stability among all the catalysts.Itkulova et al. [195], used Group VIII metals(0.25–1 wt %) along with alumina as a support to synthesise 5% bimetallic Co-based catalysts. The bimetallic Co constituted catalysts were synthesized by impregnation of Al2O3 with solutions comprising of both cobalt and platinum compounds followed by a thermal treatment. The authors investigated the stability of these catalysts by varying the temperature (300–800°C), composition of feed mixture and space velocity (SV) (500−3000 h−1) during bi-reforming as well as DRM. The authors observed that methane conversion was almost 100% at 750°C and 770°C for 5 wt % Co–Pt (9:1)/Al2O3 catalyst during both DRM and BRM, However, the results in Fig. 26 a reveal a decrease in CO2 conversion during the reforming process (Fig. 26 b) performed over the entire temperature range compared with DRM due to the suppression of CO2–CH4 reaction by the competing CH4–H2O interaction. The authors also observed a surge in the H2/CO ratio from 0.84 to 1.0 when 20 vol % steam was added to the feed with equal amounts of CH4 and CO2.The effect of Pt on BRM for various feed compositions was also investigated [195]. Results in Fig. 27 revealed an enhancement in the catalytic activity with increased platinum loading varying from 0.25 to 1 wt %. It was also noted that higher temperatures were necessary for the total conversion of CH4 when there was a decrease in platinum content in the catalyst. It was established that addition of 10–30% steam had a marked effect on the conversion of CH4, which further decreased the temperature required for conversion of methane and an increase in the ratio of H2/CO.Syngas produced during bi-reforming of methane over 5% Co–Pt/Al2O3 catalyst showed a desirable H2/CO ratio >1. Pt was responsible for the formation and stabilization of highly dispersed and reduced bimetallic nanoparticles. Itkulova et al. [196] have investigated the role of 5% Co–Pt catalysts modified with 0.25–0.5 mass% Pt supported on alumina and modified with zirconia (ZrO2) with amounts ranging from 5 to 10 mass% of Zr in the bi reforming process in the temperature range of 300–755 °C, and CO2/CH4 in 1:1 ratio. The results revealed that introduction of 20 vol% of steam into the CO2–CH4 feed was highly beneficial to the performance of the bi-reforming process. The improved performance of the 5%Co–Pt/Al2O3–ZrO2 catalysts was attributed to the synergistic effect caused by the combination of two reactions i.e. dry and steam reforming of methane.The major issue affecting commercialisation of the reforming process is coke formation, which causes deactivation of catalysts. The most effective way for decreasing coke formation is by coupling CO2 with steam. It has been established that the support plays a significant role in suppressing formation of coke on Group VIII metals during the CO2 reforming of CH4 [197–200]. Several researchers [201,202], have demonstrated that the addition of promoter such as cerium led to a marked improvement in the activity of catalyst, stability and also decreased the sintering of ZrO2 during calcination performed at high temperature. Literature reports [203–207], have also established that Pt–ZrO2 catalysts demonstrate high stability and activity under extreme deactivating environment.The activity for CO2 reforming of methane has been investigated by Noronha et al. [208], on Pt–ZrO2 (Fig. 28 a) and Pt–Ce–ZrO2 (Fig. 28b) catalysts under CH4:CO2 molar proportion of 2:1. The authors noticed that the conversions of CH4 and CO2 decreased slightly in the presence of Pt–ZrO2 catalysts (see Fig. 29).Interestingly, after the removal of water the conversion of carbon dioxide and methane remained roughly constant, and at the same level as after interaction with water (Fig. 28a). A more drastic reduction in H2/CO from 0.82 to 0.45 was also noticed after 22 h. However, the authors observed that DRM of CH4 in presence of H2O occurred differently with Pt–Ce–ZrO2 Pt–ZrO2 catalysts. Furthermore, CH4 and CO2 conversion underwent a decrease upon addition of water during the ongoing reaction on the Pt–ZrO2 (Fig. 28 a) catalysts, with the decrease in conversion of CO2 being significant. This was attributed to the reaction between excess H2 and higher amount of CO2 through water-gas shift reaction [209] in the reverse mode. The reduced stability observed in the Pt–ZrO2 catalyst was related to the diminishing of oxygen vacancies on the support and oxidation behaviour. Furthermore, Temperature Programed Oxidation analysis established that water addition enhanced the amount of mass of carbon deposited on the surface. Nevertheless, Pt–Ce–ZrO2 exhibited excellent stability in presence of H2O and its stability was due to higher number of vacancies caused by O2 on the support. Both the Pt catalysts with and without Ce were relatively stable during CO2 reforming of methane performed at 105 psig.BRM is considered to be an endothermic process which involves optimization of the temperature within the bed containing the catalyst and also a large amount of heat transfer into the reaction system occurs with the aid of external source. This suggests that catalysts used in these processes should have greater thermal conductivity, which can be attained by employing metallic supports [210]. Several authors have reported the role of numerous catalysts that operate on these supports [211,212].A promising under layer for Ni catalysts is MgO due to its high thermal stability, ability to decrease carbonisation and ability to easily form solid solutions with NiO, also aids in promoting the dispersion of reduced nickel crystallites [213]. There have been numerous studies on supported Ni catalyst with a MgO under layer: supports on metal foams [213], porous Ni plate [212] and Al2O3–SiO2 [214].Danilova et al. [215], reported the synthesis of thick porous Ni ribbon (pNirb) with a MgO under layer supported by Ni catalyst on the top. The under layer constituting magnesium oxide was synthesized by packing of the support with MgNO3 solution, then subjected to drying followed by calcination performed at 550̊ C in presence of air designated as Support 1 and the calcination performed in flowing H2 was known as Support 2. The catalyst was reduced under the atmosphere of flowing H2 at 750°C was termed catalyst I and the catalyst reduced at 900°C was termed catalyst II. Use of these supported catalysts resulted in 49% and 56% conversion of CH4 with support 1 and support 2, respectively. The greater activity of support 2 in comparison to support 1 was attributed to efficient dispersion of Ni crystallites that was produced from solid solution reduced in presence of reaction medium. The authors further remarked that catalyst II (2.7% Ni/(p Nirb +8.6% MgO) and 4.0% Ni/(pNirb +10.4% MgO) exhibited excellent stability for CH4 conversion over a period of 18 h compared with catalyst I (4.6% Ni/(pNirb +6.0% MgO) and 4.6% Ni/(pNirb + 6.0% MgO) under the following conditions: GHSV = 62.5 L h−1,CH4/CO2/H2/N2 = 35/23/39/3).Recently, nanocatalysts have attracted much attention [38]. Nanocatalysts show better selectivity, outstanding stability and higher activity, due to their special crystal structure, higher amount of surface atoms in comparison to their micro-sized counterparts and larger specific surface area [216]. Many works [217] have shown that catalyst preparation with larger surface area affected significantly the physical and chemical properties, which can only be achieved by a nanocatalyst.Khani et al. [218], synthesized novel M/ZnLaAlO4 nanocatalysts where M consists of 3%Ru, 10% Ni and 3% Pt using wet impregnation technique and characterised these by using TPR, FT-IR, TEM, XRD, FE-SEM, Thermogravimetric analysis and Differential Thermal Analysis. The authors evaluated the catalytic characteristics of the these catalysts in the SRM, DRM and BRM of methane at temperatures varying from 600 to 800°C at different gas hourly space velocities values of 10,500, 7000, 3500 h−1. TGA revealed that the nanocatalysts namely 3% Pt/ZnLaAlO4 and 3% Ru/ZnLaAlO4 did not exhibit any coke formation during SRM, which was also supported by FE-SEM (Fig. 30 ).The authors noted that an increase in temperature during the bi-reforming (BRM) of methane increased the CH4 conversion, however decreased conversion of CO2 (Fig. 31 ). Furthermore, Fig. 30 shows that 3% Ruthenium/ZincLaAlO4 demonstrated the lowest activity while 10%Ni/γ-Al2O3 showed marked activity for CO2 conversion. Additionally, among the four tested catalysts (Figs. 30), 3% Ru/ZnLaAlO4 displayed the highest catalytic CH4 conversion. The authors observed a reduction in H2/CO ratio with the rise in temperature from 600 to 700°C for the investigated catalysts used in the bi-reforming process. The authors observed 3% Ru/ZnLaAlO4 a H2/CO of 2.1 at a temperature of 800̊ C, while for 10% Ni/γ-Al2O3 catalyst a lowest value of 1.6 was obtained., 3% Ru/ZnLaAlO4 was considered as the potential catalyst for potential applications based on its resistance to formation of carbon and catalytic efficiency in BRM, dry reforming and SRM. TPR profiles of the nano-catalysts showed lowest reduction temperatures at the onset for 3%Pt/ZnLaAlO4, 10% Ni/ZnLaAlO4, 3%Ru/ZnLaAlO4 at 264°C, 333°C and 230°C respectively.Potdar et al. [219], noted that the nanocatalyst Ni–Ce–ZrO2 synthesized by employing co-precipitation technique exhibited excellent coke resistance and highly stable catalytic activity attributed to the greater mobility of oxygen in the carbon dioxide reforming of methane and higher surface area. Roh et al. [121], demonstrated that higher stability and activity of Ni/MgO–Al2O3 catalyst with nano dimensions was due to the beneficial effects of magnesium oxide (MgO) namely stronger interaction between nickel and support, basicity, enhanced steam adsorption and also the crystallite size of nanosized NiO. Sadykov et al. [220], investigated the role of catalysts made of nanocomposites in the bi-reforming reaction. Nanocomposite catalysts consisting of nickel particles implanted into an oxide matrix of Yttrium or Scandinavium-stabilised Zr (YSZ, ScSZ) mixed with doped Ce–Zr oxides or Lanthanum–Praseodymium–Manganese–Chromium–Oxygen (La-Pr-Mn-Cr-O) perovskite along with promoters namely Pd, Ru and Pt were synthesized via different routes [101].Soria et al. [221], investigated the role of H2O along with Ru/ZrO2–La2O3 catalyst placed in a fixed-bed Palladium reactor with membrane during bi-reforming of methane. The authors observed that addition of H2O along with CO2 during the reforming reaction significantly affected the catalyst activity. Fig. 32 shows that the presence of small concentrations of H2O (1–2 vol %) did not affect the conversion of CH4 appreciably, but an increase in steam to 5 vol % did result in increased CH4 conversion.Furthermore, the CO2 conversion gradually decreased with increasing concentration of H2O from 1 to 5 vol %, and the CO2 conversion exhibited lower values below 330, 375 and 450̊ C for water content ranging from 1 to 5 vol %. Furthermore, at a designated temperature, the composition of syngas (H2/CO ratio) was altered with the change in the concentration of H2O feed.The authors [221] also investigated the stability of the Ru/ZrO2–La2O3 catalyst during both bi-reforming and carbon dioxide reforming of methane at 500°C. It was observed that without steam presence in the reaction feed, the Ru/ZrO2–La2O3 catalyst was very stable and 15% of deactivation was noticed. Fig. 33 shows that the addition of water had a marked effect on the stability, which increased in a significant manner with the increase in steam amount. The deactivation values were 5%, 11%, 8% for addition of 5, 1, 2 vol % H2O, respectively.Research has indicated that steam addition to the CO2 during bi-reforming of CH4 affects the reaction parameters in a temperature-dependent manner, which is noteworthy for the generation of high purity H2 using Pd-based membrane technology.Chaudhary and Mandal [101] demonstrated the CH4 conversion of synthesis gas in presence of NdCOO3 perovskite-type oxides used as a catalyst during BRM of methane. Results from their studies revealed that H2O and CH4 conversion along with the H2/CO ratio, were greatly affected by the feed ratio of CO2/H2O during the reforming process. Furthermore, the heat of reaction was strongly affected by relative concentration of oxygen in the reactant feed, space velocity and temperature. NdCOO3 perovskite-type catalyst proved to be highly efficient for carbon-free bi-reforming process.The Sr-doped Ni–La2O3 catalyst has been reported [222] to generate the highest CH4 and CO2 activity along with the highest resistance to carbon deposition over the catalyst surface which was attributed to considerable involvement of a large amount of mobile lattice oxygen species as a result of C–H activation in dry reforming. Yang et al. [223], have reported the role of Sr addition to LaNiO3 perovskite catalysts during the bireforming process. Mineralogical characterisation by XRD revealed a distortion in the perovskite lattice and generation of alien phases such as La2−xSrx NiO3 ± δ and Sr0.5La1.5NiO4. The authors [223] observed that the reduction behaviour was affected by the presence of these phases. The results also revealed that strontium oxide adsorbed CO2 during the bi-reforming reaction and formed strontium carbonate (SrCO3), which possessed the unique ability of inhibiting carbon sources by producing La2O2CO3. The addition of Sr particles covered the support sites thereby resulting in large-sized Ni particles by decreasing the interlinkage between the support and active metals. The authors [223] recommended using a small amount of Sr in the perovskite-based catalyst for obtaining greater resistance to carbon deposition. Furthermore, larger Ni particles were formed, with diameters of 30.8, 29.9, 27.6 nm for concentrations of 10%, 50%, 30% SrO containing La2O3– NiO3 catalysts in comparison to a La2O3–NiO3 particle size of 13.5 nm.Kim et al. [224], investigated the bi-reforming reaction of methane employing mixed oxides of La, Sr and Ni packed on β-SiC catalysts loaded with Al2O3 for assessing the conversion of carbon dioxide at a certain concentration of Al2O3 used as a modifier. The authors found that though all the investigated tested catalysts provided close activation energies values, the increase in dispersion of aluminium oxide on silicon carbide with 10 wt % Al2O3 as modifier was in agreement with the higher distribution of perovskite containing La2NiO4 crystallites. Additionally, larger amounts of adsorption of carbon dioxide on the efficiently distributed basic Sr and La oxides were also responsible for enhanced carbon dioxide conversion. CO2 and CH4 conversion also correlated well to the IA1/INi ratios obtained from XPS analyses.Park et al. [225], have described the use of various foam catalyst embedded with metals to enhance the heat transfer of reaction during BRM process. The authors characterised heat transfers based on the Nusselt number and also used a pellet shaped catalyst to improve the heat transfer of the foam catalyst. The results revealed that the Nusselt number of foam catalyst packed with metals was larger than the pellet catalyst used conventionally. Additionally, uniform temperature distribution was noticed in the reformer throughout the catalyst bed along with the foam catalyst. Images of the various metallic foam catalysts are shown in Fig. 34 . Fig. 35 shows that the uncoated Al2O3 and bare Ni foam were reactive without Ni catalyst loading. Additionally, the bare Ni foam displayed greater methane conversion than the uncoated aluminium oxide. The results established that the Ni foam exhibited a significant role in improving heating characteristics of the catalyst bed and mass transfer inside the reactor. However, after wash coating of a layer of Al2O3 on the nickel foam, the conversion of carbon dioxide and methane increased to 34% and 69.1% and, respectively.The syngas flow rate along with molar ratio of hydrogen/carbon monoxide in presence of Nickel/Al2O3/Nickel foam, uncoated Nickel foam, uncoated Al2O3 bead and Al2O3/Ni foam are shown in Fig. 36 . The syngas flow rate exhibited a similar behaviour both for carbon dioxide and methane conversions. Nevertheless, the H2/CO molar ratio was different and furthermore molar ratio of H2/CO of uncoated Ni and uncoated Al2O3 bead was less than 2.0. The metallic foam catalyst is potentially useful for GTL-FPSO applications based on the enhanced mechanical properties of the catalyst and compactness of the reformer. Additionally, higher selectivity levels and activity are associated with nickel inside the coating layer, which serves as active sites for methane, water and carbon dioxide.Brush et al. [181] reported the ability of Ni/Mo2C to catalyse the bi-reforming of methane. The authors noted that by altering the ratio of carbon dioxide: water, the resulting Hydrogen: Carbon monoxide (H2:CO) ratio could be changed from 0.91 to 3.0, which covers a wide range of H2:CO ratios common to various hydrocarbon syntheses. Most importantly, the catalytic activity changed from very high (50% conversion) to very low (10% conversion) within a time interval of 10 min. Additionally, for various inlet feed compositions similar performance was exhibited by the catalyst. However, enhanced activity was followed by greater deactivation shortly after the exposure to stream.Claridge et al. [226], synthesized Mb and W carbide materials of larger surface area to assess the performance of carbide catalysts formed from non-metals for various CH4 reforming reactions including bi-reforming. Their study revealed that carbides activity was similar to those of iridium and ruthenium catalysts used in reforming of methane. Nevertheless, conversion values were in agreement with the values obtained during thermodynamic equilibrium. HRTEM images revealed the absence of deposited macroscopic on the catalysts during the reforming process.The present review article demonstrates a comprehensive review of various catalysts, including Ni, Co, Rh and Pt-based catalysts, used for BRM. It also describes the role of various promoters and supports on the conversion efficiencies of CH4, CO2 and H2/CO ratio. Ni catalysts supported by smaller nanoparticles of Ce–ZrO2 ZrO2, MgO were observed to be extremely stable and active for BRM processes. The degree of coke formation was dependent on each investigated catalyst. Coke formation was found to be more severe in commercial Ni/MgAl2O4 catalyst than Ni/MgO–Al2O3 catalyst, which was attributed to higher dispersion of Nickel over MgO–Al2O3.Bi-reforming of methane over a catalytic nickel membrane for the GTL (gas to liquid) process produced a very high conversion of methane in the range of 92.7–96% above 973 K, when the CO2/H2O feed ratios were in the range of 0–1.0. GTL process possessed two advantages. One was that carbon formation was reduced due to the oxidation of carbon precursor species and a desirable H2/CO was achieved by adjusting CH4/H2O/CO2 ratio in the feed stream. Cerium containing Ni/MgAl2O4 catalysts synthesized by both co -precipitation and impregnation technique exhibited higher metal dispersion than Ni/MgAl2O4 alone and showed marked reducibility at lower temperatures around temperature 550°C as confirmed by XPS since the redox properties of Ce4+/Ce3+ resulted in easier gasification of the settled coke on the surface of the catalyst and also helped in storage and delivering of active oxygen thereby enhancing the dispersion of Ni. The catalyst Ni/γ-Al2O3 promoted very higher conversion of carbon dioxide (82.4%) and methane (98.3%) when subjected to bi-reforming of methane for 200 h since the Ni/Al2O3 catalyst exhibited characteristics such as high metal dispersion, high catalytic activity large specific surface area, and stronger metal support interaction respectively. Presence of MgO in 20 wt % MgO/Ni catalyst was quite effective in preventing coke formation due to the formation stable MgAl2O4 spinel phase at higher temperatures. The presence of CeO2 in Ce1-x –ZrxO2 catalytic systems caused a marked improvement in redox properties, thermal stability, and promotion of metal dispersion and also enhanced the oxygen storage ability of CeO2. A drastic reduction in carbon deposition from 25.96% to 1.08% was observed for a feed composition of CH4: CO2: H2O = 1.0:0.55:0.55 when CO2 reforming was performed in conjunction with steam reforming process in presence of NiO–CaO catalyst due to its high selectivity, activity and productivity in the oxidative conversion of methane to synthesis gas. Nickel/Santa Barabara Amorphous −15 (SBA-15) exhibited enormous resistance both towards sintering and coking. 10% Nickel/3% MgO/Santa Barabara Amorphous 15 catalyst demonstrated higher catalytic performance after reaction for 620 h since the mesoporous SBA-15 support possessed uniform mesopores with thick framework walls, high thermal stability and wide specific surface area. Bimetallic systems such as Sn 0.02 Nickel/Cerium-Al catalyst displayed superior catalytic activity, increased the resistance of carbon formation and remained active over a long period of 92 h in comparison to their own counterparts. The remarkable level of stability and excellent conversions seen in the bi-reforming process has proved the versatility of Sn0.02Ni/Ce–Al catalyst which could be upgraded to variety of CO2 containing feed stocks. Ni-based catalysts supported on mesoporous MgO–Al2O3 resembling a Mg–Al hydrotalcite structure with Mg/Al ratio of 1 demonstrated excellent conversion efficiencies for CH4 (93.7%) and CO2 (75.2%) and higher resistance to coke formation due to its basic property, enhanced steam as well as CO2 adsorption, strong Ni to support interaction of these catalysts. The excellent catalytic performance of Ni 30 wt % SiO2 55 wt % MgO catalyst was ascribed to its acidic strength, enhanced basicity and structural stability under high temperatures. NdCOO3 perovskite-type mixed-oxide catalysts proved to be highly efficient for carbon-free bi-reforming process. Pt–Ce–ZrO2 catalyst exhibited excellent stability in presence of H2O and its stability was attributed to the greater number of O2 vacancies present in the support. Lanthanum promoted catalysts exhibited greater nickel dispersion than Ni/MgAl2O4 catalyst due to their enhanced interactions between the metal and support. Ni–2.5La/MgAl2O4 catalyst showed maximum sinter stability and activity due to its enhanced nickel dispersion and surface area.It is highly essential to develop Ni-based catalysts containing bi-metallic and tri-metallic configuration since Ni displayed stronger stability and enhanced activity. However, the biggest constraint in this approach is the coke formation. Bi-reforming process is endothermic and would require higher activation energy to achieve the synthesis. Future studies should be undertaken to design an appropriate bi-metallic and tri-metallic catalyst that can be suitable at lower temperatures. Furthermore, the approach towards catalyst formation has a significant influence on the catalyst's capability. Therefore, selecting a suitable catalyst and its synthesis technique can provide improved SMSI, superior activity, enhanced Ni dispersion on the catalytic support and stability against coke formation.Among the evaluated inlet feed compositions, conducting bi-reforming process under a stoichiometric feed composition (CH 4: CO 2: H 2 O = 3:1:2) is considered the ideal one for selective production of syngas within the operating temperature of 750–800 °C range. One of the major drawback associated with nano-based catalysts is that the up scaling of catalyst preparation from laboratory batches to continuous industrial production. Henceforth, development of reproducible and economical synthetic strategies is imperative for linking all advantages of nano-based catalysts to large-scale metgas generation facilities. XPS X-ray photo electron Spectroscopy XRD X-ray diffraction FT Fischer-Tropsch Synthesis pNirb Thick porous Nickel carbon H2- TPR Hydrogen Temperature Programmed Reduction CO2 –TPD Carbon dioxide Temperature Programmed Desorption TEM Transmission electron Microscopy DRM Dry reforming of methane GHG Greenhouse gas emissions SRM Steam reforming of methane POM Partial oxidation of Methane BRM Bi-reforming of methane CRM Carbon dioxide reforming of methane SEM Scanning Electron Microscopy TPR Temperature Programmed Reduction GHSV Gas Hourly Space Velocity SBA-15 Santa Barbara Amorphous-15 EISA Evaporation Induced Self-assembly WHSV Weight hourly space velocity FTIR Fourier Transfer Infrared Spectroscopy TGA Thermogravimetric analysis DTA Differential Thermal Analysis Ceria Cerium (IV) Oxide Zirconia Zirconium dioxide SGM Sol-Gel method CG Commercial method HTM Hydrothermal Method GTL-FPSO Floating Production, Storage and Offloading X-ray photo electron SpectroscopyX-ray diffractionFischer-Tropsch SynthesisThick porous Nickel carbonHydrogen Temperature Programmed ReductionCarbon dioxide Temperature Programmed DesorptionTransmission electron MicroscopyDry reforming of methaneGreenhouse gas emissionsSteam reforming of methanePartial oxidation of MethaneBi-reforming of methaneCarbon dioxide reforming of methaneScanning Electron MicroscopyTemperature Programmed ReductionGas Hourly Space VelocitySanta Barbara Amorphous-15Evaporation Induced Self-assemblyWeight hourly space velocityFourier Transfer Infrared SpectroscopyThermogravimetric analysisDifferential Thermal AnalysisCerium (IV) OxideZirconium dioxideSol-Gel methodCommercial methodHydrothermal MethodFloating Production, Storage and OffloadingThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors gratefully acknowledge the support from Edith Cowan University, Western Australia, Australia.

Today, bi - reforming of methane is considered as an emerging replacement for the generation of high-grade synthesis gas (H2:CO = 2.0), and also as an encouraging renewable energy substitute for fossil fuel resources. For achieving high conversion levels of CH4, H2O, and CO2 in this process, appropriate operation variables such as pressure, temperature and molar feed constitution are prerequisites for the high yield of synthesis gas. One of the biggest stumbling blocks for the methane reforming reaction is the sudden deactivation of catalysts, which is attributed to the sintering and coke formation on active sites. Consequently, it is worthwhile to choose promising catalysts that demonstrate excellent stability, high activity and selectivity during the production of syngas. This review describes the characterisation and synthesis of various catalysts used in the bi-reforming process, such as Ni-based catalysts with MgO, MgO–Al2O3, ZrO2, CeO2, SiO2 as catalytic supports. In summary, the addition of a Ni/SBA-15 catalyst showed greater catalytic reactivity than nickel celites; however, both samples deactivated strongly on stream. Ce-promoted catalysts were more found to more favourable than Ni/MgAl2O4 catalyst alone in the bi-reforming reaction due to their inherent capability of removing amorphous coke from the catalyst surface. Also, Lanthanum promoted catalysts exhibited greater nickel dispersion than Ni/MgAl2O4 catalyst due to enhanced interaction between the metal and support. Furthermore, La2O3 addition was found to improve the selectivity, activity, sintering and coking resistance of Ni implanted within SiO2. Non-noble metal-based carbide catalysts were considered to be active and stable catalysts for bi-reforming reactions. Interestingly, a five-fold increase in the coking resistance of the nickel catalyst with Al2O3 support was observed with incorporation of Cr, La2O3 and Ba for a continuous reaction time of 140 h. Bi-reforming for 200 h with Ni-γAl2O3 catalyst promoted 98.3% conversion of CH4 and CO2 conversion of around 82.4%. Addition of MgO to the Ni catalyst formed stable MgAl2O4 spinel phase at high temperatures and was quite effective in preventing coke formation due to enhancement in the basicity on the surface of catalyst. Additionally, the distribution of perovskite oxides over 20 wt % silicon carbide-modified with aluminium oxide supports promoted catalytic activity. NdCOO3 catalysts were found to be promising candidates for longer bi-reforming operations.

With the rapid development of the global economy, energy crisis and environment issues have become increasingly prominent. Carbon dioxide (CO2) is a primary greenhouse gas, while it could also be a valuable carbon source. In recent years, electrochemical CO2 reduction reaction (ECR) has received considerable attention among various CO2 conversion technologies due to numerous advantages [1,2]. For instance, ECR can be driven under ambient temperature and pressure using renewable energy source such as wind and solar power [3,4]. The external voltages as well as electrolytes solutions can be adjusted for the generation of specific products. Furthermore, ECR technology could not only mitigate CO2 emission, but also effectively convert CO2 to value-added chemicals and fuels, which has been regarded as an appealing technology path for closing the carbon circle [5–7]. Nevertheless, physicochemical properties of the CO2 molecule make electrochemical conversion of CO2 challenging [8,9]. In the past decades, metals or related oxides, carbon-based materials and nanocomposites have been widely investigated as electrocatalysts for ECR. Despite that great progress has been made to exploit electrocatalysts for ECR, the process is still impeded by the sluggish kinetics, poor product selectivity, catalyst stability, and high overpotential [10–13]. Taking copper as an example, it can electrochemically convert CO2 to CH4 with a high overpotential around 0.9 V, but the selectivity to specific products is low [14,15]. Therefore, it is highly desirable to develop electrocatalysts with high activity and selectivity for ECR.It has been well established that the catalytic activity can be improved by reducing the size of catalysts. Specifically, single atoms catalysts (SACs) with single atom as active centre have aroused huge interest due to maximum atom utilization and excellent performance in various catalytic reactions such as water splitting, CO2 reduction, N2 reduction, etc. [16–20]. It is worth noting that the preparation of SACs is a challenge because single atoms with high surface energy are easy to aggregate into clusters, leading to catalyst deactivation. Thus, a suitable support which could offer anchoring sites and possesses good stability will greatly improve the activity of SACs. Two-dimensional (2D) materials are appealing substrates for anchoring single transition metal atom due to their unique structures and electronic properties [21–23]. Moreover, it has been disclosed that the interaction between p-orbital of substrates and d-orbitals of single transition metal atom are beneficial for highly effective electrocatalysts [24,25]. For example, single transition metal atoms embedded into graphene, graphitic carbon nitride (g-C3N4), graphyne, boron nitride (BN), phthalocyanine (PC), etc. have been widely studied as promising SACs for CO2 reduction [26–29].Recently, a family of 2D transition metal carbides, nitrides and carbonitrides, known as MXenes, were reported and synthesized from the layered Mn+1AXn phase [30]. In Mn+1AXn, M stands for early transition metals, A denotes the group 13 or 14 elements, X denotes C or N, and n is between 1 and 3 [31]. There are a variety of MXenes that have been predicted and synthesized experimentally, which are explored for applications in many fields [32]. For instance, Li et al. reported that MXenes from the group IV to VI series are active for CO2 capture, while Cr3C2 and Mo3C2 are promising catalysts for CO2 conversion to CH4 [33]. Both simulation study and experimental work have shown that MXenes have large surface areas, excellent electronic conductivity, tunable surface composition and great stability [32,34]. Thanks to these merits, MXenes have also been demonstrated to be promising substrates for anchoring single transition metal atoms in catalytic reactions [35]. Generally, Mn+1Nn is more difficult to synthesize compared to Mn+1Cn. Interestingly, it has been demonstrated that nitride MXenes exhibit better conductivity in comparison with carbide MXenes [36]. During synthesis, the basal plane of MXenes could be functionalized by various atoms or groups including O, OH and F, which affect their inherent properties [30]. Recent investigations have demonstrated that under high temperature treatment the F group can be eliminated and OH group can be converted to O groups [37,38]. Studies have confirmed that these different functional groups could tune the work function and electronic properties of MXenes [39].Nb-based MXenes have gained great attention in energy storage and conversion [40,41]. Pt-doped Nb-based MXene has been reported to be an excellent bifunctional OER/ORR catalyst [42], while nitride Nb2N for ECR has not been studied. In this work, for the first time, we investigated the single transition metal atoms (V, Cr, Mn, Fe, Co, Ni) embedded O group terminated Nb2N monolayer (Nb2NO2) as ECR catalysts by first-principles calculation. It is found that Nb2NO2 can be an ideal support for anchoring sing TM atoms because of excellent stability and conductivity. TM@Nb2NO2 show excellent CO2 adsorption capacity, which benefits CO2 activation and reduction. Among six SACs catalysts, V, Cr and Ni@ Nb2NO2 are identified as efficient electrocatalysts for ECR to CH4, with smaller limiting potential of − 0.45, − 0.47 and − 0.28 V, respectively. Meanwhile, the origin of the ECR activity was revealed by several key descriptors.All calculations were carried out by spin-polarized density functional theory (DFT) in the Vienna Ab initio Simulation Package (VASP) with projector augmented wave (PAW) [43,44]. The generalized gradient approximation (GGA) implemented Perdew-Burke-Ernzerhof (PBE) was used to calculate the exchange-correlation energy [45,46]. The empirical correction (DFT-D3) was employed to describe the van der Waals (vdW) interactions [47]. The parameter for dipole correction along z-direction are considered in our calculations. DFT+U calculations are also considered for single TM atoms. The values of U−J were set to be 2.72, 2.79, 3.06, 3.29, 3.42 and 3.4 eV for V, Cr, Mn, Fe, Co, and Ni, respectively [18]. A 3 × 3 × 1 TM@Nb2NO2 containing 45 atoms was constructed by anchoring one TM atom in site1 (N site) and site 2 (Nb site) ( Fig. 1a). The 18 Å thickness vacuum region in the z-direction was added to eliminate the spurious interactions from periodic boundary. The cutoff energy was set to 500 eV. The K-points for geometry optimization and electronic calculations were set to be 6 × 6 × 1 and 10 × 10 × 1, respectively. The convergence of energy and force was set to be 1.0 × 10−5 eV and 1.0 × 10−2 eV/Å, respectively. Solvent effect was included in our calculations by using implicit solvent model based on VASPsol, and the dielectric constant of water was 78.4 [48]. Moreover, to explore structure stability, the ab initio molecular dynamics (AIMD) simulation was performed in NVT ensemble and phonon spectra was calculated based on the density functional perturbation theory (DFPT) [49,50]. The TM atom transition energy barrier on Nb2NO2 monolayer was calculated by the climbing image nudged elastic band method (CINEB) [51], transition states were confirmed by vibration frequency analysis. The Bader charge analysis was used to analyze electron transfer [52].The binding energy (Eb) of TM atoms on Nb2NO2 monolayer was calculated by Eq. 1: (1) Eb = E(TM@Nb2NO2) − E(TM) − E(Nb2NO2) where E(TM), E(Nb2NO2) and E(TM@Nb2NO2) denote the total energies of single TM atom, Nb2NO2, and TM@Nb2NO2, respectively. With such definition, a more negative value indicates a stronger binding of TM atoms to the Nb2NO2 substrate.The gas adsorption energy (Eads) was calculated by Eq. 2: (2) Eads = E(gas@Nb2NO2) − E(gas) − E(Nb2NO2) where E(gas), E(Nb2NO2) and E(gas@Nb2NO2) are the total energies of gas, Nb2NO2 and gas adsorbed Nb2NO2.The Gibbs free energy of ECR were calculated by the computational hydrogen electrode (CHE) method [53]. After intermediate was adsorbed on the surface of catalyst, the translation and rotation freedom could be ignored and only vibration freedom is contributed to the entropy. The free energy of H+/e- pair is equivalent to the chemical potential of H2 at standard conditions. The Gibbs free energy change (ΔG) can be obtained by Eq. 3: (3) ΔG = ΔE + ΔE(ZPE) − TΔS + ΔG(pH) + ΔG(U) in which ΔE was the energy difference between reactants and products directly obtained from DFT calculations, ΔE(ZPE) and ΔS are zero-point energy correction and entropy change at temperature T of 298.15 K. ΔG(pH) is the free energy correction due to the effect of H concentration, and was calculated by the formula ΔG(pH) = k BTln10 × pH. In this work, the pH value was set to be zero under acidic condition. ΔG(U) is the contribution of the applied electrode potentials. The limiting potential (UL) from potential-determining step (PDS) can be obtained by Eq. 4: (4) UL = −ΔGmax/e where ΔGmax is the maximum free energy change in the ECR process along the most favourable pathway.After geometry optimization, the obtained lattice parameter a of clean Nb2N monolayer is 3.11 Å, consistent with previous study [54]. Nb2N monolayer shows a hexagonal symmetry with P63/mmc space group. O was then added on the centre of three Nb atoms, similar to the O functionalized Ti2C MXene (Fig. 1a and b) [55]. The binding energy (Eb) of O on Nb2N monolayer was calculated by the equation: Eb = (E(Nb2NO2) − E(O2) − E(Nb2N))/2, where E(Nb2NO2), E(O2), E(Nb2N) are the total energy of Nb2NO2, O2 and Nb2N [56]. A negative value of Eb = –5.34 eV demonstrates that Nb2N monolayer can be easily covered by O atoms. It is possible for O group transforming to OH during ECR process. Therefore, we calculated the Gibbs free energy for H adsorption on O atoms, with ΔG*H of −0.16 eV. A moderate ΔG*H indicates that the proton can easily adsorb on and desorb from the surficial O atom, which may promote protonation of the ECR intermediates. The phonon curves and AIMD simulation were performed to check its stability, as shown in Fig. 1c and Fig. S1. There are no imaginary bands in phonon spectra. The fluctuation of the total energy of Nb2NO2 is quite small and around the equilibrium. Meanwhile, the structure does not show any obvious changes, confirming that Nb2NO2 monolayer possesses excellent stability. On the other hand, the calculated density of state of Nb2NO2 exhibits metallic behaviour, indicting good capability for electron transfer (Fig. 1d). This endows Nb2NO2 monolayer excellent electrical conductivity, a prerequisite for an ideal substrate for SACs used in ECR.The stability of TM anchored Nb2NO2 will be the key for the synthesis and application of MXene based SACs. As presented in Fig. 1a and b, there are two possible anchoring sites for single TM atoms: (1) the centre site between three neighbouring N atoms and the top of Nb atom (Nb site), (2) the centre site between three neighbouring Nb atoms and the top of N atom (N site). After structure relaxation, the anchored TM atoms have slight effects on lattice parameters a. The thermodynamic stabilities of TM@Nb2N were investigated by calculating Eb (Fig. 1e and Table S1). Notably, a more negative value of Eb on N site indicates that TM atoms prefer to bind on N site. Moreover, the transition energy barriers (ET) of single TM atoms from N to Nb site were calculated to evaluate its kinetic stability. The ET were calculated by ET = ETS − EIS, in which ETS is the total energy of transition state (TS) from N to Nb site, while EIS is the total energy of TM embedded in N site. As shown in Table S1, the ET of TM atoms are quite large in the range of 0.87–2.58 eV, implying that it is difficult for TM atoms to diffuse and aggregate into clusters. These results suggest that single TM atom can firmly anchor on N site. We therefore will only consider this site as active site for further study. In addition, Bader charge analysis show that the charge transfer from TM atoms to substrate decreases with the atomic number. Consequently, V and Cr atom present higher oxidation state (+1.08 and +1.02), while Ni atom shows lower oxidation value (+0.50).CO2 adsorption on the surface of electrocatalysts is important for CO2 activation and transformation into intermediates such as *COOH and *OCHO [57]. The optimized CO2 adsorption configurations on TM@Nb2NO2 were shown in Fig. 2. Obviously, the carbon or oxygen atom of CO2 molecule is absorbed on TM atoms. Meanwhile, it can be observed that CO2 molecule is not absorbed on TM@Nb2NO2 in linear state, but with a certain degree of bending. The corresponding adsorption energies, bond lengths of C−TM and O−TM, bond angels of CO2 molecule, and charge transfer between TM and CO2 molecule are summarized in Table 1. It is clear that the bond angle of CO2 molecule on TM@Nb2NO2 increases with the atomic number, ranging from 138.39° to 154.34°. Specially, V@Nb2NO2 greatly deviated from the linear state, which indicates higher CO2 adsorption capacity. The bond lengths of C−TM and O−TM are quite close to 2.00 Å, demonstrating strong adsorption between substrate and CO2 molecule, consistent with previous studies [26,58]. The CO2 adsorption energies on TM@Nb2NO2 range from − 0.77 and − 0.30 eV. The negative values indicate that CO2 adsorption on the SACs is thermodynamically favourable. V, Cr and Ni@Nb2NO2 exhibit relatively strong interaction with CO2. Bader charge analysis confirm that there is a significant net charge transfer from V, Cr and Ni atoms to CO2, with a value of − 0.60e, − 0.53e, − 0.31e, respectively. Thus, CO2 molecules can be effectively activated by V, Cr and Ni@Nb2NO2, and these three SACs potentially exhibit high performance for producing specific ECR products.The ECR process starts with the hydrogenation of CO2 molecule to form *COOH (* + CO2 + (H+ + e−) → *COOH) or *OCHO (* + CO2 + (H+ + e−) → *OCHO) on active centres by H atom binding O or C atom. However, the side-reaction HER (* + H+ + e− → *H) may occur due to the direct interaction between proton and TM atoms, resulting in low ECR selectivity. It has been widely accepted that the Gibbs free energy change (ΔG) for *COOH/*OCHO and *H formation can be used to evaluate the ECR selectivity versus HER selectivity [59]. Therefore, ΔG*COOH/*OCHO were calculated and compared with ΔG*H. As plotted in Fig. 3, all TM@Nb2NO2 electrocatalysts prefer ECR (below the diagonal) to HER (above the diagonal). Notably, V, Cr and Fe@Nb2NO2 are ECR selective with two favourable initial protonation processes (*COOH and *OCHO), while Ni, Co and Mn@Nb2NO2 exhibit ECR selectivity only with one favourable initial protonation step (*COOH or *OCHO). Meanwhile, ΔG*OCHO is smaller than ΔG*COOH for V, Cr and Fe@Nb2NO2, demonstrating that the formation of *OCHO is more energetically favourable. Therefore, the *COOH reduction path and the corresponding CO product will not be considered on these three SACs in the further protonation process.The reduction products from CO2 could involve C1, C2 and C3 due to complex protonation and C−C coupling. However, the formation of high carbon products (C2+) is impossible because C−C coupling will not occur on SACs. Therefore, only C1 products by accepting 2e to 8e electrons, including CO, HCOOH, HCHO, CH3OH, CH4, were investigated in this work. These different products are formed by different number of protons binding C or O atoms. A possible pathway was plotted in Fig. 4 by taking the optimized configuration of intermediates on Fe@Nb2NO2 as an example. It is obvious that only TM atom binds with the C or O atoms during the whole ECR process, demonstrating TM atom as active site.After *COOH or *OCHO formation via accepting first proton-electron pair, further hydrogenation by obtaining a second proton-electron pair will produce *OCHOH or *CO intermediates. Therefore, the binding strength between these two intermediates and active centre will decide HCOOH or CO generation. We calculated the Eads of HCOOH and CO on TM@Nb2NO2 ( Table 2). For HCOOH formation from *OCHO, V, Cr, Mn and Fe@Nb2NO2 show a large Eads with − 1.07, − 1.25, − 0.87 and − 1.51 eV, respectively. For CO formation from *COOH, the Eads of CO on Co and Ni@Nb2NO2 are − 1.95 and − 1.73 eV, respectively. It means that both HCOOH and CO could be further protonated on these SACs instead of desorbing from the SACs as final products. In addition, the generation of *OCHOH or *CO on Fe, Co, Cr, Ni and V@Nb2NO2 are overall exothermic. The generation of *OCHOH is only slightly endothermic on Mn@Nb2NO2, benefiting the further reduction of intermediates.For further hydrogenation of *OCHOH or *CO, three possible intermediate including *CHO, *COH and *OCH could be generated. Notably, *COH from *CO (*CO + (H+ + e−) →*COH) on Co and Ni@Nb2NO2 underwent a larger energy uphill in comparison with the formation of *CHO (*CO + (H+ + e−) →*CHO). Similarly, For V, Cr, Mn and Fe@Nb2NO2, the formation of *OCH from*OCHOH (*OCHOH + (H+ + e−) →*OCH + H2O) are more energy consuming than the production of *CHO. Thus, it can be concluded that *CHO will be the key intermediates for the third hydrogenation process.*OCH2 and *CHOH intermediates can be produced after *CHO accepting the fourth proton-electron pair. It is evident from Fig. 5 that the ΔG of *CHOH on these six SACs show energy uphill, while the ΔG for the formation of *OCH2 on these SACs show energy downhill. Therefore, these TM atoms exhibit strong oxophilicity to form TM−O bonds. The four-electron product HCHO will be desorbed from the electrocatalyst if the interaction between *OCH2 and TM atom is too weak. The calculated Eads of HCHO on these SACs are in the range of − 1.59 to − 0.96 eV, suggesting that it is difficult for HCHO to desorb and thus can be further reduced.*CHOH then accepted the fifth proton-electron pair to produce *CH and *CH2OH intermediates, while the hydrogenation products of *OCH2 are *OCH3 and *OHCH2. However, ΔG of *CH and *OHCH2 are energetically unfavourable and will not form. In contrast, *CH2OH and *OCH3 will be the key intermediates and participate in later hydrogenation. CH3OH is the six-electron product via *OCH3 + (H+ + e−) →*OHCH3 → * + CH3OH. Nevertheless, the formation of *OHCH3 only show energy downhill on Fe and Mn@Nb2NO2. The Eads on V, Cr, Mn, Fe, Co, Ni@ Nb2NO2 is − 1.15, − 1.17, − 0.88, − 0.75, − 1.07 and − 1.03 eV, respectively. Thus, CH3OH can still be stably bonded with SACs and further reduced. The eight-electron product CH4 can be generated from diverse paths such as *CH3 + (H+ + e−) → * + CH4, *OCH3 + (H+ + e−) → CH4 + *O and *OHCH3 + (H+ + e−) → CH4 + *OH. Remarkably, the Eads of CH4 on TM@Nb2NO2 are significantly smaller than the other C1 products, ranging from − 0.47 to − 0.23 eV, indicating that CH4 can easily desorb from the SACs and become the final product. According to principle of minimum free energy increase at each step, the optimized paths for ECR to CH4 on TM@Nb2NO2 were concluded as below (Fig. 5): (I) V,Cr and Fe@Nb2NO2:*+CO2 → *OCHO → *OCHOH → *CHO → *CHOH → *CH2OH → *CH2 → *CH3 → * + CH4 (II) Mn@Nb2NO2:*+CO2→ *OCHO → *OCHOH → *CHO → *OCH2 → *OCH3 → *OHCH3 → *OH + CH4 → * + H2O (III) Co and Ni @Nb2NO2: * + CO2 → *COOH → *CO → *CHO → *OCH2 → *OCH3 → *OHCH3 → *OH + CH4 → * + H2O Thus, TM@Nb2NO2 can be promising candidates in electrochemically converting CO2 to CH4.To evaluate the ECR performance of TM@Nb2NO2, the PDSs and the corresponding UL were summarized in Table 2. Generally, the lower the value of UL, the higher the activity of SACs. In path I and II, *OCHOH → *CHO was identified as PDS for V, Cr, Mn and Fe@Nb2NO2. The UL for CH4 generation on these four SACs are − 0.45, − 0.47, − 0.62 and − 0.89 V. The PDS of Co and Ni@Nb2NO2 in path III is *CO → *CHO, and the corresponding UL are − 0.57 and − 0.28 V. Intriguingly, UL for the ECR to CH4 on V, Cr, Co and Ni@Nb2NO2 are lower than that the state-of-the-art catalyst Cu (211) (−0.74 V) [60], demonstrating potentially excellent performance of TM@Nb2NO2 for ECR to CH4. Particularly, the UL of Ni@Nb2NO2 is among the best reported in literature. Finally, we investigated the stability of Ni@Nb2NO2 by AIMD simulations with a time step of 3 fs at the temperature of 300 K for 18 ps (Fig. S2). It can be found that Ni atom can still stay at the vacancy, which evidenced that diffusion will not occur.The excellent activity of TM@Nb2NO2 for CH4 generations is mainly related to the interaction between active TM atom and substrate. As shown in Table S1, the Nb2NO2 monolayer is negatively charged by TM atoms. The different amount of charge transfer indicates the different interaction strength. Consequently, the single TM atom with different positive charge will contribute to different catalytic activity. After intermediate adsorption on the single TM atom, the binding strength between them will directly determine UL. According to the Sabatier principle, too strong or too weak binding strength will result in low catalytic activity [61]. As shown in Fig. 6, the PDOS of key intermediates from PDSs of TM@Nb2NO2 exhibit deferent interaction between TM and C or O atoms. For instance, the Mn-3d orbitals and the O-2p orbitals of *OCHOH in Mn@Nb2NO2 have slight overlap, contributing to weak binding strength. Fe-3d orbitals interacts greatly with O-2p, exhibiting strong interaction. The corresponding adsorption energy of OCHOH on these two SACs is − 0.95 and − 1.51 eV, suggesting that too strong or weak interaction could increase the free energy of PDSs.We further investigated the activity origin on TM@Nb2NO2 by using descriptors. The PDSs of TM@Nb2NO2 can be assigned to *OCHOH and *CO, therefore we distinguish them by two different areas (palegreen and slateblue in Fig. 7). Since the d-band centre of TM atoms has often been used to correlate the catalytic properties, the locations of d band centres (ε) were calculated and plotted against UL, as shown in Fig. S3 and Fig. 7a. With the increase of the TM-d electron number, ε shifts to a more negative energy level, resulting in the increase of UL. When the key intermediate is *OCHOH, there is a good linear relationship between ε and UL (UL = 0.35ε – 0.40, R2 = 0.97). For *CO as key intermediate, only Co and Ni@Nb2NO2 are distinguished. Generally, the more negative the value of ε, the weaker the adsorption between intermediates and catalysts. For example, it can be found that Mn@Nb2NO2 shows a lower ε, while the Eads for *OCHOH is smaller, indicating weak adsorption and a large UL. However, ε is not associated with Eads for a specific TM atom in a small range, because of the neglect of the d-band shape and the effect of the TM-s and p orbitals. Thus, the linear relationship is not apparent (Fig. 7b). For *CO intermediate, the higher ε of Co atoms contributed a strong Eads of *CO and high UL.The crystal orbital Hamilton populations (COHP) were employed to analyse the bonding and antibonding states of the TM and key intermediates *OCHOH and *CO [13]. Meanwhile, the integrated COHP (ICOHP) was calculated to give a more quantitative explanation (Fig. S4). For O atom bonding with V, Cr, Mn and Fe@Nb2NO2, it shows obvious antibonding states below Fermi level, demonstrating weak adsorption. The corresponding ICOHP values are − 1.32, − 1.44/− 1.57, − 1.47/− 1.75, − 1.28/− 1.61 eV, respectively. V and Cr@Nb2NO2 have similar antibonding states in spin up state, resulting in similar UL. For C atom bonding with Co and Ni@Nb2NO2, there is no antibonding state below Fermi level with value of − 2.56/− 2.66 and − 2.40 eV, respectively, indicating strong adsorption. The more negative the ICOHP, the more stable of bonding, thus Fe@Nb2NO2 shows a large UL. A good linear relationship between ICOHP and UL was obtained for V, Cr, Mn and Fe@Nb2NO2 (UL = 1.58Φ + 1.70, R2 = 0.86), disclosing the role of different metal centres in the bonding/antibonding orbital populations.Recently, charge transfers of active atoms have been reported as descriptor to explain the performance of catalysts [62]. Herein, we calculated the valence state (δ) of TM atoms after adsorbing intermediates. The δ of different atoms for different binding atoms vary in a range from + 0.55 to + 1.32, indicating an increase of charge transfer from TM atoms after intermediates adsorption and different interaction strength between them. Fe atom had the largest Δδ increase of 0.37 after intermediates adsorption, implying a possible strong interaction between Fe and *OCHOH and a large UL. Meanwhile, an approximate linear relationship (UL = −3.02Δδ −0.31, R2 = 0.86) was obtained, demonstrating that binding strength between catalysts and intermediates can be represented by Δδ. Therefore, ε, Φ and Δδ can be used as descriptors to describe the activity origin well. Meanwhile, the Eads can be a nominal descriptor for the ECR activity to CH4 due to the close connection between energy and electronic structure, while ε, Φ and Δδ can quantitively describe the intrinsic activity of ECR to CH4 on TM@ Nb2NO2. Overall, the results show that Ni@Nb2NO2 is the best ECR catalyst for CH4 generation, while Fe@Nb2NO2 is not an ideal catalyst.Single TM atoms (V, Cr, Mn, Fe, Co and Ni) anchored Nb2NO2 monolayer as potential SACs for electrochemical CO2 reduction were studied by first-principles calculation. Results demonstrate that TM atoms can be stably embedded into N site and will not aggregate into clusters. CO2 molecules can be effectively activated by V, Cr and Ni@Nb2NO2 due to charge transfer and large adsorption energy. All TM@Nb2NO2 electrocatalysts exhibit high selectivity for ECR in comparison with HER. The Eads of C1 products (CO, HCOOH, HCHO, CH3OH) is too large for them to desorb from the surface of catalysts, while CH4 can easily desorb due to the small Eads. The PDS on these SACs for ECR to CH4 can be divided into two categories: *OCHOH to *CHO for V, Cr, Mn and Fe, and *COOH to *CHO for Co and Ni. The UL for CH4 generation on V, Cr and Ni@Nb2NO2 SACs are − 0.45, − 0.47 and − 0.28 V, exhibiting high performance for ECR to CH4 and is even better than the Cu (211) electrocatalyst. Furthermore, the adsorption energy of the key intermediates (Eads) can serve as a nominal descriptor to indicate ECR activity, while d band center (ε), ICOHP (Φ), the change of valence state (Δδ) can quantitatively describe the ECR activity. This work demonstrated that MXene based earth abundant metal SACs are promising for electrocatalytic CO2 reduction. Song Lu: Methodology, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization. Yang Zhang: Investigation, Formal analysis, Writing – review & editing. Fengliu Lou: Investigation, Formal analysis, Writing – review & editing. Zhixin Yu: Conceptualization, Formal analysis, Validation, Resources, Supervision, Writing – review & editingThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work is supported by the Norwegian Ministry of Education and Research. The computations were performed on resources provided by UNINETT Sigma2 - the National Infrastructure for High Performance Computing and Data Storage in Norway.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jcou.2022.102069. Supplementary material .

The design of highly efficient catalysts for electrochemical reduction CO2 (ECR) to value-add chemicals and fuels is important for CO2 conversion technologies. In this work, earth abundant transition metal (TM = V, Cr, Mn, Fe, Co and Ni) atoms embedded into two-dimensional (2D) Nb2NO2 (TM@Nb2NO2) as single-atom catalysts (SACs) for ECR was investigated by first-principles study. We demonstrated that Nb2NO2 can be an excellent substrate for anchoring single TM atom due to its excellent stability and electronic conductivity. Besides, V, Cr and Ni@Nb2NO2 could effectively promote CO2 adsorption and reduction. All TM@Nb2NO2 exhibit high selectivity towards CH4, and V, Cr and Ni@Nb2NO2 show low limiting potentials. The activity origin was revealed by analysing adsorption energy, d band centre, bonding/antibonding population and the change of valence state of TM atoms.

municipal solid wastewater-gas shiftvolume percentageweight percentageSyngas obtained during biomass/municipal solid waste (MSW) gasification is mainly a mixture of carbon monoxide, carbon dioxide, hydrogen, methane and nitrogen, which can be utilized for electric power generation or liquid fuel synthesis [1]. The biomass- and MSW-derived syngas, however, contains significant concentrations of impurities such as tar, HCl, alkali chlorides, particulate matter, ammonia, HCN and sulfur compounds. Tar, consisting of a mixture of aromatic hydrocarbons, causes equipment failure by its condensation and corrosion upon cooling of syngas [2–4]. The techniques that can efficiently remove tar compounds to the acceptable levels are still under development. One of the prospective techniques is catalytic steam reforming which converts tar into H2 and CO [5–7]. Different types of natural minerals (e.g., dolomite, olivine and clay minerals) and synthetic catalysts (e.g., char, activated alumina and metal-based catalysts) were proposed for tar reforming, among which Ni-based catalysts are the most common and commercially available. The utilization of Ni-based catalysts enhances syngas production due to steam reforming of hydrocarbons and other catalyzed reactions, including dry reforming, WGS and Bodouard reactions [8–12]. Furthermore, Ni-based catalysts facilitate simultaneous decomposition of NH3 and HCN into N2 and H2 during the reforming process, resulting in lower NOx emissions [13,14].Besides nickel, other metals as well as bimetallic and polymetallic composites have been extensively investigated as reforming catalysts [15–23]. For instance, monometallic Fe and bimetallic Ni-Fe catalysts have shown satisfactory reforming activity and high catalytic stability during reforming of tar compounds under certain conditions [24–28]. Noichi et al. [24] found that higher Fe content in Fe-Al catalysts enhanced the catalytic steam reforming activity by increasing naphthalene conversion efficiency. In NiO–Fe2O3–Al2O3 catalysts developed by Dong et al. [25] and Margossian et al. [26], syngas production and dry reforming activities of methane were influenced by Fe content. Furthermore, catalysts with optimized Fe content were reported to enhance thermal stability of the Ni-Fe catalysts by mitigating coke formation during tar reforming [25,26]. This superior effect was attributed to the formation of Ni-Fe alloys enriched with Fe-O species at the surface of nanoparticles that could catalyze coke oxidation [27,28].It is well known that impurities present in syngas (e.g., particles, sulfur and chlorine species) can poison the catalysts during steam reforming process [29–35]. H2S is notorious poison for catalysts and only a few ppmv of H2S could rapidly deactivate a Ni-based reforming catalyst [29–32]. Upon contact with Ni-based catalyst, sulfur species (e.g., H2S) chemisorb on metal sites forming NiS according to the reaction (1) decreasing the accessibility of active sites to hydrocarbons [29]: (1) Ni + H 2 S ⇌ Ni - S + H 2 Yung et al. [30] have attempted to regenerate the spent Ni catalyst which was contaminated during catalytic tar reforming at 850 °C by 43 ppmv H2S in syngas produced from gasification of white oak. It was found that the Ni-S species in catalyst could not be completely removed during the steam/air regeneration procedure. As a result, the catalytic activity of Ni was only partially recovered and was lower than its initial activity levels [30]. The low melting point and high surface mobility of NiS can accelerate sintering [31], which may deteriorate the activity of catalyst. Furthermore, sulfur species can increase the carbon deposition on catalyst surface, which also decreases the catalytic activity [32].The presence of HCl in syngas was reported to decrease the reforming and WGS activities of Ni catalysts [33–37]. Richardson et al. [33] found that the conversion of methane was extremely inhibited in the presence of HCl, due to the chemisorption of HCl by Ni. Coute et al. [36] demonstrated that HCl induced detrimental effect on WGS activity during steam reforming of chlorocarbons. Veksha et al. [37] investigated the mechanism for the activity loss of Ni catalysts during naphthalene reforming in the presence of 2000 ppmv HCl and demonstrated that naphthalene conversion is not influenced by HCl while WGS activity was poisoned due to the sintering of Ni. In the above mentioned studies, either H2S or HCl were present in gas streams during the reforming while in real syngas, these impurities are present simultaneously. To what extent the co-existence of both H2S and HCl in the gas can influence the catalytic activity during steam reforming has not yet been investigated.The purpose of this work was to investigate the influence of H2S and HCl on the poisoning of synthesized and commercial catalysts during steam reforming of tar. It has been well known that Ni is an excellent metal for steam tar reforming. In this study, the addition of Fe is attempted, because Fe is a low cost material and Fe species has high redox activity [38]. Furthermore, the addition of Fe to Ni had beneficial effect on the performance of bimetallic catalysts under certain experimental conditions [24–28]. Four synthesized catalysts with different loadings of Ni and Fe on alumina support and two commercial catalysts were tested in a fixed bed reactor at different temperatures with varying contents of H2S and HCl in gas. Naphthalene was used as a model tar compound as it is one of the major tar species [39] which also has high stability during tar reforming [40–42]. In this study, 50 ppmv H2S and 300 ppmv HCl were used as they are in the range of typical concentrations of H2S and HCl present in syngas produced from biomass/MSW [29–31,34–37]. The individual and combined effects of impurities on reforming and WGS activities of the catalysts and the reversibility of the catalyst poisoning are discussed.Four catalysts with different Ni and Fe contents were synthesized using the method described elsewhere [37]. Briefly, the catalysts were prepared by impregnation of aluminum hydroxide (H3AlO3·xH2O, Sigma-Aldrich) having particle sizes of 0.56–1.18 mm with known concentrations of Ni(NO3)2·6H2O (Sigma-Aldrich) and Fe(NO3)3·9H2O (Sigma-Aldrich) in aqueous solution. After evaporation of water in a rotary evaporator (Hei-Vap Precision, Heidolph Instruments, Germany), the materials were dried overnight in an oven at 105 °C and calcined at 500 °C for 2 h (heating rate 2 °C min−1) in air, followed by sieving to obtain particle sizes between 0.56 and 1.18 mm. The synthesized catalysts are denoted as xNi – yFe, where x and y represent calculated molar contents of metals per 100 g of the resulted catalyst.Two commercial catalysts from different manufacturers (6-holes monoliths from Xian Sunward Aeromat Co., China and 19 × 19 × 10 mm rings from Pingxiang Hualian Chemical Ceramic Co., China) were crushed and sieved to obtain 0.56–1.18 mm particles, and used as the reference materials. Fig. 1 shows the experimental setup for catalytic naphthalene reforming. A fixed bed reactor with a quartz frit (50–90 μm openings) was used. In a typical run, 0.5 mL of a catalyst was loaded into the reactor and heated at 15 °C min−1 in reducing atmosphere containing 20 vol% H2 – 80 vol% N2 (total gas flow 50 mL-STP min−1) to the reforming temperature of 790, 850 or 900 °C. Once the temperature was reached, the gas flow was maintained for 30 min to reduce the catalyst and then the flow was switched to 200 mL-STP min−1 (space velocity 24,000 h−1) of a model syngas containing 0.14 vol% naphthalene, 10 vol% H2, 26 vol% H2O, 0 or 300 ppmv HCl, 0 or 50 ppmv H2S and N2 (balance). In this study, 0.14 vol% naphthalene was used as it is within the range of typical concentrations of naphthalene produced from biomass/MSW gasification [43,44]. Naphthalene vapors were generated by purging N2 gas through an evaporator containing heated naphthalene. H2, N2 and H2S were supplied from gas cylinders using mass flow controllers. The steam and HCl vapor were generated from aqueous solution of HCl injected by a syringe pump into an evaporator. During experiment, all the gas lines were heated and kept above 150 °C to avoid vapor condensation. After reforming reactor, the model gas passed through two water traps to capture HCl and one silica gel trap to remove moisture, and then was collected in Tedlar gas bags for analysis (collection time 5 min). Concentrations of CO, CO2 and C1-C5 hydrocarbons were measured by a calibrated gas chromatograph (Agilent 7890B, USA) equipped with one flame ionization and two thermal conductivity detectors. Steam reforming of naphthalene is analogous to steam reforming of other hydrocarbons [10,45]: (2) C 10 H 8 + 10 H 2 O = 10 CO + 14 H 2 Due to WGS activity, CO is partially converted to CO2 over Ni catalysts [1,28,45]: (3) CO + H 2 O = CO 2 + H 2 Naphthalene conversion can be calculated by the following equation: (4) Naphthalene conversion (%) = ( n C O 2 + n CO ) n naphthalene × 10 × 100 % where: n CO and n C O 2 are the molar concentrations of CO and CO2 generated during steam reforming of naphthalene, mol min−1, and n naphthalene is the molar concentration of naphthalene in the feed, mol min−1. All experiments were triplicated and the results are presented as averages of three experimental runs.The catalysts were characterized by X-ray diffraction analysis with Cu-Kα radiation source (XRD, Bruker AXS D8 Advance), X-ray photoelectron spectroscopy with a dual anode monochromatic Kα excitation source (XPS, Kratos Axis Supra), X-ray fluorescence spectroscopy (XRF, PANalytical Axios mAx), transmission electron microscopy at 120 kV (TEM, JEOL JEM-2010) and N2 adsorption at −196 °C (Quantachrome Autosorb-1 Analyzer). Binding energies of elements in XPS spectra were corrected against an adventitious carbon C 1s core level at 284.8 eV. The processing of XPS peaks was carried out in the CASA XPS software. TEM images were used to measure the size of Ni nanoparticles in spent catalysts. The diameters were calculated using ImageJ software by analysing 150–200 Ni nanoparticles in each sample and assuming that nanoparticles have ideal spherical shape. Specific surface area of catalysts was calculated from N2 adsorption isotherms using BET model. Total pore volumes were calculated from N2 adsorption volume at P/P0 = 0.96. Temperature programmed reduction (TPR) was performed in a 5% H2/N2 gas mixture at 30 mL min−1 flow rate with a temperature ramp of 10 °C min−1 up to 900 °C. Carbon content in the catalysts was measured by CHNS elemental analyser (Vario EL Cube).The properties of pristine Ni, Fe, Ni-Fe and commercial catalysts are presented in Table 1 . Ni and Fe contents were determined from XRF analysis and used to calculate molar quantities of Fe and Ni. The molar Ni and Fe loadings per 100 g of catalysts were close to the corresponding theoretical values of x and y in xNi-yFe samples. The amount of Ni in 0Fe-0.4Ni and two commercial catalysts loaded into the reactor for steam reforming of naphthalene was nearly the same (approx. 1.90 mmol 0.5 mL−1 catalyst) due to the differences in bulk density, allowing comparison between the activities of Ni in the synthesized and commercial catalysts. The synthesized catalysts had higher BET specific surface areas and total pore volumes compared to the commercial catalysts. According to the high N2 adsorption volumes at relative pressures P/P0 > 0.1 and hysteresis loops between adsorption and desorption branches of isotherms (Fig. S1), the synthesized materials were mesoporous. Among them, 0Fe-0.4Ni had the largest porosity (i.e. 213 m2 g−1 and 0.31 mL g−1, respectively) (Table 1), which is one of the reasons for its better catalytic performance stated in the following study. The specific surface areas and total pore volumes of the synthesized catalysts decreased with increasing Ni + Fe contents, which can be attributed to the impregnation of the porous alumina with loaded metal species. X-ray diffraction (XRD) patterns of the synthesized catalysts in Fig. 2 consist of broad peaks with no distinct XRD peaks and also show no sharp XRD peaks indicating that alumina, nickel and iron oxides have non-crystalline and/or nanosized structures, so that alumina provides surface area for better dispersion of catalyst. On the other hand, in commercial catalysts, the XRD peaks of NiO (commercial 1 and 2) and α-Al2O3 (commercial 1) can be clearly identified. Fig. 3 depicts the Ni 2p and Fe 2p core level spectra of the four pristine synthesized catalysts. Ni 2p spectra of 0Fe-0.4Ni, 0.1Fe-0.4Ni, 0.2Fe-0.4Ni and 0.5Fe-0Ni contain shake-up satellite peaks with binding energy (BE) of approx. 862 eV and peaks with BE of approx. 856 eV. In Ni-based catalysts, the binding energy of Ni2+ typically increases with the strength of NiO–Al2O3 interactions from approx. 854 eV for unsupported and weakly bound to the support NiO to approx. 856 eV for strongly bound to the support NiO [46–49]. At high NiO–Al2O3 interaction levels, the binding energy of Ni2+ in NiO of alumina supported catalysts becomes similar to that in NiAl2O4 spinel (855.8 eV) [48,50]. Due to the shift in binding energy, it is uncertain whether the Ni2+ state in the catalysts NiO or NiAl2O4 solely based on XPS spectra. The similar binding energies of Ni2+ in all synthesized catalysts suggest that independently on Ni + Fe loading, strong interactions between NiO and alumina were maintained in the catalysts and there was no formation of new compounds with Fe species. The same can be concluded from Fe 2p core level spectra. Binding energies of Fe 2p for all catalysts were similar regardless of the presence of NiO (i.e. 711.0 eV for 0.1Fe-0.4Ni, 711.4 eV for 0.2Fe-0.4Ni and 711.0 eV for 0.5Fe-0Ni) and corresponded to Fe3+ in Fe2O3 [50].TPR profiles provide useful information regarding reducibility of Ni and Fe oxides in the synthesized catalysts (Fig. 4 ). The reduction of catalysts occurred in a wide temperature range between 300 and 800 °C. The catalysts 0.2Fe-0.4Ni and 0.5Fe-0Ni contained a distinct reduction peak at 475 °C, which corresponds to the reduction of Fe2O3 [51], i.e. the main Fe species in catalysts according to XPS. According to the TPR profile of 0Fe-0.4Ni, most of nickel was reduced at 500–700 °C with the maximum reduction temperature at 590 °C, which can assigned to highly dispersed NiO having strong metal-support interactions [48,52]. Small shoulder peaks at 350, 425 and 770 °C were also observed. The reduction at 300–400 °C is typically attributed to bulk and/or unsupported NiO, while the reduction >700 °C could be attributed to nickel aluminates formed due to sintering of NiO with Al2O3 [48,53,54], indicating that minor quantities of these species could be also present in the synthesized catalysts. According to the similar positions of H2 consumption peaks in the catalysts, the reducibility of Ni species was not influenced by the addition of Fe2O3 and vice versa.XPS and TPR data of the commercial catalysts are shown in Fig. S2. As it was reported elsewhere [37], in both catalysts, Ni2+ was in the form of NiO. However, in Commercial 2, NiO was more strongly bonded to the support compared to Commercial 1. Considering the similar Ni loading per 0.5 mL catalyst bed for Commercial 1, Commercial 2 and 0Fe-0.4Ni, this allows investigation of the effects of H2S and HCl on the activity of catalysts with different strength of NiO–Al2O3 interactions determined by the differences in porosity, crystalline structure, NiO dispersion etc. The addition of Fe to Ni-based catalyst provides further insight about the influence of H2S and HCl on the activity of catalysts with different metal composition. Fig. 5 depicts naphthalene conversion using the six catalysts in the presence and absence of H2S and HCl at 850 °C. CO and CO2 were the only reaction products. No formation of C1–C5 hydrocarbons was observed during the process. Naphthalene conversion over catalysts fluctuated during the first 30 min of experiment and was stabilized thereafter. In all catalysts containing Ni, the reforming activity was lower in the presence of H2S and HCl due to the poisoning effect (data in Fig. 5a against Fig. 5b). Furthermore, naphthalene conversion by 0.5Fe-0Ni was approx. 12% in the absence of H2S and HCl, and decreased to approx. 8% in the presence of H2S and HCl, suggesting the poisoning of Fe. Regardless of the presence of H2S and HCl, naphthalene conversion was stable during 5 h tests. The synthesized 0Fe-0.4Ni showed comparable conversion efficiency with commercial catalysts, which was likely due to the same amount of Ni loading per 0.5 mL bed in the three catalysts. These results suggest that there was similar poisoning effect on the naphthalene reforming activity for the catalysts with different strength of NiO–Al2O3 interactions. Reforming activity of 0.1Fe-0.4Ni was similar to 0Fe-0.4Ni, while the higher content of Fe in 0.2Fe-0.4Ni resulted in the decreased naphthalene conversion. This could be attributed to the decreased porosity and specific BET surface area with the higher Fe content due to the occupation of surface sites (Table 1). Unlike Ni-based catalysts, 0.5Fe-0Ni merely achieved approx. 8% of naphthalene conversion, indicating that Ni is more active catalyst for naphthalene reforming compared to Fe. The lower catalytic toluene reforming activity due to Fe addition to Ni/zeolite catalyst was reported by Ahmed et al. [38], who found the depletion in basicity strength of this Fe-Ni/zeolite catalyst leading to suppressed steam reforming.Elemental CHN analysis (Table S1) of the pristine and spent catalysts suggests that there was no significant increase in the amount of carbon after the reforming, indicating no coking happened in the presence of H2S and HCl. This can be attributed to the relatively high content of steam in the model gas (i.e. 26 vol%) that could assist in carbon gasification. Fig. 6 shows the TEM image of fresh 0.2Fe-0.4Ni after preheating in 20 vol% H2–80 vol% N2 and spent 0.2Fe-0.4Ni after 5 h of reaction at 850 °C in the presence of H2S and HCl. The comparison of the morphologies of fresh and spent catalyst indicates the absence of carbon deposition during naphthalene reforming, which is consistent with CHN analysis. After reforming, Ni was present in the form of discreet spherical nanoparticles. This is attributed to the sintering of Ni during the process [37]. On the contrary, Fe was evenly distributed over the catalyst surface (Fig. 6a). Fig. S3 shows that in other Fe-containing catalysts, Fe also remains in the dispersed state. The coverage of entire surface of the spent 0.2Fe-0.4Ni catalyst by S and Cl indicates that the chemisorption of these species occurred both on the Ni and Fe sites (Fig. 6b) [29,55,56], which explains the poisoning effect of HCl and H2S both on the reforming activity of Ni and Fe. Fig. 7 shows the XRD patterns of the spent catalysts after naphthalene reforming at 850 °C in the presence of 50 ppmv H2S and 300 ppmv HCl. In the spent samples containing Ni element, the formation of metallic Ni phase was observed as suggested by the labelled NiO XRD peaks. As there were no XRD peaks of NiO in the fresh catalysts, these results indicate that upon reduction and reforming, Ni undergoes sintering into larger size crystalline nanoparticles, which is consistent with TEM data in Fig. 6a. Unlike NiO, the formation of crystalline FeO in 0.5Fe-0Ni was not observed as suggested by the absence of corresponding XRD peaks in this sample and even distribution of Fe in the TEM images of spent catalysts (Figs. 6a and S3). According to TPR data (Fig. 4), the reforming temperature was sufficient for the reduction of Fe2O3 to FeO. Therefore, it is likely that in the spent catalysts iron was in metallic non-crystalline state. These observations are consistent with scanning TEM data in Fig. 6, showing the differences in Fe morphology compared to Ni.There was no change in the position of NiO XRD peaks in 0Fe-0.4Ni, 0.1Fe-0.4Ni and 0.2Fe-0.4Ni with the addition of Fe (Fig. 7), which would have been observed with the formation of Ni-Fe alloys [27,57], indicating that there was no alloying between Ni and Fe in the spent catalysts. The amount of chemisorbed sulfur and chlorine species during reforming was typically low which explains the absence of XRD peaks corresponding to metal chlorides and sulphides in all catalysts.The reaction temperature is one of the most important operating variables for steam reforming. 0Fe-0.4Ni, 0.1Fe-0.4Ni and Commercial 1 were further selected to investigate the effect of temperature on catalytic activity. Fig. 8 shows naphthalene conversion at 790, 850 and 900 °C in the presence of H2S and HCl. Except for the decrease in conversion within the initial 30 min at 790 °C, the activity of catalysts remained constant thereafter indicating that it is possible to maintain stable conversion efficiency of naphthalene in the presence of H2S and HCl within the studied period of time at each temperature. The catalytic activities of the three catalysts were similar in the presence of H2S and HCl at each temperature regardless of the strength of NiO–Al2O3 interactions (0Fe-0.4Ni vs. Commercial 1) and the addition of Fe (0Fe-0.4Ni vs. 0.1Fe-0.4Ni). The reforming activities of all catalysts were greatly influenced by the reforming temperature, increasing from approx. 40% to approx. 100% efficiencies with the increase in reaction temperature from 790 to 900 °C, respectively. These results can be attributed to the increased reaction rate of naphthalene with steam and the decreased H2S poisoning effect at higher temperature. It has been well known that H2S poisoning is caused by sulfur adsorbed on the nickel surface in the catalyst according to reaction (1). This reaction is reversible. With the increasing temperature desorption of H2S increases releasing surface active sites for the steam reforming reaction [58].To determine the respective and relative roles of H2S and HCl in the catalyst poisoning effect observed in Figs. 5 and 8, the naphthalene reforming at four different conditions was compared: (1) 50 ppmv H2S and 300 ppmv HCl, (2) 0 ppmv H2S and 300 ppmv HCl, (3) 50 ppmv H2S and 0 ppmv HCl, and (4) 0 ppmv H2S and 0 ppmv HCl. The experiments were carried out at 790 °C, as the poisoning was the most prominent at this temperature. According to Fig. 9 , in the absence of H2S, naphthalene conversion was approx. 100% both at 0 and 300 ppmv HCl. In the presence of 50 ppmv of H2S, naphthalene conversion decreased to approx. 40% both at 0 and 300 ppmv HCl. These results suggest that the poisoning of naphthalene reforming was caused by H2S, while HCl had negligible effect on this reaction. Furthermore, since the naphthalene conversion in the presence of H2S was similar at 0 and 300 ppmv HCl, it can be concluded that H2S and HCl had no synergistic effect on the poisoning of reforming activity when both impurities were present in the stream.Based on the obtained data, during the reforming of naphthalene from gas streams containing both H2S and HCl, the poisoning of catalysts is mainly caused by H2S and can be attributed to the decreased accessibility of surface active sites for hydrocarbons due to H2S chemisorption [29]. The poisoning effect on naphthalene reforming activity was similar for the catalysts with different strength of NiO–Al2O3 interactions and Ni + Fe contents. Increasing reaction temperature could effectively improve catalytic activity of Ni and Ni-Fe based catalysts in the presence of H2S and HCl leading to approx. 100% naphthalene conversion (Fig. 8). Fig. 10 shows the ratios between CO and CO2 in the gas during naphthalene reforming over the catalysts at 850 °C in the presence of H2S and HCl. Steam reforming of hydrocarbons is typically presented as the combination of two reactions, namely, partial oxidation of hydrocarbon by steam into CO and H2 (reaction 2) followed by WGS reaction (3) [1,8,10,28,45]. Consequently, the lower CO/CO2 ratio is probably attributed to the higher conversion of CO into CO2 over catalysts via WGS reaction (3). Dashed black line shows the CO/CO2 ratio at thermodynamic equilibrium (CO/CO2 = 0.52 at 850 °C). For all catalysts, the CO/CO2 ratios were higher than 0.52 indicating that thermodynamic equilibrium was not attainable. This is due to the lower space velocity and longer residence time required for the equilibration of WGS reaction over the catalysts [37]. There were significant differences in the kinetics of WGS reaction as suggested by the different CO/CO2 ratios for the catalysts. The CO/CO2 ratios of synthesized Ni and Ni-Fe catalysts increased from 0.9 to 1.0 to 1.2–1.5 during the 5 h tests, depending on the sample. These changes were much lower compared to Commercial 1 and Commercial 2 catalysts (i.e. from 0.7 to 3.9 for Commercial 1 and from 1.0 to 2.2 for Commercial 2), suggesting higher stability of WGS activity of the synthesized catalysts. Based on the similar CO/CO2 ratios for 0Fe-0.4Ni, 0.1Fe-0.4Ni and 0.2Fe-0.4Ni, the addition of Fe to catalysts did not alter the WGS activity of catalysts. Furthermore, the lower CO/CO2 ratios over the synthesized Ni containing catalysts compared to 0.5Fe-0Ni indicate that the WGS activity over Ni was higher compared to Fe during naphthalene steam reforming.Although 0Fe-0.4Ni, Commercial 1 and Commercial 2 had similar NiO loading per catalyst bed volume, the strength of interactions between NiO and alumina support was different in the catalysts, eventually, leading to the different Ni-support interactions in the reduced catalysts. Specifically, the strength of interactions increased from Commercial 1 to Commercial 2 and, finally, to 0Fe-0.4Ni which is consistent with the increase in WGS activity in the same order (Fig. 10). One reason behind the observed phenomenon is the mechanism of WGS reaction over Ni based Al2O3 catalysts. By combining density functional theory and microkinetic modelling, it was demonstrated that Ni-support interface provides catalytically active sites for WGS reaction, serving as a storage for oxygenated Ni2+ species [59]. Therefore, the decrease in the strength of metal-support interactions in catalysts can result in the observed loss of WGS activity. In comparison, for the steam and dry reforming reactions of methane, the importance of metal-support interactions was found to be less important as the active sites for these reactions seem to be different. [59,60] Assuming that the mechanisms for reforming of hydrocarbons are similar [45], this could explain the negligible differences in naphthalene conversion over 0Fe-0.4Ni, Commercial 1 and Commercial 2 (as shown in Fig. 5).For WGS reaction, oxygenated Ni2+ sites are required [59], while higher reforming temperatures favor the reduction of NiO to metallic Ni. TPR profiles of synthesized and commercial catalysts (Figs. 4 and S2) show that the reduction temperature of Ni2+ increased from Commercial 1 to Commercial 2 followed by the synthesized catalysts indicating that the synthesized catalysts can have the higher density of oxygenated Ni2+ sites at the reforming conditions due to the stabilization of NiO by the support [48,52,61,62]. To confirm that, thermodynamic calculations using HSC Chemistry 9 software were carried out to calculate the content of oxygenated Ni2+ in catalysts in the absence and presence of NiO-support interactions. For the simplicity of calculations, it was assumed that in the absence of interactions with the support, nickel can only be oxidized into NiO. In the presence of strong metal-support interactions, Ni can form stable oxidized species at the NiO-support interface. γ- and α-Al2O3 were selected as the representative support materials, which allow for the formation of NiAl2O4 spinel [62] .Under steam reforming conditions, H2O acts as an oxidant and the oxidation of Ni can be described by the following reactions: (4) Ni + H 2 O ⇌ NiO + H 2 (5) Ni + γ - , α - Al 2 O 3 + H 2 O ⇌ NiAl 2 O 4 + H 2 Fig. 11 a shows the standard Gibbs reaction energies (ΔG°) for the oxidation of Ni as the function of reforming temperature. It can be seen that ΔG° increases with temperature suggesting that higher temperature causes the formation of metallic Ni. However, at the same temperature, ΔG° is lower when γ- and α‐Al2O3 participate in the reaction, indicating that in the catalysts with strong NiO-support interactions, there is a higher content of oxygenated Ni2+. From the corresponding thermodynamic equilibrium constants, the content of Ni2+ can be calculated at the experimental conditions. According to Fig. 11b, in the absence of NiO-support interactions, the content of Ni2+ slightly increases with temperature and is 1.3%, 1.4% and 1.5% at 790, 850 and 900 °C, respectively. In the presence of NiO-support interactions, the content of Ni2+ is much higher at each reforming temperature (Fig. 11b). Notably, the γ- Al2O3 favors the stabilization of Ni2+ to a larger extent compared to α‐Al2O3 highlighting the importance of alumina material for the design of catalysts with tailored WGS activity. The provided thermodynamic calculations confirm that at the reforming temperatures, NiO-Al2O3 interactions can indeed stabilize nickel in the oxidized form due to the participation of support in the reaction, which could be in turn responsible for the higher WGS activity on the synthesized catalysts. Since, the XRD patterns of the spent catalysts contain only metallic Ni phase (Fig. 2), it is likely that the oxygenated Ni2+ species are mainly present at the NiO-Al2O3 interface.Previously, it was proposed that the exposure of catalysts to high concentration of HCl (2000 ppmv) during steam reforming of naphthalene causes the chemisorption of HCl on Ni followed by the sintering of Ni species into larger size nanoparticles. This process is irreversible and leads to a permanent loss of WGS activity [37]. The poisoning of WGS activity of catalysts by low concentrations of H2S and HCl (i.e. 50 and 300 ppmv, respectively) has not been investigated. Fig. 12 presents the CO/CO2 ratios for 0Fe-0.4Ni, 0.1 Fe-0.4Ni and Commercial 1 at 790, 850 and 900 °C. Among the tested catalysts, Commercial 1 showed the lowest WGS activity at each temperature. With the increase in temperature, CO/CO2 ratios for Commercial 1 catalyst decreased, indicating that WGS activity of this catalyst could be improved by increasing the reforming temperature. Since the content of oxygenated Ni2+ species is relatively high at all reforming temperatures, this could be attributed to the faster reaction rate that allows to approach closer to the thermodynamic equilibrium and/or enhanced desorption of S- and Cl-species at higher temperature. Nevertheless, the CO/CO2 ratios for Commercial 1 remained high compared to those corresponding to thermodynamic equilibrium. The CO/CO2 ratios for Ni and Ni-Fe catalysts were lower than that of Commercial 1 and closer to thermodynamic equilibrium at all temperatures, indicating higher WGS activity.Since the poisoning of Commercial 1 was more pronounced at lower temperature, the individual and combined effects of H2S and HCl on WGS activities of two representative catalysts, namely, 0Fe-0.4Ni and Commercial 1, were compared at 790 °C. Fig. 13 a presents the CO/CO2 ratios at four experimental conditions: (1) 50 ppmv H2S and 300 ppmv HCl, (2) 0 ppmv H2S and 300 ppmv HCl, (3) 50 ppmv H2S and 0 ppmv HCl, and (4) 0 ppmv H2S and 0 ppmv HCl. In the context with respect to WGS reaction, the presence of H2S and HCl had negligible effect on the poisoning of 0Fe-0.4Ni indicating high stability of the WGS activity to the action of both impurities. The deterioration of WGS activity of Commercial 1 was observed even in the absence of H2S and HCl. This could be attributed to the lower strength of NiO–Al2O3 interactions in this catalyst compared to 0Fe-0.4Ni. As shown in Fig. 13b and c, the sizes of Ni nanoparticles were larger in the spent Commercial 1 compared to 0Fe-0.4Ni after using condition 4, which could result in the lower WGS activity [59].For Commercial 1, CO/CO2 ratios increased both under condition 2 (HCl only) and condition 3 (H2S only), indicating that both impurities contributed to the poisoning of WGS activity (Fig. 13a). The poisoning of WGS activity in the presence of H2S was faster compared to HCl as demonstrated by the rapid increase in CO/CO2 ratio within the first 60 min of reaction (i.e. condition 2 against condition 3). The poisoning of catalyst was more pronounced in the presence of both H2S and HCl (condition 1), indicating the detrimental synergistic effect of impurities. According to Fig. 13b and c, at low concentrations of H2S and HCl, there was no change in the sizes of Ni nanoparticles of 0Fe-0.4Ni and Commercial 1. These data suggest that unlike at 2000 ppmv HCl in literature [37], low concentrations of H2S and HCl are unable to enhance Ni sintering, and the detrimental effect on WGS activity of Commercial 1 was most likely associated with the poisoning of catalyst surface solely via chemisorption. This could explain the increase in WGS activity of Commercial 1 with the increase in the reaction temperature from 790 to 900 °C in Fig. 12 as higher temperature typically decreases chemisorption. If this hypothesis is correct and chemisorption is the main reason for the catalyst poisoning, after desorption of S and Cl species, the WGS activity of catalyst can be restored. On the other hand, if sintering causes the poisoning as observed for high concentrations of HCl (i.e. 2000 ppmv), the loss of WGS activity would be irreversible [37]. To test the hypothesis, the spent Commercial 1 and 0Fe-0.4Ni after 5 h of naphthalene reforming at 790 °C in the presence of 50 ppmv H2S and 300 ppmv HCl (denoted as Exp. 1 in Fig. 14 ) were respectively used for the subsequent 5 h naphthalene reforming at 790 °C in the absence of H2S and HCl (denoted as Exp. 2 in Fig. 14). According to Fig. 14a, while at the end of Exp. 1 naphthalene conversions by 0Fe-0.4Ni and Commercial 1 were both only approx. 40%, they were restored to approx. 80% by commercial 1 and approx. 85% by 0Fe-0.4Ni during Exp. 2 when H2S and HCl were removed from the gas stream. This improvement can be attributed mainly to the desorption of H2S, that has detrimental effect on the steam reforming of hydrocarbons as it was demonstrated in Section 3.2. Despite the two times increase in the catalytic activity, naphthalene conversion during Exp. 2 was still lower compared to that of the fresh catalysts utilized in the absence of H2S and HCl (i.e. approx. 100%). These data suggest that desorption of H2S was incomplete. According to Fig. 14b, for 0Fe-0.1Ni, CO/CO2 ratio during Exp.2 was similar with that during Exp.1 (i.e. 1.1) indicating that the presence of H2S and HCl had negligible effect on WGS activity of 0Fe-0.4Ni. This observation is consistent with Fig. 13a showing high stability of the WGS activity to the action of both impurities. However, after Exp. 1, CO/CO2 ratio by Commercial 1 was 4.4 and the CO/CO2 ratio drastically decreased to 2.5 during the first 30 min of Exp. 2 and remained stable for 4.5 h. This value (i.e. 2.5) is comparable with the CO/CO2 value obtained for the fresh Commercial 1 utilized in the absence of H2S and HCl (i.e. 2.8), indicating that at low concentrations of impurities, the poisoning effect on the WGS catalytic activity was reversible, thus, confirming the hypothesis.The structure of catalyst played an essential role in WGS reaction but not in reforming reaction (which was strongly influenced by temperature). The stronger NiO–Al2O3 interactions provided beneficial effect to catalytic activity which could be probably attributed to the formation of larger content of oxygenated Ni2+ species that serve as active sites for WGS reaction [59]. The poisoning effect of HCl and H2S on WGS was more pronounced in a catalyst with weakly bonded NiO to the Al2O3 support. At low H2S and HCl concentrations, the poisoning of WGS activity proceeds via chemisorption of S and Cl species and the loss of catalytic activity is reversible when H2S and HCl are removed from the gas stream.The effects of H2S (50 ppmv) and HCl (300 ppmv) on catalytic steam reforming of naphthalene were investigated using Ni, Ni-Fe and Fe catalysts supported on alumina at 790, 850 and 900 °C. Ni had higher reforming and WGS activities compared to Fe and the activities of Ni were not significantly influenced by the addition of Fe. H2S poisoned naphthalene reforming activity of the catalysts, while the addition of 300 ppmv to gas stream had no effect on this reaction at 0 and 50 ppmv H2S. On the contrary, both HCl and H2S could poison WGS activity of the catalysts and the poisoning effect was more pronounced when both impurities were present in the gas stream. The poisoning by H2S could be only partially restored by removing H2S from the gas stream indicating the strong chemisorption of H2S on Ni. However, H2S poisoning effect could be prevented by carrying out reforming of naphthalene at higher temperatures. Specifically, the increase in temperature from 790 °C to 900 °C increased naphthalene conversion from approx. 40% to approx. 100%. The poisoning of WGS activity during naphthalene reforming was significantly influenced to the structure of catalyst. The stronger NiO–Al2O3 interactions provided beneficial effect minimizing the loss of WGS activity. This beneficial effect could be attributed to the formation NiO-support interfaces upon reaction serving as active sites for WGS reaction. At these concentrations of H2S and HCl (i.e. 50 and 300 ppmv, respectively), the loss of WGS activity was reversible when H2S and HCl were removed from the gas stream.This research is supported by the National Research Foundation, Prime Minister’s Office, Singapore and the National Environment Agency, Ministry of the Environment and Water Resources, Singapore, under the Waste–to–Energy Competitive Research Programme (WTE CRP 1501 105). The authors also acknowledge the management of Nanyang Environment and Water Research Institute and Economic Development Board, Singapore for the support.Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuel.2018.12.119.The following are the Supplementary data to this article: Supplementary data 1

H2S and HCl are common impurities in raw syngas produced during gasification of biomass and municipal solid waste. The purpose of this study was to investigate the poisoning effect of H2S and HCl on synthesized and commercial catalysts during steam reforming of naphthalene. Four synthesized catalysts with different loadings of Ni and Fe on alumina support and two commercial catalysts were selected and evaluated in a fixed bed reactor at 790, 850 and 900 °C. The obtained results revealed that reforming and water-gas shift (WGS) activities of catalysts did not benefit from the Fe addition. The activities were influenced differently by H2S and HCl indicating that the reactions were catalyzed by different active sites on the nickel surface. In the presence of H2S and HCl, the poisoning of naphthalene reforming activity was caused by H2S and was not affected by HCl when both compounds were present in the gas. H2S chemisorbs on nickel surface forming NiS and decreasing the accessibility of active sites to hydrocarbons. The poisoning effect was only partially reversible. On the contrary, the poisoning of WGS activity could be caused by both H2S and HCl, and the activity could be completely restored when H2S and HCl were removed from the gas. Unlike naphthalene reforming activity, which was comparable for catalysts with similar Ni loadings, WGS activity depended on the catalyst structure and was less susceptible to poisoning by H2S and HCl in case of the catalyst with strong NiO-support interactions.

Data will be made available on request.Supplying most of the energy consumed by our society from fossil fuels is at risk of critically affecting global warming due to net CO2 emissions into the atmosphere [1]. An energy matrix transformation from the currently used net emission sources to CO2 neutrality is essential to achieve energy sustainability. An increased production of energy from renewable energy sources such as wind-, solar-, hydro-, or geopower would here be highly desired. However, several of these energy resources are highly intermittent, geographically spread, or seasonally dependent. Achieving an efficient and large-scale compatible way of storing the produced energy would be highly beneficial. In this context, ammonia emerges as a promising candidate both as a fertilizing chemical and as a potential energy vector that benefits from its high hydrogen content and easy liquefaction. Currently, ammonia is industrially produced via the Haber-Bosch process, which demands a large amount of energy and releases CO2 into the atmosphere, thus aggravating the greenhouse effect. One strategy to circumvent this is to produce ammonia in an eco-friendly process, which could be an electrochemical synthesis of ammonia from nitrogen gas using sustainably produced electricity. A severe bottleneck of electrochemical ammonia synthesis, is the low ammonia production rates of about microgram per hour per square centimeter-level, often lower than 10% Faradaic efficiencies (FEs), and stability issues. Therefore, the development of suitable materials plays an important role in mitigating such issues and achieving industrial application [2]. In this context, high-entropy alloys (HEAs) emerge as a new class of catalysts that provide unprecedented compositional diversity that hold the promise to tune reaction pathways and, thus, selectivity and rates, alongside entropic stabilization of the material.The concept of multi-component alloys with entropy stabilization came out around 2004 when two independent research groups showed that multiple-element materials containing at least five different species could be formed into a homogeneous phase [3,4]. The thermodynamically and kinetically stabilized structures of HEAs provide high fracture resistance, ductility, and physicochemical stability, thus enabling employment in harsh environments [5–7]. Concerning their application in catalysis, these alloys form a promising new material class and is a rapidly growing research field [5,6]. As an advantage, the multi-component form of HEAs can provide several active sites on a catalytic surface and structural stability. The complexity of the catalytic surface enables the possibility of breaking the symmetry rules imposed by the scaling relationships [8], which in principle opens the possibility of finding highly active catalysts for different reactions.It is notoriously challenging to reduce nitrogen to form ammonia electrochemically due to the scaling relationships and simultaneous ability to reduce hydrogen in a protonic system, and thus a competition between the nitrogen reduction reaction (NRR) and the hydrogen evolution reaction (HER). Although N2 is the most abundant molecule in the atmosphere, its triple bond and the lack of dipole moments make it a highly inert specie and, hence, very hard to catalyze due to the lack of nitrogen fixation on catalytic surfaces [9]. Several strategies have been reported in the literature to enhance NRR, where they all more or less are related to nitrogen fixating surfaces with either proton deficiency or electron starving (and thus low rate) [9–14]. Especially interesting for this work are the results reported by D. Zhang et al. [15]. They were one of the first groups reporting that HEAs could be used to reduce nitrogen, where 38.5% FE was achieved using HEA RuFeCoNiCu nanoparticles with a small size of 16 nm and signifies the promises of the approach, although the remaining challenge is to perform the reaction with NRR selectivity versus HER also at high rates.The present work focuses on two important aspects of the application of HEAs to NRR: i) a rational strategy to screen over an ample search space of quinary HEAs formed with Mo-Cr-Mn-Fe-Co-Ni-Cu-Zn working as novel catalysts for NRR, without the inclusion of Ru or other platinium group metals (PGMs), and ii) identifying relationships between HEAs intrinsic properties and their catalytic activity. The search for alternative catalysts for NRR in the wide range of compositional space found in the quinary HEAs is performed by employing the framework of the density functional theory (DFT), machine learning techniques and a probabilistic approach developed by T. A. Batchelor et al. [16]. Similar strategies were also applied, successfully, by T. A. Batchelor et al. [16], J. K. Pedersen et al. [17] and W. A. Saidi et al. [18]. They used DFT to train machines over hundreds of adsorption energies on HEAs microstates and, further, used these machines to screen for selective and active catalysts for oxygen, carbon dioxide, carbon monoxide reduction reactions and also ammonia oxidation, respectively. The estimated catalytic activities are, further correlated with intrinsic properties of the HEAs, like the average valence electron concentration and their electronegativity. This might help to understand the main properties that dictate the electrochemical transformations and it is also a simplified path to selecting promising catalysts. Finally, since a clear reaction pathway on the surface of a HEA is not possible due to their inherent randomness, a statistical analysis of the reaction pathway will be performed for the selected HEAs.The associative pathways (distal/alternating and enzymatic) are the most favorable ones when electrochemical NRR is in focus [19]. This is due to the high activation barriers to dissociating N2* into 2 N*, that, for instance, is of the order of 1.77 eV for the case of Ru(0001) [20]. The other two likely options are the distal/alternating pathways and the enzymatic pathway [21]. Pedersen et al. [22] recently showed that species on threefold hollow sites of HEAs could partially circumvent scaling relations with the adsorption of species on top sites due to the different coordination of threefold sites compared to on top sites (also confirmed in this work). Moreover, Singh et al. [19] and Montoya et al. [23] showed that the two potential limiting steps of the NRR reaction on transition metal surfaces are the hydrogenation of N2* forming NNH* and the desorption of NH* forming NH2 * . In the distal pathway, the N2* adsorbs on the top position while the NH* adsorbs on the threefold hollow site. Hence, the scaling relationship between these two steps can be circumvented due to the randomness of the HEA surfaces. That allows us to seek highly active HEAs for NRR that deliver strong N2 * bond interaction in the distal position (adsorbing on the top site), but still with lower desorption of the NH* intermediate. Therefore, we will focus on identifying HEAs that optimize the catalytic activity and selectivity towards NRR following the distal/alternating pathway. Moreover, only for the selected catalysts, we explicitly compute, with the DFT framework, the potential limiting steps, N2(gas)+*→N2*, N2*+H++e-→NNH* and NH*+H++e-→NH2* for 100 microstates of the referent HEA and show how the reaction pathways can be depicted based on the statistical analysis.Tuning to the approach to characterize the reaction steps, Singh et al. [19] and Montoya et al. [23] have shown that the N* is a descriptor of the NRR catalytic activity on transitions metals where, for the case of close-packed structures, Fe is placed on the top of the volcano curve [19]. Therefore, we can employ this descriptor to optimize HEAs elemental concentrations that maximize the local sites with similar adsorption energies as in the case of Fe, for instance. That must lead to optimal cases with reasonable potential limiting steps – at least similar to the value displayed by Fe. Moreover, it allows us to build a much more efficient screening strategy since the number of parameters is reduced. That in itself should be enough to deliver promising HEA catalysts for NRR. However, due to the break of scaling relationships, the top of the volcano curve is not completely known and, hence, including the N2* also in the optimization process might lead to cases with even higher activity and that can also mitigate the N2 fixation issue. It is also important to highlight that the activations barrier of the NRR following the distal pathway tends to be very similar to the computed thermodynamical barriers. E. Tayyebi et al. [20] and A. B. Höskuldsson et al. [24] have shown that including activation barriers in the calculation of N2 reduction pathways leads to the same electrochemical paths predicted with thermochemistry, for Ru(0001) and W(110). Moreover, the transition states computed for small molecules like N2, are often resembling the final state of the reaction [25]. Therefore, confirming that thermodynamical steps, in this case, can be used as an effective parameter to estimate the rates of the reaction.The approach to model the reactions and to screen over the HEA’s elemental concentration pool is depicted in Fig. 1 and based on the following steps: 1. Quinary HEAs formed with the elements Mo-Cr-Mn-Fe-Co-Ni-Cu-Zn are randomly created. DFT calculations are performed over 1200 microstructures formed with the above-mentioned elements. For each microstructure, the adsorption energies of N2 * and N * (descriptors of the reaction) are computed and stored in a database (DFT details are in the section “ DFT Calculations”). 2. Representation models of the microstructures are created and used to build neural network models that are trained on the adsorption energies from the DFT calculations. These models can compute adsorption energies almost instantaneously and, overcome the time-consuming task of performing thousands of adsorption energies with the DFT approach (details are in the section “ Deep neural networks” ). 3. Using the deep neural network models, we calculate the adsorption energy of N * and N2 * (descriptors of the catalytic activity) on 2000 microstates of each of the 3000 HEAs considered here (an impossible job if DFT would be directly employed). The constraint that species concentration must be lower than 50% is used. A higher concentration of a specific species reduces the entropic effects that stabilize these catalysts. Hence, increasing the probability of structural dissociation into multiple phases, for instance. The probabilistic approach is, hence, used together with the adsorption energies to estimate catalytic activities (as described in the section “ Towards active HEAs for NRR” ) and also selectivities (as described in the section “ Towards Selectivity” ). 4. Inherent properties of the HEAs like the averaged valence electron concentration (VEC), averaged electronegativity (ELE) and averaged working function (WF) are correlated with the estimated activities to unravel the properties controlling the activity and selectivity. 5. The thermodynamical barriers of the potential limiting steps are calculated for 100 microstructures of a selected HEA. These are shown in the form of box plots and compared with the case of Fe (111) (details are in the section “ Potential limiting steps of the selected HEA” ). Quinary HEAs formed with the elements Mo-Cr-Mn-Fe-Co-Ni-Cu-Zn are randomly created. DFT calculations are performed over 1200 microstructures formed with the above-mentioned elements. For each microstructure, the adsorption energies of N2 * and N * (descriptors of the reaction) are computed and stored in a database (DFT details are in the section “ DFT Calculations”).Representation models of the microstructures are created and used to build neural network models that are trained on the adsorption energies from the DFT calculations. These models can compute adsorption energies almost instantaneously and, overcome the time-consuming task of performing thousands of adsorption energies with the DFT approach (details are in the section “ Deep neural networks” ).Using the deep neural network models, we calculate the adsorption energy of N * and N2 * (descriptors of the catalytic activity) on 2000 microstates of each of the 3000 HEAs considered here (an impossible job if DFT would be directly employed). The constraint that species concentration must be lower than 50% is used. A higher concentration of a specific species reduces the entropic effects that stabilize these catalysts. Hence, increasing the probability of structural dissociation into multiple phases, for instance. The probabilistic approach is, hence, used together with the adsorption energies to estimate catalytic activities (as described in the section “ Towards active HEAs for NRR” ) and also selectivities (as described in the section “ Towards Selectivity” ).Inherent properties of the HEAs like the averaged valence electron concentration (VEC), averaged electronegativity (ELE) and averaged working function (WF) are correlated with the estimated activities to unravel the properties controlling the activity and selectivity.The thermodynamical barriers of the potential limiting steps are calculated for 100 microstructures of a selected HEA. These are shown in the form of box plots and compared with the case of Fe (111) (details are in the section “ Potential limiting steps of the selected HEA” ). DFT calculations: The projected augmented wave method was used to solve the Kohn-Sham equations implemented in the Vienna ab initio Simulation Package (VASP) [26,27]. The wave functions were expanded using plane waves with a cutoff energy of 400 eV while a (4×4×1) k-point mesh was used to sample over the Brillouin zone. Smearing of 0.2 eV was employed to obtain partial occupations using the Methfessel-Paxton scheme of second order. Spin-polarized orbitals were used in the ferromagnetic (FM) state and the Bayesian error estimation functional with van der Waals correlation (BEEF-vdW) [28] was utilized to describe the Kohn-Sham Hamiltonian’s exchange and correlation term. The BEEF-vdW has been reported to be one of the most accurate functionals to describe adsorption energies on transition metal surfaces [29,30], and is the approach chosen for this study. The structural models were built into a 2×2×4 face-centered cubic (FCC) (111) slab with a vacuum of 20 Å to avoid interaction amongst periodic images, allowing the two topmost layers to geometrically relax. In contrast, the two bottom layers were fixed to the optimized bulk structure. Atoms positions were optimized until a maximum force of 0.08 eV/Å was obtained. It is common for calculation where single atoms are used forces convergence of the order of 0.03–0.01 eV/Å. However, the randomness of the HEAs adds complexity and lower convergence parameters need to be used. Other references have also employed this value [16,17]. Moreover, we tested for one case of N * with force convergence of 0.03 eV/Å. We got a difference of 0.02 eV in the adsorption energy. Lattice parameters of the slabs were set on a weighted average basis and assuming species has FCC bulk structures, similar to the work of T. A. Batchelor et al. [16]. Moreover, Clausen et al. [31] showed that possible remaining strain effects on the adsorption energy of small molecules are alleviated by the inherent distortion of the lattice in HEAs. Bulk optimizations were performed with a k-point mesh of 15×15×15 in an FCC structure, and the obtained lattice parameters are summarized in Table S1. Deep neural network: Although the values of ALPHA could be estimated under any first-principle approach, the high amount of possible microstructures makes the calculation of the N and N2 adsorption energies non-feasible from a computational time point of view. To circumvent this issue, a representation model of the microstructures that enables establishing a deep neural network (DNN) model permitting the computation of N and N2 adsorption energies almost instantaneously was built and used together with the DFT calculations. Though the DFT calculations were performed on 1200 microstructures, after data cleaning, the DNN was trained using 784 adsorption energies for N atoms sited on the hexagonal-close-packed sites. We have shown that N adsorbs strongly on this site (see Fig. S2), which also corroborates with the results of W. A. Saidi et al. [18]. For the case of N2, 784 adsorption energies of the N2 molecule on randomly created slabs were used to train the DNN model. The representation used to feed the DNN involves the specification of four regions of the HEAs microstructures and frequency counting of species on each specific region ( Fig. 2). These regions are then concatenated into a vector defining a regression problem, ∆ E N , N 2 = ∑ p R ∑ k metals C p , k N p , k i , where N p , k i is the number of atoms of specie k in the region p and R is the total number of regions, solved with the DNN. Each built vector represents one microstructure of a HEA of a specific concentration.The DNNs were built using the Keras library [32]. The data was trained in several networks where the best models were composed of dense sequential layers. The input layers were set with two hundred neurons and a linear activation function for the N adsorption energy and one hundred neurons together with a linear activation function for the dinitrogen adsorption energy. Six hidden layers composed of two hundred neurons each and a “relu” activation function were employed for the N adsorption training, while for the N2 adsorption training, one layer with 50 neurons and a “relu” activation function (L2 regularization function were employed in both cases). The output layers were built with a linear function. Loss function (mean squared error, MSE) between predicted adsorption energies and DFT computed adsorption energies were minimized using an Adam optimizer with a learning rate of 0.014. Our dataset utilized to build the DNN was randomly divided into a training set (80%) and test set (20%) for both N and N2 adsorption energies. The evolution of the loss function with the epoch number (training steps) is shown in Fig. S4 and confirms no overfitting phenomena. Towards active HEAs for NRR: A probabilistic approach based on the adsorption of N2 molecules and N atoms on the HEAs surface is employed to estimate the catalytic activity of HEAs. The basic principles of the approach were firstly proposed by T. A. Batchelor et al. [16] and are here further expounded on and extended to the use in NRR. The first assumption is that bonds formed between small molecules and catalytic surfaces have a local character, hence, are determined by the microstructure of the local site. This means that the vast composition of the HEA can be approached as an average over microstructures of the HEA, where each microstructure will contribute to the activity in a specific way, depending on the adsorption energy of N in the respective site. This is an approximation that concurs with the experimental situation with a random mixing and atomic dispersion in a real HEA. Moreover, the N2 fixation issue is accounted for by introducing the N2 adsorptions energy in the model. In a technical sense, to maximize the number of randomly created sites that deliver: i) Adsorption energies of N in between the obtained values for Fe and Ru. Ru has been proven to show activity towards NRR [33–35], while Fe is a known to be an efficient electrocatalyst for ammonia production [36,37] and appears at the top of the volcano plot – lower limit potential step [19]. Therefore, maximizing microstructures with similar N adsorption energy should ensure high catalytic activity for the specific HEA towards NRR. ii) To identify the number of sites that adsorbs N2 exothermically – better N2 fixation. These assumptions can be formulated into a probabilistic approach with: (1) P ( N 2 ) = ∑ Microstructures with ∆ E N 2 < − 0.5 eV ∑ All considered Microstructures (2) P ( N ) = ∑ Microstructures with ∆ E N ( Fe ) < ∆ E N < ∆ E N ( Ru ) ∑ All considered Microstructures (3) ALPHA = P N 2 × P ( N ) where P ( N 2 ) is the probability of finding sites with exothermic adsorption for N2, P ( N ) is the probability of finding microstructures with energy between ∆EN(Fe) and ∆EN(Ru) ALPHA is the probability of the two events happening. A HEA with high ALPHA should thus deliver high activity, while each microstructure is randomly created with the constraint that its species concentration reassembles a specific HEA concentration.The Gibbs free energy variation of the reaction N2 + * → N2 * can be calculated as ∆ G = ∆ E + ∆ ZPE + ∆ H vib + ∆ H rot + ∆ H trans − T ∆ S vib + ∆ S rot + ∆ S trans where ∆ ZPE is the variation on the zero-point energy, ∆ H and ∆ S are the variations of enthalpy and entropy, respectively, and ∆ E is the electronic energy change. To further validate and assess the parameters, we have suggested that a complete vibrational frequencies calculation is performed using DFT to assess the thermal effects. Performing this for conformations in the set, we find that (4) ∆ ZPE + ∆ H vib + ∆ H rot + ∆ H trans − T ∆ S vib + ∆ S rot + ∆ S trans ≈ 0.5 eV Therefore, we can estimate that ∆ G will be exothermic only when ∆ E is lower than − 0.5 eV, which is settled as a limit in Eq. 1. This condition for activity towards NRR is also found in the work of C. Ling et al. [38]. For other reactions, the analogous assessment of thermal effects is required with the corresponding change in Eq. 1.The adsorption energies of N atoms and N2 molecules on the HEAs microstructures were calculated as (5) ∆ E N = E N − E * − N 2 gas − phase 2 and ∆ E N 2 = E N 2 − E * − N 2 gas − phase The calculation of the first constraint used for the N adsorption, ∆ E Fe , was performed considering 2×2x5 BCC slabs on the (110) and (100) directions where results were − 1.1 eV and − 0.88 eV, respectively. ∆ E Ru was calculated using a HCP structure in the (001) direction resulting in adsorption of − 0.78 eV. The recent work by Megha Anand et al. [37] highlights that the best NRR catalyst Ru, is followed by Fe in terms of effectiveness in catalytic activity. Instead of using the exact values of Fe and Ru adsorption energies as the constraints in Eq. 2, we set those to be − 0.7 eV and − 0.9 eV, therefore slightly shifted towards the Ru instead of Fe. If other reactions are targeted, with key rate-limiting steps in the adsorption energies, adjustments of the targeted training parameters are required and can also be chosen from other rate-limiting parameters without loss of generality. Towards Selectivity: It is well known that most of the catalysts suffer from poor selectivity towards NRR due to the competing HER – protons being more likely to be activated and reduced on the catalytic surface than dinitrogen (N2). Selectivity can, hence, be ranked by analyzing the averaged value of ∆ E ( H ) and ∆ E N 2 over the microstructures of a HEA. In a first assessment, this allows the selection of the best catalysts as the ones with more positive values of ∆ E H − ∆ E N 2 [39]. We have tested, for 10 microstructures, if N2 would be adsorbed exothermically once H* atoms are on the catalytic surfaces (hydrogenated surface). Unfortunately, for all cases, N2 does not adsorb exothermically. Therefore, there is a competition between N2* adsorption and H* adsorption and, this is mitigated if the term ∆ E H − ∆ E N 2 gets more positive. Moreover, the energetics of both adsorbates, H and N atoms, scale linearly (see Fig. S1 (error in the energy of hydrogen) for details, both adsorbs on a threefold site). While the relation ∆ E H − ∆ E N 2  might be of interest to assess absolute values of selectivity, the main concern is to rank the HEAs faithfully based on such parameters. To approach this, and, using the scaling relation between H and N adsorption, the energetics of the hydrogen adsorption, ∆ E H can be exchanged by the energetics of N atoms adsorption, ∆ E N , to predict selectivity and, leading to: selectivity =  ∆ E N − ∆ E N 2 . This parameter will now be named SELE. Potential limiting steps of the selected HEA: For the selected catalyst, a statistical approach is employed to estimate the thermodynamical barriers of the potential limiting steps. (6) N2 + *→N2* (7) N2* + (H++e-) →NNH* (8) NH* +(H++e-) →NHH* The computational hydrogen electrode approach, as proposed by Nørskov [40], was applied to model the electrochemical reactions. This approach assumes a coupled electron-proton transfer simplifying the demanding calculation of solvation energies of ionic species. The free energy variation of each electrochemical/chemical reaction was calculated for 100 microstructures of the selected HEA using DFT as: (9) E ad = E adsorbate * − E * − ∑ i n i μ i where E adsorbate * is the self-consistent-field (SCF) energy of the adsorbed intermediate corrected by the zero-point energy (ZPE) of the adsorbate, E * is the SCF energy of the pure slab and n i is the number of species i with chemical potential μ i . Moreover, μ H , μ N are the chemical potentials of hydrogen and nitrogen, respectively, that are obtained as: (10) μ H = 1 2 E H 2 (11) μ N 2 = E N 2 (12) μ N = E N 2 2 (13) E H 2 , N 2 = E scf + ZPE + H vib + H trans + H rot − T S vib + S trans + S rot + PV The usual approach to depict the energy landscape of reaction pathways on transition metal surfaces needs to be adapted to fulfill the restriction imposed by the randomness of the HEAs. Indeed, every microstate of the structure (that together resemble the HEA surface) delivers one different energetics for the concerning reaction step. Therefore, what we can get is a distribution of energies for each associated transformation. Box plots are a common tool to report the overall patterns of a group. This summarizes important information about the group as the minimum, the first quartile, the median, the third quartile and the maximum. Here, the computed thermodynamical barriers are shown as a box plot.In this section, the accuracy of the developed DNN is discussed and compared to preview data reported in the literature (See the deep neural network model section). We also performed a deeper analysis of the relations between ALPHA, SELE, and intrinsic HEAs properties, with the resulting HEA(s) activities given in the subsection Computed electrocatalytic activity (ALPHA). Finally, the most promising novel HEAs for NRR are pointed out in The selected HEAs section and its potentials limiting steps investigated. The deep neural network model: The DNN employed here to estimate adsorption energies of N2 molecules and N atoms on the surface of HEAs displayed reasonable accuracy with mean absolute errors (MAEs) of 0.09 eV and 0.20 eV, respectively (Figs. 2a and 2b). One can note that the mean absolute error of N2 adsorption is below the typical resolution of DFT while the N adsorption shows a slightly higher error. In context to this, we would like to remind the reader that we have used the hypothesis that bond formation is a local process, and hence that the adsorption energies can be obtained by specifying elements close to adsorbates and their location. However, the model assumes symmetric adsorption sites as input parameters. This might be one of the causes of the better accuracy of the model found for N2 adsorption than the model found for N adsorption since N2 sits on the top site in a very symmetric environment (only one species is accounted for in region one, Fig. 2). On the other hand, N atoms sit on threefold HPC sites. Therefore, symmetry breaking would be observed depending on the coordinating species, leading to higher errors in the built model (three species are considered in region one for this case). Still, others have reported ML predicted MAEs of about 0.2 eV regarding the DFT adsorptions energies [41], which inherently also has an error of about 0.2 eV within Beef-vdW functional [28,29]. Therefore, it pays off the employment of these models in pro of a considerable gain in computational time, allowing the removal of unpromising catalysts to be experimentally processed or by DFT calculations. Computed electrocatalytic activity (ALPHA): ALPHA and SELE were calculated for three thousand randomly created quinary HEAs of the elements Mo, Cr, Mn, Fe, Co, Ni, Cu, and Zn. The relationship between ALPHA and SELE is displayed in Fig. 3(a) and (b). For this task, two thousand microstates of each created HEA were considered to assess the averaged quantities for SELE and the probabilities associated with ALPHA. Higher and lower values of SELE lead to lower values of ALPHA. An optimal value is obtained when SELE is between − 0.25 and 0.0, building a volcano-shaped relationship. Interestingly, the shape obtained for this relationship is also reproduced for ALPHA vs. averaged N adsorption energies. Hence, the averaged N adsorption energies emerge as the main influencing factor for SELE. The averaged N2 adsorption appears as an almost fixed shift in SELE since they are computed as the average of cases with adsorptions higher than 0.5 eV. Hence, minimal variance is revealed when comparing distinct HEAs. As expected, the cases with higher ALPHA have averaged N adsorption energy around − 0.75 eV – the ∆ E N Ru = −0.78 eV since ALPHA is set to maximize the probability of sites with adsorption similar to Ru.Something interesting differentiates the case studied here from the volcano-shapes reported for NRR in the literature [23,36]. HEAs with the same averaged N adsorption (SELE) can display different activities (Fig. 2a), producing a volcano relation where data is spread inside the volcano shape. Two characterizing cases were selected for further analysis to gain insights into the obtained relationship. The first case, Mo0.38Fe0.31Co0.19Ni0.06Cu0.06, has high ALPHA of 0.14 with a SELE value of − 0.15, while the second case, Mo0.25Cr0.06Mn0.31Cu0.06Zn0.31, has ALPHA 0.00 and SELE − 0.15 eV. Though both have similar averaged N adsorption energies of − 0.75 eV and − 0.68 eV (hence, similar SELE), their distributions are completely different (Fig. 3d). Most cases end up in the required P(N) region for Mo0.38Fe0.31Co0.19Ni0.06Cu0.06. For Mo0.25Cr0.06Mn0.31Cu0.06Zn0.31, the cases are distributed on high energy and low energy values, leading to low ALPHA, yet similar SELE. The problem faced here has a multi-dimensional character, and due to the need to get averaged quantities, information is lost, thus, explaining the filled volcano-shaped relation between ALPHA vs. SELE.Adsorption energies are widely employed to characterize activities in distinct fields of electrocatalysis [23,36,37]. Here, the direct application of the similar quantity, averaged N adsorption as the descriptor of catalytic activity of HEAs towards NRR, is shown to be insufficient to uniquely characterize each HEA, as discussed above. The plot of the averaged valence electrons in the occupied d orbitals ( γ ) vs. the averaged adsorption energies of N and N2 brings insights into how to properly explore and develop a unique descriptor for the activity of HEAs ( Fig. 4). N and N2 adsorption strength correlate with the conventional approach to analysing the d-band center of transition metals [42]. This is in our view closely associated with the number of valence d electrons in the system and, of course, the energetic position of the states. The results displayed a close linear relation for the case of averaged N adsorption energies with R2 of 0.76 (R2 = goodness of the linear relation), but a widespread data point for the case of N2 averaged adsorption energies, R2 of 0.45. Here, calculations of N2 adsorption on the microstructures are performed on the top site. So, the value of valence electrons in the occupied d orbital of the species where N2 is adsorbed must have a much stronger influence than the averaged relation inherent in γ . Hence, γ alone cannot fully describe the averaged N2 adsorptions. On the other hand, this issue is mitigated in the description of averaged N adsorption by the stronger influence of the three atoms coordinating the adsorbate, threefold HCP. Hence, γ results in a better descriptor for the averaged N adsorption energy. Again, information is lost when performing the averages and there exists a need to introduce a second property to better correlate the averaged adsorption and γ .Electronegativity measures the electron affinity of certain elements when a covalent bond is formed. Under the assumption that electronegativity would influence the redistribution of d electrons during the bond formation between adsorbate and catalytic surface, H. Xu et al. [43] showed that electronegativity could be employed together with the number of d electrons of a species as a descriptor of the O and OH adsorption energies on single metal catalysts. Using the HEA’s averaged electronegativity in the plot of N2 vs. γ as a color map, one sees that thought at same γ , and different averaged N2 values are observed. Moreover, these values mainly vary with the weighted-averaged electronegativity of the HEAs (we will name the weighted averaged electronegative ELE from now on). Generally, higher ELE leads to more negative values of averaged N2 adsorption, and inspecting the relationship between the averaged N2 adsorption energy vs. γ / ELE (Fig. S5) a better R2 of 0.58 is obtained in comparison to the previews value of R2 0.45 for averaged N2 adsorption energy vs. γ . This reflects in the volcano plot (Fig. 3a), where higher activities are found for the cases with higher ELE once the probability of finding N2 adsorption exothermically increases under these circumstances.As the conventional (Pauling) electronegativity scale is defined from covalent bonding, it causes concerns in metallic alloys with domination of metallic bonding. Therefore, we have also assessed how the obtained relations behave when changing the electronegativity scale (Fig. 3a-b and S6). Clearly, no change is observed when varying from the Pauling scale to the Mülliken scale defined from the arithmetic mean of the ionization energy and the electron affinity (these electronegativity measures scale linearly for the species investigated here, Fig. S7). On the other hand, no correlation between ALPHA and the electronegativity employing the Allen scale (The Allen scale is the average one-electron energy of the perceived valence shell electrons in the ground state in the free atom) is observed (Fig. S6d). Moreover, as the electronegativity could be considered a non-ideal descriptor in a metallic alloy, we explored the possibility of using the averaged work functions of the HEA as a further modification of γ as a descriptor of ALPHA where the work functions are computed for the pure bulk phases and, further, weight-averaged for the HEAs. This approach displayed no correlation with ALPHA (Fig. S6c), and together with the lack of correlation with the average one-electron energy in the Allen electronegativity, one can summarize that the local environment and effects beyond single atom properties are vital in constructing a descriptor for charge transfer in-between elements and catalytic activity of the HEAs.The obtained relations indicate that ALPHAs of HEAs can be conveniently described by ELE and γ , properties easily assessed by knowing the HEAs composition and concentrations. The relation between activity (ALPHA) as a function of γ and ELE shows that higher activities are more likely to be obtained when γ is between 6 and 6.5 and ELE is higher than 1.9 (Fig. S8). It is also important to emphasize that calculations were performed here in an FCC (111) surface, and, experimentally, the HEAs phases can vary. To, somehow, capture this information, another construction for the description of the catalytic activity is constructed on the VEC of HEAs (VEC and γ scales linearly, hence, similar relation with ALPHA). Sheng Guo et al. [44] showed that HEAs with VEC higher than eight must likely form FCC structures. This similarity allows the removal of cases where the HEAs come with VEC smaller than 8, increasing the probability of getting an FCC phase upon synthesis procedure (emphasis here is given to an FCC lattice just because calculations were performed with this structure). Fig. 5 graphically shows this analysis by plotting the values of ELE vs. VEC of each HEA together with their APLHA as the color map. Higher ALPHA values are more likely to be obtained when ELE is higher than 1.9 and VEC is between 7.5 and 8.5. Hence, a map towards higher ALPHA(S) is found only using intrinsic properties of the HEAs like VEC and ELE.The selected catalysts’ cases presenting ALPHA higher than 0.1 are summarized in Table 1. We have grouped HEAs in two sets: i) the three cases with the highest ALPHA and ii) the cases with ALPHA higher than 0.1 and, hence, ranked based on SELE. The highest ALPHA is obtained for Mo0.38Fe0.31Co0.19Ni0.06Cu0.06. On the other hand, C. J. H. Jacobsen et al. [45] have shown that MoCo is a promising catalyst for electrocatalytic ammonia production. Therefore, it is not surprising that Mo0.38Mn0.06Fe0.13Co0.38Ni0.06 and Mo0.31Mn0.06Fe0.13Co0.44Cu0.06, having balanced values between Mo and Co and minor concentrations of other species, display high activity. This also confirms the robustness of the screening strategy employed in this work.The second set of compounds ranked based on selectivity displayed Mo0.44Co0.38Ni0.06Cu0.06Zn0.06 as the best option. Again, the balanced Mo-Co ratio leads to high activity while introducing Zn, Cu, and Ni in small quantities, leads to an increased value of SELE. Furthermore, species with higher VEC (VEC ∝ γ ) present lower bond strength between N atoms and the catalytic surfaces, pushing the SELE to more positive values. Interestingly, all selected cases presented ELE close to 2 since, generally, this pushes the N2 adsorption towards more negative values, thus, resulting in higher ALPHA.Though HEAs formed from quinary components of the elements Mo, Cr, Mn, Fe, Co, Ni, Cu, and Zn were randomly created in this work, the best HEAs serving as novel catalysts for NRR are mainly formed of Mo-Fe-Co and with minor or non-quantities of other species. Moreover, at least 30% of the HEAs are made of Mo for all cases. N2 molecules on transition metal surfaces correlate with the d band center position with respect to the Fermi level of the transition metal due to the “push-pull” mechanism with σ-donation and π * -back donation [42]. Amongst the investigated species, Mo is the one with d band center closer to the Fermi level [46], thus, resulting in a higher probability of delivering strong N2 adsorption. This in turn increases the value of ALPHA. Aiming to confirm this hypothesis, the adsorption energy of N2 is calculated (see Fig. S3 for details) and displays the stronger adsorption on Mo as compared to other species. Therefore, Mo can be considered as the main N2 molecules fixating center on the catalytic surface of the HEA during the NRR cycling. While higher amounts of Mo (yet still inside the high entropy alloy stability zone) would probably assist in the activation of N2 molecules, but would also result in higher adsorption energies for the N atoms, scaling with the H adsorption and thus increased HER. This implies that a too strong N adsorption leads to: i) very slow rates of NRR reaction, ii) catalytic surface poisoning. To retain selectivity and not to have predominant HER reactions, the concentration of Mo has to be balanced by introducing Co and Fe species. Considerable concentrations of Cr and Mn in the HEA also deliver strong adsorption of N as compared to other species. Hence, these are not optimal options to make this balance. On the other hand, Ni, Cu, and Zn can contribute to weaker adsorption energy values.The histograms of the HEAs Mo0.38Fe0.31Co0.19Ni0.06Cu0.06 and Mo0.44Co0.38Ni0.06Cu0.06Zn0.06, cases selected as the best alternatives on the two sets presented in Table 1, are displayed in Fig. 6. The relation between the probability of N2 adsorption is directly proportional to the concentration of Mo species on the HEA, as discussed above. Here, Mo0.38Fe0.31Co0.19Ni0.06Cu0.06 presents 42% of its active sites presenting N2 adsorption in the exothermic region while Mo0.44Co0.38Ni0.06Cu0.06Zn0.06 has 50% of the cases into the exothermic region (Fig. 6). The increment in the probability of finding exothermicity in the N2 adsorption is, here, due to the increment in Mo concentration. On the other hand, the chance of finding sites with N adsorption energy close to the obtained for Ru is smaller for the case Mo0.44Co0.38Ni0.06Cu0.06Zn0.06, 20% of the sites, in comparison to Mo0.38Fe0.31Co0.19Ni0.06Cu0.06 presenting 32% of the sites in the optimal region. Mo0.44Co0.38Ni0.06Cu0.06Zn0.06 exhibits sites with N adsorption energy as positive as 1 eV, and this is due to the higher concentration of species with higher VEC like Ni, Cu, and especially Zn. While this is positive to the selectivity of the HEA pushing the average N adsorption to − 0.57 eV as compared to − 0.76 eV in Mo0.38Fe0.31Co0.19Ni0.06Cu0.06, this comes at the price of lower activity.The tendency of a HEA to form a solid solution instead of dissociating into multiple phases can be determined via either combination of (Caloric and electrochemical) experimental measurements or theoretically via quantum mechanical calculations of alloy bonding, effects of lattice entropy from mixing, and temperature effects. X. Yang et al. [47] have demonstrated that estimations can be obtained via empirical data that estimates atomic sizes, formation enthalpy and configurational entropy. When the terms δ = ∑ i = 1 N C i 1 − r i r ave 2 ≤ 6.6 % and Ω = T m Δ S mix Δ H mix ≥ 1.1 the HEA might form a solid solution. Here, δ is a parameter gauging the atomic size difference that depends on, C i , the atomic percentage of ith component, r i atomic radius of ith component and r ave the averaged atomic radius. Ω parameter depends on the concentration weighted averaged melting temperature, T m , the configurational entropy Δ S mix = − R ∑ i = 1 N C i ln C i and mixing enthalpy Δ H mix = ∑ i , j N C i C j 4 H i , j where H i , j is the mixing enthalpy of binary alloys computed based on Miedema macroscopic model and obtained in the work of A. Takeuchi et al. [48].Apart from predicted activity, the individual concentrations of the elements and their respective atomic radius need to be taken into account also for the predicted HEAs. As such, it is a compromise to retain an entropically stabilized structure and, at the same time, change the composition to strive for higher activity without sacrificing too much of the entropic stabilization and thus increasing the risk of precipitation and phase separation for some of the elements. For all pointed HEAs the values of δ are smaller than 6.6%, and values of Ω are higher than 1.1 (Table 1). Hence, these HEAs would likely form a solid solution as previously described in the introduction.We selected the best case in Table 1, Mo0.38Fe0.31Co0.19Ni0.06Cu0.06, to perform a comparative analysis of the thermodynamical barriers of the potential limiting steps with the case of Fe(111). The energetics of the reactions for Eqs.(6–8) are displayed in box plot format for the HEA and as red lies for the case of Fe(111) ( Fig. 7).The first step, the N2* adsorption, is endothermic on the Fe(111) surface, while for at least 25% of the 100 microstates of the HEA, this reaction becomes exothermic. Since, N2 capturing is one of the main issues in NRR, the existence of local sites on the HEA surface with strong N2 bonds is considered a plus for the electrochemical NRR. The activation of the N2 * is the second investigated reaction transformation. There, in the case of Fe, the thermodynamical barriers is 1.1 eV. This means, based on the computational hydrogen electrode approach, that at leads a potential of 1.1 V vs. RHE is needed to activate N2* and form NNH* on iron. This picture changes for the case of the HEA. There, the lowest observed case displays a thermodynamical barrier of about 0.74 eV, while at least 25% of the 100 tested microstates of the HEA display barriers lower than 1 eV. Finally, for the desorption of the NH* and forming NH2 * , thermodynamical barriers go from 0 eV up to 0.6 eV for the HEA vs. 0.4 eV for the case of Fe. Certainly, considering the distal pathway, the first activation of the N2* molecule is the PLS. As demonstrated here, the randomness of the HEA surface opens up the possibility of lower PLSs as compared with the case of Fe(111) and still keeping the desorption of the NH* in a reasonable energetic value.All the above results are expected to be directly applicable for NRR in gas-phase or in H2O/N2 vapor conditions as in a gas diffusion cell, while several additional considerations have to be taken into account in a practical application in a solid-liquid reaction cell. First, one needs to consider the low N2 solubility in water (1.3 ×10−3 mol/L) [13] together with the high adsorption energies of H2O* , OH* and H* on the catalytic surface that might create a water coverage, hydroxylation or hydrogen coverage in aqueous electrochemical cells depending on the conditions regarding electrolyte pH and the employed electrochemical potential. These points can limit the N2 coverages on the catalytic surface, hence, deteriorating the delivered FE and activity (besides the dominant kinetics of HER over NRR [49]). In this context, even with an optimized catalyst, values of activity and FE could be way off from the expected due to the lack of available catalytic sites for the NRR reaction to proceed. One way to mitigate such issues is to work with a gas-diffusion electrode (GDE) that increases the N2 coverage by adjusting the back N2 pressure. In conjunction with an optimized catalyst, this strategy can facilitate the activity towards NRR due to the increased N2 coverage and the suppressed H2O presence that inhibit HER and surface coverages with water or hydroxyl groups, hence, increasing FE towards NH3. Another option is the application of an aprotic electrolyte with increased N2 solubility [50,51]. This would also promote the N2 coverage and mitigate HER. Though HEAs concentration and compositions were optimized to deliver higher catalytic activity and selectivity towards NRR, most of the discussions presented here need to be carefully evaluated when an aqueous electrochemical cell is considered. As long as the underlying electronic properties of the HEA surface is consistently scaled to lower N2 and N adsorption energies upon hydroxylation of the surface, the results can be directly transferrable. However, also potential decrease of the frequency of competing N2 fixation and effects from differently induced kinks and terraces in-between different compositions would be required to behave in a scalable way compared to the flat surfaces screened here. For differences in any of these scalings, a case-to-case investigation has to be performed for the HEAs and their corresponding hydroxylated surfaces and surface Pourbaix diagrams, to evaluate and rank the most interesting HERs.Another point that has to be carefully evaluated is the selection of parameters used in this work to define what is an efficient HEA for NRR. Within the probabilistic approach, we selected HEAs compositions that maximize the number of sites endothermically adsorbing N2 and also adsorbing N atoms (descriptor of catalytic activity) with adsorption energy close to Ru, as explained in the methods section. Even though this approach leads to the identification of active HEAs towards NRR, it also selects compositions with considerably high H* . This means that the selectivity can be deteriorated and also the activity due to the low coverage of N2. Kani et al. [13] hypothesized that the most efficient catalyst for NRR, instead, would be the one with lower hydrogen adsorption H* providing lower H coverage and also lower HER activity – higher activation barrier. This option would allow the application of lower cathodic potentials without fully covering the catalytic surface with H atoms. Moreover, hydroxylation of the catalytic surface would be partially suppressed due to the destabilization of the OH groups adsorption, hence, facilitating the existence of active sites for the N2 adsorption and lower potentials would likely lead to higher current densities. In our opinion, however, this would lead to intrinsically low NRR rates due to the increment of the thermodynamical limiting step of the NRR path. Therefore, the definition of what is a highly efficient catalyst for NRR is a gray zone that depends, amongst other things, on the experimental conditions in place and which rate one wishes to achieve. The approach presented in this work is flexible, however, and can easily be modified toward desired selectivities and rates to fit experimental conditions beyond the ones proposed in this work.By employing DFT together with Machine Learning and deep neural network techniques, a screening protocol enabling a rational selection of novel catalysts for NRR was developed to search over a large compositional space of HEAs. Activities and selectivities were computed by a probabilistic approach that incorporates the adsorption of N, H, and adsorption of N2 atoms. The computed HAE(s) activities reveal a volcano-shaped relationship with Mo0.38Fe0.31Co0.19Ni0.06Cu0.06 located on the top of the volcano. Moreover, a rank based on selectivity and activity pointed to Mo0.44Co0.38Ni0.06Cu0.06Zn0.06 as an alternative option that balances activity and selectivity. We also include a critical analysis of different aspects of electronegativity in connection to the work function of the elements showing that the local composition and charge transfer are necessary to formulate key descriptors of catalytic activity. Instead, valence electron concentration of HEAs with either different energy d-states or electronegativity, forms descriptors of the catalytic activity. The approach shows a promising pathway to conveniently screen candidates for catalytic activity and selectivity for a given catalytic reaction, here exemplified by the NRR reaction. The screening disclosed and quantified existing relationships between HEAs composition and catalytic activities towards NRR that bears the promise of accelerating the search for complex NRR catalysts.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was financially supported by the European Union’s Horizon 2020 research and innovation programme under the call H2020-LC-SC3-2020-RES-RIA in the TELEGRAM project [grant agreement No 101006941]. The computations were enabled by resources provided by the Swedish National Infrastructure for Computing (SNIC) via the project SNIC 2021/5-282, and funding by the Swedish Research Council through grant agreement no. 2019-05591. Supplementary information is available in the online version.Supplementary data associated with this article can be found in the online version at doi:10.1016/j.nanoen.2022.108027. Supplementary material. .

A computational approach to judiciously predict high-entropy alloys (HEAs) as an efficient and sustainable material class for the electrochemical reduction of nitrogen is here presented. The approach employs density functional theory (DFT), adsorption energies of N atoms and N2 molecules as descriptors of the catalytic activity and deep neural networks. A probabilistic approach to quantifying the activity of HEA catalysts for nitrogen reduction reaction (NRR) is described, where catalyst elements and concentration are optimized to increase the probability of specific atomic arrangements on the surfaces. The approach provides key features for the effective filtering of HEA candidates without the need for time-consuming calculations. The relationships between activity and selectivity, which correlate with the averaged valence electron concentration and averaged electronegativity of the reference HEA catalyst, are analyzed in terms of sufficient interaction for sustained reactions and, at the same time, for the release of the active site. As a result, a complete list of 3000 HEAs consisting of quinary components of the elements Mo, Cr, Mn, Fe, Co, Ni, Cu, and Zn are reported together with their metrics to rank them from the most likely to the least likely active catalysts for NRR in gas diffusion electrodes, or for the case where non-aqueous electrolytes are utilized to suppress the competing hydrogen evolution reaction. Moreover, the energetic landscape of the electrochemical NRR transformations are computed and compared to the case of Fe. The study also analyses and discusses how the results would translate to liquid-solid reactions in aqueous electrochemical cells, further affected by changes in properties upon hydroxylation, oxygen, hydrogen, and water coverages.

The efficient conversion of biogas to liquid fuels (Bio-GTL) could become a key industrial process in the future bio-economy. Biogas is an attractive renewable energy source and can be produced by anaerobic digestion of organic wastes in the presence of microorganisms. Methane and carbon dioxide are the major constituents of biogas, although traces of ammonia, hydrogen, oxygen, hydrogen sulfide, water vapor and other impurities can be present depending on the feedstock used (landfill, sewage sludge, agricultural waste, etc.) [1]. Although biogas can have different uses [2], its high concentration of CO2 and CH4 makes it ideal for syngas production by dry reforming with minimal purification. Syngas is a CO and hydrogen-rich gas mixture that can be readily converted to high-performance liquid fuels by the Fischer-Tropsch process [3,4].The catalytic reforming of biogas has been widely studied for the last two decades. Currently, the main limitation to industrialize this technology is the low catalytic stability associated with coke deposition and metal sintering under reaction conditions. Owgi et al. [5] surveyed different catalytic systems for the dry reforming of methane, analyzing the effect of active metals, support materials, promoters, and preparation methods, and concluded that the design of cost-effective and stable catalysts for biogas reforming is an unsolved challenge. Recently, significant progress has been made in understanding catalysts under working conditions using in situ and operando techniques [6–8]. The investigation of catalysts under realistic conditions allows correlating the dynamic structural changes of the catalyst surface in the presence of reactants with catalytic performance [9], providing unique insights that could inform the rational design of better catalysts.Nickel catalysts are favored in industrial applications for their low-cost. Different strategies are used to control the nickel particle size and mitigate coke deposition, such as the use of basic supports and doping of the catalyst with alkaline metals, which enhances the metal dispersion and favors coke gasification by the reverse Boudouard reaction [10,11]. Indeed, the addition of potassium as a promoter on nickel catalysts has been extensively investigated. However, the role of potassium in Ni-based catalysts during reforming is not fully understood. Borowiecki et al. [12] proposed that the effect of potassium as a promoter is directly related to its location and chemical state on the catalyst surface. They showed that only those potassium atoms close to nickel sites promote gasification of carbonaceous deposits produced in methane cracking. Frusteri et al. [13] suggested that potassium induces an electronic effect on Ni/MgO catalysts, inhibiting the rate of coke formation, carbon nucleation, and carbon diffusion through the nickel sites. In a previous study, we reported that introducing a potassium promoter on a Ni/MgAl2O4 catalyst changed the nature of the catalytic active sites and enhanced their coking resistance [14]. We proposed that the existence of Ni-K containing phase favors the gasification of carbonaceous deposits by the reverse Boudouard reaction and reduces the sticking probability of CO/CO2 in dissociative adsorption.At present, several questions about the effect of potassium in dry reforming remain unsolved, which limit our ability to design better catalysts. Where is the potassium located on the catalyst? What properties of the material are affected? Is potassium a carbon gasifier? In this work, we investigate how the addition of potassium affects the structural properties, reducibility, and chemical state of Ni-based catalysts, as well as its relationship with the catalytic activity and stability in the dry reforming of methane. To this end, we study a series of xK-Ni/MgAl2O4 catalysts with identical amounts of nickel but different potassium loads (x between 0 and 5 wt%). The study includes a combination of in situ X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), temperature-programmed reduction (TPR), and diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) coupled with mass spectrometry (MS) to investigate the structural and redox changes occurring during the dry reforming of methane.The support, a stoichiometric aluminium magnesium spinel (MgAl2O4), was synthesized by co-precipitation and treated at 900 °C for 24 h following a previously described procedure [15]. The solid obtained was labelled as MgAl.The catalysts were prepared by wet co-impregnation of potassium (KNO3) and nickel (Ni (NO3)2·6H2O) salts. Adequate amounts of these salts to obtain 1, 3 or 5 wt% potassium and 10 wt% nickel loading were diluted in water and mixed with the support. Excess water was removed by roto-evaporation at 60 °C. The samples obtained were dried overnight at 100 °C (fresh catalysts) and then calcined at 550 °C for 4 h (calcined catalysts). The catalysts were designated as xK-Ni/MgAl, where x indicates 1, 3 or 5 wt% of K2O. For comparison, an unpromoted catalyst was also synthesized by impregnating the bare support with the nickel salt solution and designated as Ni/MgAl.For every characterization study, a calcined sample of catalyst from the original synthesis batch was used, except for those techniques using spent catalysts.Nitrogen adsorption-desorption experiments of the calcined catalysts were carried out in Micromeritics Tristar II instrument to evaluate their textural properties. Prior to measurement, the samples were outgassed in vacuum at 250 °C for 4 h. The specific surface area and pore volume were estimated using the Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) models, respectively.UV–Visible spectroscopy characterization of the calcined catalysts was carried out in a Shimadzu 2101 spectrometer with a diffuse reflectance accessory. The electronic spectra were recorded in the wavelength range from 190 nm to 900 nm. The bandgap energies were estimated from the intercept of the tangents to the plots of (α * hν)1/2 against the photon energy [16]. In situ X-ray diffraction and reducibility analyses of the calcined catalysts were performed in a high-temperature chamber Anton Paar HTK 1200 coupled with an X'Pert Pro Philips diffractometer equipped with Ni-filtered Cu Kα radiation (40 mA, 45 kV) and a X'Celerator detector. The powder XRD patterns were recorded with a 0.05° step size in the 10° to 90° 2θ range and 30 s time steps. Measurements were carried out every 100 °C in the 25–900 °C temperature range while flowing 100 NmL min−1 of 5% H2 diluted in argon through the chamber.XPS measurements of the calcined catalysts were carried out in a SPECS spectrometer equipped with a PHOIBOS 150 MCD analyzer working at fixed pass energy of 40 eV and 0.1 eV resolution for the studied zones. Al Kα radiation (1486.6 eV) was used at 250 W and 12.5 kV. Prior to analysis, each sample was pressed into a thin disk. All XPS spectra were recorded at room temperature with the binding energy calibrated with Mg 2p at 50 eV. The samples were reduced in situ at 800 °C during 1 h in 5%H2/Ar mixture in a high-pressure treatment cell (HTHP Cell). After reduction, the evacuation of the gases was maintained overnight before the spectrum was acquired at room temperature under vacuum. The analytical chamber operates under ultra-high vacuum (10-10 mbar). In situ TPR-DRIFTS analysis of the calcined catalysts was performed using a high-temperature environmental reaction chamber with ZnSe windows supported in a Praying Mantis (Harrick) optical system and coupled to a Thermo Nicolet iS50 FTIR spectrometer with MCT detector. The spectra were recorded as an average of 64 scans with 4 cm−1 of spectral resolution per spectrum. 80 mg of finely ground calcined catalyst was loaded in the cell for each experiment. The temperature-programmed reduction (TPR) was performed by feeding a flow of 5% H2 in Ar (50 NmL min-1) and increasing the temperature from room temperature to 750 °C at a rate of 10 °C min-1. The spectra were recorded in continuous series mode using the OMNIC 9.1 software and the temperature was simultaneously monitored using a software written in Labview. The effluent gases were analyzed on line by mass spectrometry (PFEIFFER Vacuum Prisma Plus).A quartz U-shaped reactor fitted in a homemade temperature programmed device equipped with a TCD detector was used to analyze the reduction (TPR) of the calcined catalysts. For the TPR quantitative analysis, the TCD signal was calibrated with a CuO pattern (Strem Chemicals 99.999%). Water and other condensable gases were trapped in a cryogenic bath of dry ice and acetone. An analogous experiment was performed by omitting the cold trap and substituting the TCD by an on line mass spectrometer (Pfeiffer Vacuum Prisma Plus) to monitor the gases evolved during the TPR experiment. The modelling of the TPR was performed using the optimization toolbox in MATLAB 2020a. The modelling involved the numerical optimization of kinetic parameters with the resolution of a system of ordinary differential equations (ODEs), which describes the TCD signal over time (see Supporting Information for details). Prior to modelling, experimental data were conditioned by removing the electric noise of the TCD using a loess filter. Background subtraction was applied using a piecewise cubic Hermite interpolating polynomial. The time window used in the fitting was clipped to that in which reduction processes were observed. The resolution of the system of ODEs was carried out with an adaptive Runge-Kutta algorithm. The optimization started from guess values for the parameters provided manually. The parameter values were then iteratively refined using a combination of Nelder-Mead and Levenberg-Marquardt optimization algorithms to minimize the sum of squared residuals between the experimental TPR signal and the model prediction. The confounding between fitted parameters was assessed by their standard errors, estimated by numerically computing the Jacobian of the objective function at the optimum.A cold-cathode Hitachi® S4800 SEM-FEG microscope was used for SEM analysis of reduced (800 ºC during 1 h in 5% v/v H2/N2) and spent catalysts. High resolution transmission electron microscopy (HR-TEM) micrographs of reduced catalysts were taken with a Talos® F200S FETEM microscope operated at 200 kV. At least 200 particles were measured to assess the diameter of nickel particles. The volume-surface average diameter was estimated assuming a spherical morphology with the following expression [17]: (1) D p = ∑ n i d i 3 ∑ n i d i 2 where n i represents the number of particles with diameter d i .The nickel exposed surface area and the metallic dispersion of the catalysts were estimated by H2 chemisorption pulses based on the quantity of hydrogen chemisorbed over nickel surface atoms. The experiments were performed in a U-shape quartz reactor loaded with 800 mg of sample. Firstly, the sample was reduced in situ at 800 °C under 50 NmL min-1 of 5% H2/Ar for 1 h, and then the temperature was decreased to 60 °C in Ar flow. At this temperature, calibrated volume pulses of 250 μL of pure hydrogen were successively introduced until saturation was achieved, indicating that hydrogen species covered the nickel metallic surface thoroughly. The pulses were monitored from the H2 signal (m/z= 2) by mass spectrometry (PFEIFFER Vacuum Prisma Plus).Temperature-programmed oxidation (TPO) experiments were carried out using approximately 50 mg of spent catalysts to analyze the carbon deposits formed in the reforming reactions. The same equipment was used as in the TPR measurements. Samples were heated from room temperature to 900 °C with a rate of 10 °C min-1 under 50 mL min-1 SPT of O2/He (10% v/v) flow. The signals of CO2 (m/z= 44), O2 (m/z= 32), and CO (m/z= 28) were followed on line by mass spectrometry in a PFEIFFER MS Vacuum Prisma Plus.A dispersive Horiba Jobin Yvon LabRam HR800 Confocal Raman microscope using a green laser (532.14 nm) working at 5 mW and with a 600 grooves mm-1 grating was used to record the Raman spectra of spent catalysts. A 50x objective (Olympus) was used in the microscope with a confocal pinhole of 1000 µm.Powder X-ray diffraction (XRD) analysis of spent catalysts was performed on a Siemens D-500 diffractometer using a Ni-filtered Cu Kα radiation at 40 mA and 45 kV. The diffraction patterns were recorded in the 2θ range from 10–90° using 0.05° step size and 300 s time steps.The catalytic activity in the dry reforming of methane (DRM) reaction was evaluated at atmospheric pressure in a Microactivity Reference (PID Eng & Tech) reactor coupled to a microGC (Varian 4900) equipped with Porapak Q and MS-5A columns and TCD detectors. A fixed-bed tubular reactor of 9 mm inner diameter made of Hastelloy was used for all the experiments. Prior to each test, 200 mg of non-diluted calcined catalyst (sieved to 100–200 µm) was reduced at 800 °C for 1 h in 100 NmL min-1 of 50% H2/N2. All catalytic tests were carried out at 30 NL g-1 h-1 space velocity using a molar ratio CH4/CO2 = 1 at 650 °C for 48 h. Experiments were performed at three different partial pressures of 20, 30 and 40 kPa for each reactant, using nitrogen to balance the total pressure at 100 kPa.Potassium addition leads to a change in the textural properties of the Ni-based materials prepared. Fig. S1 summarizes the textural properties of the materials after calcination. The mesoporous support presents a monomodal pore size distribution peaking at ~20 nm, which remains unchanged upon nickel impregnation. The observed BET specific surface area (SBET) for the support is 83 m2 g-1. After nickel incorporation, the specific surface area decreases slightly to 71 m2 g-1, suggesting that nickel gets well distributed throughout the pores of the support. The surface area strongly decreases upon adding potassium to the catalyst (Table S1). This may indicate that, impregnation with potassium leads to reconstruction of the support surface, probably with formation of strong basic sites [18]. Fig. 1A shows the UV–Vis spectra of the calcined catalysts in the 200–800 nm region. It contains bands of different widths, corresponding to different nickel-support interactions. The undoped Ni/MgAl material presents a broad featureless band centered at ~250 nm, which is associated to O2p→Al3sp electronic transitions [19]. When increasing the K content above 1 wt%, this band splits in two at 252 and 212 nm, respectively. The splitting strongly suggests that potassium stimulates spinel inversion, modifying the acid-base properties of the support. Spinel inversion is the partial interconversion of A and B sites in the spinel structure, giving rise to oxygen vacancies. The simultaneous addition of K and Ni precursor salts to the spinel support in the synthesis is necessary for Ni ions to participate in the defect creation process. The absorption spectrum also suggests that some nickel ions are incorporated in the spinel structure. Both doped- and undoped-Ni materials present broad bands at 380, 407, 650 and 737 nm, likely associated with Ni2+ ions in tetrahedral and octahedral coordination sites [20,2122]. The low energy bands at 652 and 737 nm (ascribed to 3 T 1 (F)→ 3 T 2 (P) and 3 T 1 (F)→ 1 T 2 , 1 E(D) transitions, respectively) are indicative of tetracoordinate nickel species. This points to the incorporation of nickel ions in the spinel structure (NiAl2O4), since only hexacoordinated Ni2+ ions can be accommodated in the rock salt-like structure of NiO. On the other hand, hexacoordinated Ni2+ ions in NiO or Ni1−xMgxAl2O4 may account for the bands at 380 and 407 nm. The NiO UV–Vis absorption spectrum is dominated by the valence band (VB) to conduction band (CB) transition at 355 nm, whereas the d-d transition in the visible region is hardly visible [23], in agreement with absorption coefficients previously reported (Table S2). The small shoulder at 380 nm is associated with trapped holes associated with Ni2+ species.The deposition of Ni and K introduces defects (oxygen vacancies) in the spinel support. It was reported that the MgAl2O4 indirect bandgap is 7.8 eV. However, the presence of oxygen vacancies results in interband energy states and an electron transition at 4.75 eV (260 nm) [24]. Solid non-stoichiometry, disorder or impurities may result in complex defects, including the generation of interstitial oxygen atoms and spinel inversion. These defects originate an optical transition at 5.3 eV (230 nm) [25]. Fig. 1B indicates the estimated band gap values for all the calcined catalysts. Notably, the band gap decreases from 3.37 eV in Ni/MgAl sample to 3.24 eV in the case of 5 K-Ni/MgAl. This indicates that potassium doping favors the formation of oxygen vacancies. Wrobel et al. [26] demonstrated that potassium may be incorporated in NiO nanostructures, resulting in hole-doped materials. The presence of defects may result in coordination numbers below six for nickel ions: tetracoordinated [NiO4] polyhedral units (square-planar in K2NiO2 and tetragonal in K9Ni2O7) have been reported, the latter also containing Ni3+ ions [27]. Laporte forbidden d-d transitions of Ni2+ ions in K2NiO2, characterized by a broad band at ~625 nm [28], cannot be excluded.The materials prepared may change when they are used under the high temperature and reducing conditions of a dry reforming reactor. To investigate this, in situ X-ray diffraction experiments were conducted while exposing the catalysts to increasing temperature under a reducing flow of 5% H2/Ar ( Fig. 2). The results indicate that the phases present initially in the materials are spinel phases (MgAl2O4 and likely some NiAl2O4), nickel oxide (NiO with bunsenite structure), and traces of aluminium oxide. In addition, potassium-rich samples (3 wt% and 5 wt%) also present some potassium nitrate (KNO3). The small amounts of potassium nitrate present in the starting materials are rapidly reduced at 400–500 °C (Fig. 2). Notice that these nitrate species had not decomposed during the oxidative calcination in the preparation of the materials. Virtually no K2O was detected in any measurement.NiO reduction starts in all samples at 600 °C, which is accompanied by an increase in reflections characteristic of metallic nickel at 2θ values of 44.5° and 51.8° (Fig. 2). The potassium loading does not seem to affect the onset temperature for the NiO reduction process. However, as the temperature is increased, the Ni peaks become sharper and more intense in the 3 K-Ni/MgAl and 5 K-Ni/MgAl. This suggests that metallic nickel sintering may be stimulated by high potassium loading. In agreement with this, Mross reported that the incorporation of alkali ions in the NiO lattice diminishes the activation energy of recrystallization and provokes the sintering of the metal particles [29]. Also, El-Shobaky et al. [30] observed that the incorporation of monovalent ions in the NiO lattice induces the formation defects, facilitating the diffusion of ions in the outermost surface layers and the agglomeration of nickel particles.Potassium may favor the mobilization of species from the spinel at high temperature. As shown in Fig. 2, in the samples with high loads of potassium, the diffraction peak at 2θ = 42.7°, associated with NiO/MgO phases, persists even at 800–900 °C. This suggests that the incorporation of potassium in the catalyst may promote the migration of Mg2+ and Ni2+ ions from the bulk spinel lattice to the outer surface layers. We believe that, at temperatures above 600 °C, potassium interacts with nickel particles and helps with the reduction and mobilization of NiAl2O4 and MgAl2O4, leading to higher final amounts of metallic Ni and NiO/MgO oxides. Also note that, in the starting materials, the relative amount of NiO also increases with the potassium loading, suggesting that potassium may exert a similar mobilizing effect under the oxidant calcination conditions.Remarkably, in potassium-rich materials (3 wt% and 5 wt%), a new crystalline phase is detected under high-temperature, reducing conditions. As the potassium loading is increased, a new diffraction peak appears at 32.8° at temperatures above 700 °C. The diffractograms at high temperature did not show any crystalline diffraction related to known potassium species, suggesting that K+ cations could have entered the NiO crystal lattice to form a new Ni-O-K structural phase. In more detail, Fig. 3 shows the diffractograms recorded at 800 °C during in situ reduction for all catalysts. All the samples lack the peaks expected for the (220), (311), (222), (200), and (400) reflection planes of the MgAl2O4 spinel lattice. Moreover, the (111) reflection of well-crystallized metallic nickel (2θ = 44.5°) is observed in all samples. High potassium loads (3 wt% and 5 wt%) lead to a new diffraction peak at an angle of 32.8°. The assignment of this peak is unclear: it might be attributed to the formation of K-Al-O phases or K-doped MgAl2O4 spinel. However, the large ionic radius of potassium would hamper the formation of this type of K-Al-O phases. Considering the UV–Vis analysis discussed above, we tentatively assign these peaks to the formation of nickel potassium oxide composite layers (Ni-O-K).Praliaud et al. [31] suggested that K is mainly present in the K+ form and that a Ni-O-K surface complex is formed on the surface of nickel particles. The addition of alkali metals on the NiO surface has been extensively studied, and the formation of Ni3+ oxidation states in these phases has been proposed to occur by the following reaction: (2) K2O2 + 2 NiO → 2 KNiO2 The formation of potassium nickelate phases leads to the stabilization of nickel in the trivalent formal oxidation state. Kim et al. [32] described the KNiO2 structure illustrated in Fig. 3. Interestingly, Ni3+ occupies pyramidal sites located between potassium layers with an adequate K-K distance to accommodate the nickel cations, minimizing the electrostatic repulsions between potassium ions. Our results suggest that a new phase, designated as Ni-O-K, is formed during the reduction treatment, which may be compatible with proposals like KNiO2.XPS measurements were conducted to investigate the electronic state of nickel both in the presence and absence of potassium and confirm the existence of a new phase designated as Ni-O-K that contains nickel in state trivalent. Fig. 4 A includes the Ni 2p3/2 spectra recorded for both Ni/MgAl and 5 K-Ni/MgAl samples without treatment and after in situ reduction at 800 °C. The two samples show a main peak at 854.1 eV with the corresponding shake-up satellite peak at 860.2 eV. These binding energies are typical of nickel oxide (NiO) disperse on the catalyst surface [33]. Apparently, both unpromoted and K-doped samples present similar Ni 2p3/2 spectra. However, after reduction treatment, notable changes become visible in both samples. XPS analysis of the reduced Ni/MgAl sample indicates the formation of metallic nickel (850.3 eV) and the presence of a weak peak at 857.1 eV, which is related to Ni2+ cations in NiAl2O4 spinel. This high binding energy stems from the strong metal-support interaction [34]. By contrast, the reduced 5 K-Ni/MgAl displays a broad peak at 857.1 eV, while the peak assigned to Ni0 is very faint. The broad peak may be affected by the complex main line splitting due to multiplet contributions in oxides, although its high binding energy and broadening clearly indicate that the oxidation state of nickel is formally Ni3+ [35,36]. Different authors have postulated that potassium doping inhibits the reduction of nickel and shifts the binding energy to higher values due to the presence of ionic potassium and the formation of Ni-O-K complexes [31]. Carley et al. [37] investigated the interaction between potassium and nickel single-crystal (100) surface by XPS measurements, evidencing the formation of Ni3+ unambiguously after annealing the surface at 600 K. The authors suggested the formation of species between nickel and potassium where the chemical state of nickel is formally + 3. These observations are in good agreement with our results.The contribution from multiplet splitting, satellite peaks and plasmon loss structures often complicates the interpretation of XPS results, particularly for nickel species in different surface environments. Biesinger et al. [38] reported that additional insights can be obtained from the Ni LMM Auger peak-shape. Fig. 4B shows the Ni LMM Auger spectra for the fresh and reduced samples. The peak-shapes observed in both fresh Ni/MgAl and 5 K-Ni/MgAl samples are typical of NiO species, while the reduced Ni/MgAl sample presents a LMM Auger peak-shape characteristic of metallic Ni. By contrast, the reduced 5 K-Ni/MgAl sample shows a Ni LMM Auger spectrum with a significant broad peak which can be fitted to NiOOH oxyhydroxides (i.e. Ni3+), constituted by stacking faults with intercalated alkali cations [38,39]. Moreover, the confirmation of formation of a Ni-O-K oxide phase was also verified from the XPS spectra recorded in the K2p region for both fresh and reduced 5 K-Ni/MgAl samples. As illustrated in Fig. S2, the deconvolution of the K2p XPS spectra show that the reduced sample has two different potassium phases. This observation can be directly related to the formation of the Ni-O-K layer.Based on these observations, we suggest that potassium interacts with nickel surface particles forming a core surrounded by a Ni-O-K phase (Ni@Ni-O-K). This phase likely presents an alkali-nickelate-type structure with nickel is stabilized in oxidation state + 3. In previous work, we demonstrated that CO is hardly adsorbed on K-promoted nickel catalysts [14]. Ni-O-K sites are accessible to hydrogen adsorption but not to CO adsorption. As discussed below, the importance of these sites in coke gasification is crucial for developing more stable dry reforming catalysts.To gain additional insights on the origin of the new Ni-O-K phase, we conducted in situ DRIFT spectroscopy during a reduction experiment. Fig. 5 shows the evolution of the IR spectra recorded during the H2-TPR reaction from room temperature to 750 °C for both Ni/MgAl and 5 K-Ni/MgAl samples, respectively. It also displays the evolution of the main gaseous species followed by MS as a function of time-on-stream and temperature for both samples.With regards to the unpromoted Ni/MgAl sample (Fig. 5a), a complex set of bands attributable to polydentate (1510–1407 cm-1), bidentate (1560–1360 cm-1), and monodentate (1547–1373 cm-1) carbonate species were initially detected in the 1600–1300 cm-1 region [14,40]. The bands at 1637 cm-1 and 3467 cm-1 are characteristic of physisorbed water. Early in the TPR, the release of physisorbed water leads to a band at 3730 cm-1, characteristic of isolated hydroxyl species bound to tetrahedral Mg2+ cations on the MgAl2O4 surface [41]. The thermal stability of carbonate species increases from monodentate to polydentate species, and the most thermostable carbonates are only fully removed above 500 °C (Fig. 5b). All carbonate surface species were entirely removed by 750 °C. In agreement with this, the products detected by MS (Fig. 5c) include the release of H2O and CO2 in two steps at approximately 200 and 400 °C, suggesting the desorption of two types of labile carbonates. Above 500 °C, reduction of the most thermostable carbonate-like species was clearly accompanied by the formation of CH4 and CO and the consumption of hydrogen.On the other hand, the IR evolution for the 5 K-Ni/MgAl sample was significantly different to that of the unpromoted sample (Fig. 5d). Firstly, the spectra of 5 K-Ni/MgAl show that potassium addition completely neutralizes the hydroxyl surface species, in agreement with our previous work [14]. Consequently, multiple overlapping absorption bands were detected in the 1800–1200 cm-1 range at room temperature. At temperatures above 600 °C, most of these bands disappear and only the most thermostable species remain on the surface (Fig. 5e). These correspond to bulk polydentate or highly ionic carbonate species (1720–1407–1444 cm-1), although monodentate carbonate species formed on very strong basic Mg-O-K sites (1584–1323 cm-1) are also thermally stable [14,41,42]. Note that the presence of carbonate species on the samples likely stems from contact with CO2 in air during preparation of the material (calcination), and this would be favored by the presence of basic potassium sites. On the other hand, not only labile carbonate species are removed during the reduction, but also nitrate species, observed in the 1550–1350 cm-1 region, disappear at temperatures between 500 and 600 °C. In terms of gases evolved during the reduction of 5 K-Ni/MgAl (Fig. 5f), an appreciable CO2 and H2O release is first observed without hydrogen consumption, indicating that labile carbonates are decomposed into CO2 and water below 400 °C. By contrast, an important hydrogen consumption was detected between 350 and 550 °C, peaking at 450 °C. This hydrogen depletion matches the production of NO, CO, and H2O, corresponding to the reduction of nitrates and carbonates in this temperature range. Increasing the temperature further resulted in higher CO release, indicating that the reduction of carbonates by the reverse water gas shift reaction (RWGS) was accelerated above 600 °C. Notably, with this material, CH4 was not detected at any temperature. Previous mechanistic studies suggested that adsorbed CO is an important reaction intermediate of CO2 methanation [43]. The presence of Ni-O-K sites may inhibit CO adsorption and the subsequent production of methane.A closer inspection of the IR spectra during reduction of 5 K-Ni/MgAl at 250–650 °C ( Fig. 6) reveals the appearance of two bands at 2038 and 2158 cm-1 accompanied by the simultaneous decrease of the bands associated with bidentate carbonate species (1540 and 1372 cm-1) and potassium nitrate-like species (1565, 1418, 1510 and 1363 cm-1) [44]. The bands at 2038 cm-1, observed between 250 and 450 °C, can be assigned to CO linearly adsorbed on small particles of metallic nickel, whereas the band at 2158 cm-1 had been assigned to linear NCO adsorbed forming cyanate-nickel complexes species [45]. However, we consider it is more reasonable to assign this band to nitrosyl(carbonyl) complexes formed on reduced nickel sites, given the CO and NO released during the reduction. The formation of these nitrosyl(carbonyl) intermediates deserves attention since it allows to understand the formation of Ni-O-K sites.The interaction between NiO particles and adsorbed potassium species is accompanied by an initial reduction of small particles of nickel oxide to metallic nickel, which begins at temperatures below 250 °C. The driving force of this step is the reduction of potassium carbonate species to generate highly stable alkali peroxide species and CO-linearly adsorbed on metallic nickel sites [37]. Under reducing conditions at 300–500 °C, potassium nitrate species are reduced, and a nitrosyl(carbonyl) complex is formed on reduced nickel sites. The thermal decomposition of this nitrosyl(carbonyl) intermediate occurs around 500 °C releasing CO and NO simultaneously. Presumably, the formation of the formally Ni3+ oxidation state evidenced by XPS can be attributed to the reduction of the alkali peroxide previously formed in the low temperature reduction. This step occurs at temperatures above 550 °C in which the oxygen of alkali peroxide is expended at high temperature to form trivalent alkali-nickelates (Ni-O-K sites). Fig. 6 sketches the plausible mechanism of formation of Ni-O-K sites. A similar process has also been proposed on the basis of FTIR studies of CO and adsorption experiments performed over analogous catalysts based on iron and potassium [46]. The authors hypothesized a similar mechanism to explain the formation of new phases (KFeO2 and similar) constituted by oxidized iron and potassium in close contact. In agreement with the XRD and XPS results above, it is reasonable to assume that reduction of 5 K-Ni/MgAl catalyst at high temperature leads to the formation of an active phase composed of a metallic nickel core covered by Ni-O-K sites.The modelling of H2 temperature-programmed reduction (TPR) allows to evaluate the impact of potassium on the reducibility of the different materials. The H2-TPR profiles for the materials are shown in Fig. 7. All samples display a broad reduction peak with multiple underlying components across the range 600–900 °C. We modelled the TPR process by considering three first-order reduction processes in an unsteady reactor model, resulting in a system of algebraic and ordinary differential equations (see Supporting Information for details). The resulting components fitted from the model are displayed in Fig. 7 and the corresponding parameters are included in Table S3. The three-component model recapitulates the TPR signal from the different materials very successfully and allows estimating the apparent activation energy for each reduction step (Fig. 7).Potassium interacts with nickel and enhances its reducibility in the Ni/MgAl2O4 catalysts, as shown in Fig. 7. The TPR results indicate that the different reduction processes are initiated at lower temperatures as potassium loading is increased. This points to an interaction between nickel and potassium species (Fig. 7). The intermediate temperature component β (550–750 °C range) is associated with the reduction of NiO species with a moderate interaction with the support, whereas the component γ (> 750 °C) is assigned to the reduction of NiOx complex species with a very strong metal-support interaction [47,48]. In fact, this high-temperature component can be associated with Ni2+ ions migrated into the MgAl2O4 matrix forming the non-stoichiometric nickel aluminate spinel (NiAl2O4) phase discussed above. Notably, the introduction of potassium above 3 wt% significantly increases the activation energy for the reduction of the γ component, indicating that potassium interacts with the NiAl2O4 phase and makes its reduction more temperature sensitive. The results also support the existence of a weaker interaction between K and NiO, with the activation energy for the reduction of NiO increasing slightly with K loading.On the other hand, a reduction component α is observed below 500 °C, and its area under the curve notably increases with the potassium loading. Some authors have ascribed this reduction peak to bulk nickel oxide crystallites presenting very weak interactions with the support [49,50], and accordingly, it would indicate that a large amount of nickel should be reduced in the 5 K-Ni/MgAl sample. However, this would be inconsistent with the characterization results above. Thus, to understand the origin of this peak, a MS spectrometer was coupled to analyze on line the gases released during the TPR process. As shown in Fig. S3, an intense peak of H2 consumption around 420–430 °C was observed along with the formation of CO (m/z= 28), NO (m/z= 30) and H2O (m/z= 18) in the samples with high potassium load. This indicates that the α component in the TPR results from the reduction of potassium nitrate and carbonate species. In addition, CO2 (m/z= 44), CH4 (m/z= 15), and CO (m/z= 28) were also detected during its reduction, and they decreased with the potassium loading (Fig. S3).In summary, we conclude that Ni-O-K sites are not generated directly from the reaction of potassium oxide and nickel oxide on the K-Ni/MgAl2O4 catalysts but, instead, require an intermediate complex formed from potassium nitrate and carbonate species. These data provide additional evidence for the Ni-O-K phase. Ni-O-K sites seem available for hydrogen adsorption but not for CO adsorption.To explore the functional implications of this new Ni-O-K phase, we investigated the catalytic performance of all prepared samples in dry reforming at 650 °C for 48 h, using different partial pressures of methane and CO2 (20, 30 and 40 kPa) and a CH4/CO2 molar ratio equal to 1. Fig. 8 shows the CH4 and CO2 conversion over all the prepared catalysts against time-on-stream for the three partial pressures studied. Notably, Ni/MgAl and 1 K-Ni/MgAl catalysts suffered a drastic deactivation after 6 h when the reaction was performed at high partial pressures (30 and 40 kPa) of reactants. This was due to the rapid accumulation of coke, which even led to reactor plugging. It is well known that coke formation is related to CH4 cracking and that dissociative adsorption of CHx* species is a rate-determining step sensitive to CH4 partial pressure. By contrast, the catalysts promoted with high potassium loads showed stable CH4 and CO2 conversions during 48 h, even at high partial pressures of both reactants. This clearly indicates that potassium mitigates the carbon deposition or accelerates the gasification of carbon deposits. Fig. S4 shows the evolution of the H2/CO molar ratio over time during the tests performed at 20 kPa for all catalysts. The Ni/MgAl catalysts present a gradual increase of H2/CO molar ratio over time, indicating that the unpromoted catalyst deactivates rapidly. Similarly, the Ni-1 K/MgAl catalyst shows an initial increase in the H2/CO molar ratio, but it then stabilizes. This suggests that small amounts of K promote an equilibrium between the Boudouard reaction and carbon gasification, slowing catalyst deactivation. On the other hand, the H2/CO ratios remain stable over time for the 3 K-Ni/MgAl and 5 K-Ni/MgAl catalysts. Both samples show H2/CO molar ratios below the stoichiometric value of 1, pointing to stable carbon gasification and RWGS reactions. Consequently, no deactivation was observed for both potassium-rich catalysts. In agreement with these results, it has been reported that potassium promotes the RWGS reaction as it activates the CO2 molecules via carbonates and subsequent reduction into CO [51,52].To quantify the amount of nickel exposed in the different materials, H2 chemisorption experiments were carried out ( Table 1). The results indicate that high potassium loads reduce the amount of exposed nickel sites. This may be due to i) the coverage of Ni sites by formation of a Ni-O-K layer and ii) the somewhat increased particle size (Fig. S8 and Table 1) upon K loading. The catalyst average particle sizes are slightly larger for samples with high potassium load. Some authors have reported an increased metal particle size upon addition of promoters such as Mg, K or Ce [53]. This suggests that potassium may favor the formation of larger nickel particles as a result of a weaker interaction with the MgAl2O4 support, in agreement with our TEM results (Fig. S5).We also investigated the effect of potassium on the catalyst turnover frequency (TOF), which expresses the activity of catalysts in terms of moles transformed per time unit and per mol of exposed nickel. Apparent TOF values were estimated from the run tests at 20 kPa using initial reaction rates and the dispersion of nickel estimated from H2 chemisorption (Table 1). Notably, the samples containing 3 and 5 wt% K2O displayed superior TOF values for both reactants. This indicates that, although the number of nickel sites exposed is reduced upon K addition, the C-H bond cleavage and the CO2 activation in carbon gasification are greatly facilitated, consistent with the formation of a new Ni-O-K active phase.In Fig. 9, the CH4 and CO2 consumption rates are compared with the H2 and CO production rates, respectively, in terms of Ni metal surface exposed. As can be observed in Fig. 9A, CH4 consumption and H2 production rates are directly correlated with the Ni surface area exposed and thus Ni/MgAl and 1 K-Ni/MgAl catalyst, in which higher fraction of metallic nickel sites are exposed, favored these reactions. Likewise, the addition of potassium decreases the amount of exposed nickel sites due to the formation of Ni-O-K layer and both methane consumption and H2 yield are less favored. On the other hand, Fig. 9B shows that CO2 consumption and CO formation rate are also nickel structure-sensitive reactions. Remarkably, in the absence of potassium, the CO produced is significantly lower than the CO2 consumed. When potassium is added, the CO produced matches the CO2 consumed. This suggests that a sizeable proportion of CO produced on the unpromoted catalyst is dissociated to C* and O* species whereas, in the presence of Ni-O-K, CO does not adsorb and remains in the effluent. Therefore, the addition of potassium avoids the CO dissociation on nickel sites and thus decrease the accumulation of carbonaceous deposits. Moreover, this effect could open the door to enhanced low-temperature RWGS catalysts since CO dissociation favors the methanation reaction against reverse water gas shift reaction [54].Finally, the amount and nature of the carbon deposits after dry reforming of methane were studied as a function of potassium loading. The TPO profiles obtained for all samples are displayed in Fig. 10. The m/z= 32, m/z= 44, and m/z= 28 signals were chosen to analyze the evolution of O2, CO2, and CO, respectively. The area under a TPO profile is proportional to the amount of deposited carbon, and it is clearly seen that the 5 K-Ni/MgAl catalyst has the lowest amount of carbon deposited. As observed in Fig. 10, Ni/MgAl and 1 K-Ni/MgAl catalysts show maximal CO2 production at 613 °C, which is associated with the oxidation of highly structured carbon species such as whiskers or carbon filamentous. This type of carbon is the main one responsible for pore blocking and metal particle encapsulation [10,55]. Note that there is an abrupt decline in the CO2 signal on both Ni/MgAl and 1 K-Ni/MgAl catalysts at 716 °C and 750 °C. The amount of carbon deposited in these samples was so high that all oxygen of stream was fully consumed, and the CO2 produced then reacted as an oxidant through the reverse Boudouard reaction (C* + CO2 → CO). This explains the concurrent formation of CO gas. The TPO profiles obtained for 3 K-Ni/MgAl and 5 K-Ni/MgAl catalysts show that the carbon oxidation processes occur at lower temperatures. The TPO profile of 3 K-Ni/MgAl catalyst peaks at 593 °C, consistent with the oxidation of carbon with a higher graphitic degree, probably small filamentous species [56]. The oxidation step occurring around 330 °C is ascribed to the oxidation of amorphous carbon, and this is the only process detected on the 5 K-Ni/MgAl spent catalyst. Figs. S6 and S7 shows the structural analysis obtained by XRD and Raman spectroscopy, respectively, for all the spent catalysts. The XRD pattern (Fig. S6) obtained for Ni/MgAl spent catalyst present an intense and well-defined peak at 2Ɵ = 26.5°, associated with graphitic carbon species (JPDS 00-025-0284). This peak was less intense in the spent K-promoted catalysts, and it was absent in the Ni-5 K/MgAl sample. On the other hand, potassium loading increased the intensity of peaks associated with Ni0, MgAl2O4 and Mg(Ni)O phases compared to the unpromoted sample.The Raman spectra of Ni/MgAl, 1 K-Ni/MgAl and 3 K-Ni/MgAl catalysts, Fig. S7, are characterized by two main features at 1349 and 1578 cm-1, ascribed to the vibrational modes D and G of carbon species, respectively. The D-band is characteristic of structurally disordered carbon while the G-band corresponds to C-C vibration stretching of structured carbon, such as graphite [57]. The D-to-G intensity ratio (I(D/G)) is thus a valuable parameter to characterize the carbon disorder degree [58]. The I(D/G) ratios estimated for the spent catalysts follows the sequence 3 K-Ni/MgAl > 1 K-Ni/MgAl > Ni/MgAl (Fig. S7), indicating that potassium loading decreases the graphitization degree. These bands were absent in the 5 K-Ni/MgAl spent sample.The morphology of spent catalysts was studied by SEM analysis (Fig. S8). As depicted in Fig. S4, the SEM micrographs show that the unpromoted catalyst becomes covered of carbon species with whisker or filamentous structure. Meanwhile, the amount of whisker carbon and the diameter of the filamentous decreases notably with the increment of potassium loading becoming wholly absent for 5 K-Ni/MgAl catalyst. These observations are consistent with the XRD and Raman spectroscopy results.Taken together, the results reveal that, while the unpromoted catalyst became fully covered by carbonaceous deposits, increasing the loading of potassium led to less carbon deposition, which was also less graphitic in nature. Remarkably, deposits were fully prevented on the 5 K-Ni/MgAl catalyst, and its diffractogram coincides with the pattern of the reduced catalyst prior to reaction (Fig. 2), confirming an exceptional improvement in stability accompanying the new Ni-O-K phase.The efficient conversion of biogas to syngas by dry reforming is a very promising route to produce liquid fuels, but low catalytic stability prevents its establishment as an industrial process. In this work, Ni-based catalysts promoted with potassium were tested in the dry reforming of methane and were exhaustively characterized to understand the key role of potassium in carbon deposition suppression. By means of different characterization techniques, we have established that, in the presence of 5 wt% of potassium, the nickel particles form a core surrounded by a Ni-O-K inter-layer (Ni@Ni-O-K) during the reduction of the catalyst. Likely, this layer presents a structure type alkali-nickelate (KNiO2), in which nickel is stabilized in oxidation state + 3. The Ni-O-K phase formation induces essential changes in the electronic properties of nickel. The presence of Ni-O-K sites leads to coke-resistant catalysts with excellent activity and stability. Specifically, these new sites do not catalyze the dissociation of CO, thus avoiding the formation of methane and coke and greatly enhancing the yield of syngas and the catalytic stability. The study also provides the first insights on the formation process of the Ni-O-K phase, providing a new direction to design high-performance dry reforming catalysts for sustainable syngas production.Textural properties; XPS K2p; TPR modeling details; SEM/TEM results; H2/CO ratios results for DRM; Characterization of spent catalysts (XRD, Raman spectra and SEM); Figs. S1–S8 and Tables S1–S3. L. Azancot, V. Blay: Conceptualization, Methodology L. Azancot, L.F. Bobadilla: Data curation, Writing – original draft preparation. V. Blay, R. Blay-Roger, L. Azancot, A. Penkova, M.A. Centeno: Visualization, Investigation. L.F. Bobadilla, J.A. Odriozola : Supervision. L.F. Bobadilla, L. Azancot: Writing – review & editing, J.A. Odriozola, M.A. Centeno : Funding acquisition.The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Financial support for this work has been obtained from the Spanish Ministerio de Economía y Competitividad – MINECO (RTI2018-096294-B-C33) co-financed by FEDER funds from the European Union and the Universidad de Sevilla-Junta de Andalucía Program under contract US-1263288. Lola Azancot acknowledges the MINECO for her associated Ph.D. fellowship (BES-2016-0077475).Supplementary data associated with this article can be found in the online version at doi:10.1016/j.apcatb.2022.121148. Supplementary material .

Liquid fuels produced via Fischer-Tropsch synthesis from biomass-derived syngas constitute an attractive and sustainable energy vector for the transportation sector. This study focuses on the role of potassium as a promoter in Ni-based catalysts for reducing coke deposition during catalytic dry reforming. The study provides a new structural link between catalytic performance and physicochemical properties. We identify new Ni-O-K chemical states associated with high stability in the reforming process, evidenced by different characterization techniques. The nickel particles form a core surrounded by a Ni-O-K phase layer (Ni@Ni-O-K) during the reduction of the catalyst. This phase likely presents an alkali-nickelate-type structure, in which nickel is stabilized in oxidation state + 3. The Ni-O-K formation induces essential changes in the electronic, physical, structural, and morphological properties of the catalysts, notably enhancing their long-term stability in dry reforming. This work thus provides new directions for designing more efficient catalysts for sustainable gas-to-liquids processes.

Lignin is an abundant biomass source that can be converted into value-added aromatic platform chemicals. 1 To use biomass to its fullest, the challenges involved in valorizing lignin need to be overcome. 2–5 Conventional lignocellulose delignification methods, such as Kraft and organosolv pulping, require the complete breakdown of the C–O bonds in a series of steps. 6–8 However, the ability to deconstruct lignin directly from raw biomass has transformed the conventional concept of the biorefinery by capturing high-value products from lignin in the first step, 9 a process that is particularly effective under reductive conditions. 3 , 10–12 The lignin is depolymerized and selectively converted to monomers with high retention of carbohydrates in the pulp. 10 , 13 , 14 The monomers obtained are valuable platform chemicals with a range of applications. Further use of the remaining cellulose and hemicellulose provides additional value-added chemicals. 15–17 Using noble metal catalysts (e.g., Rh, Ru, Pd, Pt), the yields of lignin monomers can reach close to the theoretical maximum. 18–20 However, these noble metal catalysts are expensive and can be susceptible to poisoning by CO or coking. 21 Efforts to maximize the efficiency of these noble metal catalysts with improved selectivity has led to the development of atomically dispersed heterogeneous catalysts. 22–24 All of the reaction steps take place at single-atom sites and, compared with metal nanoparticles (NPs), the reaction kinetics with single-atom catalysts are rate limited by the low concentration of H atoms available in the active atomic sites.Progress toward catalyst design to combine the activity of noble metals and low cost of earth-abundant metals is popular. 25 Some earth-abundant metals such as Ni have low energy barriers for both the dissociation of H2 and the diffusion of H atoms. 26 If H2 dissociation on Ni NPs and diffusion to active single noble metal atoms is facile, then the rate-limiting addition of H atoms to the substrate should be accelerated, improving the overall reaction kinetics and reducing the amount of the precious metal required. Such a single-atom alloy (SAA) concept has been described for a PdCu system in which facile hydrogen dissociation and spillover take place. 27 Because the dissociation of H2 and reaction sites on SAAs are decoupled, SAAs may not be confined to linear scaling relationships, exceeding the reactivity limit and selectivity of many catalysts.Despite numerous reports of SAAs, their application is mostly limited to catalytic process involving small substrates and in reactions such as C-C coupling, hydrogenation, and electrocatalytic processes. 22 , 28–33 There are only limited studies describing SAAs as catalysts for biomass transformations, with the focus on lignin model compounds as substrates. 34 , 35 Direct depolymerization of lignin using SAAs has not, to the best of our knowledge, been reported.Here, we describe a highly active catalyst for reductive lignin depolymerization based on Pt single atoms anchored onto Ni NPs supported on C (denoted as Pt1Ni/C, where Pt1 represents single Pt atoms). Using this catalyst, a yield of lignin monomers of 37% for birch sawdust was achieved under 5 MPa H2 in methanol (MeOH) at 200°C, which is significantly higher than that using single Pt atoms supported on active carbon Pt1/C or Ni NPs on active carbon Ni/C.The Pt1Ni/C catalyst was obtained by anchoring Pt atoms on Ni NPs supported on C (Ni/C) through galvanic replacement (Scheme 1 ). In the synthesis, Pt(acac)2 dissolved in toluene was added to a suspension of Ni/C in ethanol. C was chosen as the support as it is inexpensive and has a high surface area and cavities where H2 can be adsorbed. 36 Following washing with ethanol and hexane, the Pt1Ni/C catalyst was obtained as a black powder. Inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analysis of the Pt1Ni/C catalyst gives weight percentages of the Pt and Ni as 0.3 and 4.4 wt%, respectively. Thermogravimetric analysis (TGA) of Pt1Ni/C in air (Figure S4) confirmed that the loading of metal NPs is ∼5%. The specific surface area of Pt1Ni/C is 90.4 m2/g (Figure S5). Using transmission electron microscopy (TEM), the diameter of the Ni NPs was found to be ∼6.9 nm with a narrow size distribution (Figure S1), a size that is similar to Ni/C (Figure S7).The distribution of Pt on Ni/C within the Pt1Ni/C matrix was analyzed by high-angle annular dark-field scanning TEM (HAADF-STEM) (Figure 1 ), which confirms that the Pt atoms are highly dispersed (the isolated Pt atoms are manifested by brightness and marked by circles; Figure 1B). The Pt single atoms on the surface of the Ni/C particles were further characterized by extended X-ray absorption fine structure (EXAFS) analysis in the R space (Figures 2 A and S8; Table S1). The oscillation manners of the Pt L3-edge in the R space for the Pt1Ni/C differ from those of the Pt foil. The Pt−Pt bond at 2.76 Å was not observed in Pt1Ni/C. Compared to the Pt L3 edge of Pt foil probes by X-ray absorption near-edge spectroscopy (XANES) (Figure 2B), the adsorption edge for Pt1Ni/C is ∼11,570 eV, indicative of Pt(II) species. X-ray photoelectron spectroscopy (XPS) of Ni 2p and Pt 4f indicate that the majority of surface Ni and Pt species are in the 2+ oxidation state (Figure S2). 37 The X-ray diffraction (XRD) profile of the Pt1Ni/C catalyst shows a characteristic peak at 44° corresponding to (111) reflections of face-centered cubic Ni NPs (Figure S3), at the identical position observed in Ni/C, presumably as the low content of highly dispersed Pt does not influence the XRD pattern.The H2-temperature-programmed reaction (TPR) profiles (Figure 2C) indicate that alloying takes place in the Pt1Ni/C catalyst. With the addition of single Pt atoms, the reduction temperature of Pt1Ni/C (261°C) is shifted to lower regions compared with Ni/C (275°C). The adsorption of H2 on the surface of the Pt1Ni/C catalyst was investigated through H2 temperature-programmed desorption (TPD) measurements (Figure 2D). The first peak (at 425°C) in the Pt1Ni/C catalyst is at a much lower temperature than that observed for Ni/C (at 698°C), indicating that the Pt atoms provide low-barrier exit routes for H2 during the desorption process.The performance of the Pt1Ni/C catalyst was investigated in the depolymerization of lignin using birch sawdust as the substrate. For comparison, single Pt atoms and Ni NPs supported on C (Pt1/C and Ni/C) were prepared and applied in the same depolymerization reaction under identical conditions (Figure 3 ). The monomer yields are used as a measure of depolymerization efficiency of the catalysts. Under 5 MPa H2 in methanol at 200°C, the yield of total monomers with the Pt1Ni/C catalyst is 37% after 18 h (Figure 3, entry 1), quite close to the theoretical maximum monomer yield, which ranges from 44 to 56 wt%. 13 The yield of total monomers is significantly higher than that with the control catalysts (Figure 3, entries 2 and 3) and reported Ni/C, with monomer yields of 24% at 200°C in methanol. 38 Further analysis of the product distribution shows that a combined selectivity toward 4-n-propylsyringol (S) and 4-n-propylguaiacol (G) exceeds 90% within the monomer fractions, whereas 4-n-propanolguaiacol (G-OH) and 4-n-propanolsyringol (S-OH) accounts for <5% of the monomers (Figure 3, entry 1). Similar to other reported single-atom catalysts (i.e., Co1/C, 39 Ru1/ZnO/C, 40 and Pd1/CNx 41 ), the aromatic rings are preserved with the Pt1Ni/C catalyst even under more forcing reaction conditions, unlike pure NP catalysts that lead to ring hydrogenation. 42 , 43 Saturated compounds were not observed with the Ni/C catalyst at 200°C, although traces of saturated compounds were detected at 300°C. It has been shown that Pt NPs are more active than Ni NPs in hydrogenolysis. 44 As the hydrogenolysis of C–O bonds is highly metal dependent, 45 the overall yield in S and G may be attributed to the high activity of Pt atoms. Note that in the absence of the metallic sites the monomer yield is very low (Figure 3, entry 4).After separating the liquid products and drying the solid residue of birch sawdust and catalysts, 0.1 g birch sawdust was added to perform the recycling test. The decrease in the activation of the catalyst may be caused by coking on the surface of the catalyst (Table S2).The mechanism of reductive fractionation involves solvolysis of the C–O bonds, with the catalyst hydrogenating the reactive intermediate products generated, preventing re-polymerization. 46 , 47 Based on the similar temperatures used, both Ni NPs and single Pt atoms contribute to H2 dissociation (Figure 2C). Compared with Pt1/C, which is less active with hydrogen dissociation, both Ni NPs and single Pt atoms in Pt1Ni/C served as active sites in H2 dissociation and adsorption of H atoms, so that a more abundant amount of H atoms can be produced on the surface of Pt1Ni/C. Compared with Ni/C, the better performance of Pt1Ni/C is due to the lower hydrogen binding energy of Pt atoms than Ni atoms. The H2-TPD analysis shows a large decrease in the H2 desorption temperature of the Pt1Ni/C catalyst (425°C) (Figure 2D), compared with that of Ni/C (698°C), indicating that the single Pt atoms serve as active sites. Moreover, due to the single-atom nature of the Pt in the Pt1Ni/C, the aromatic structure of the phenyl rings of lignin monomers are preserved without further hydrogenation. Since ring hydrogenation requires coordination of the aromatic ring over a trimetal face, NP catalysts would lead to the hydrogenation of aromatic rings, 42 , 43 while single Pt atoms can achieve hydrogenolysis of the C–O bonds without the hydrogenation of the phenyl ring. 48 As such, Pt single atoms played a pivotal role in Pt1Ni/C in enhancing the catalytic activity while keeping the high selectivity in the lignin depolymerization, a benefit that cannot be achieved by using Ni/C only. Other reasons for the high activity, such as support effects, cannot be excluded.The reaction conditions were optimized to obtain monophenolic compounds in higher yields (Figure 3). The yield of monophenolic compounds increased from 12 wt% at 150°C to 43 wt% at 300°C at a H2 pressure of 5 MPa in methanol after 18 h (Figure 4 A), with high selectivity to S and G (>90%, Figure 4B). At lower H2 pressures (and in the absence of H2), G-OH and S-OH are preferentially formed instead of G and S (Figure 4C).A comparison of product distributions in water, methanol, ethanol, 1-propanol, 1-butanol, and ethylene glycol (Figure 4D) shows that solvent has a remarkable impact on the monomer yield as well as the product distribution, as observed elsewhere. 49 , 50 The solubility of lignin in different solvents has been extensively studied, 51 with the solubility in ethylene glycol being the highest followed by methanol, ethanol, 1-propanol, 1-butanol, and H2O. 52 The monomer yields basically decrease with the decreasing solubility of lignin in the solvent. As in our case, a comparably lower yield of 32% in the pure ethylene glycol was obtained than that from methanol, probably due to the high viscosity and low solubility of H2 in ethylene glycol.We describe a selective hydrogenation catalyst in which monodispersed Pt atoms are anchored on the surface of Ni NPs supported on active carbon. The Pt1Ni/C catalyst affords monophenolic compounds in a 37 wt% yield at 200°C, which is close to the theoretical maximum yield. The remarkable activity and selectivity of the Pt1Ni/C catalyst may be attributed to the synergistic effects between the Ni NPs and the single Pt atoms.C powder (Vulcan XC 72R, 1 g) was mixed with 50 mL 5 M nitric acid at 80°C and stirred for 18 h. The solids were separated by centrifugation and were washed with distilled water until constant pH was achieved. The pre-treated activated carbon (200 mg) was dispersed in ethanol (20 mL) under vigorous stirring at room temperature. To the resulting suspension, a solution containing Ni(NO3)2·6H2O (50 mg, 0.2 mmol) in ethanol (5 mL) was slowly added and stirring was continued for 12 h at room temperature. The reaction mixture was further heated at 40°C under stirring until all of the solvent had evaporated. The remaining solid was heated to 400°C for 1 h under H2 in a tube furnace, to afford Ni/C.The Pt1Ni/C catalyst was prepared via galvanic replacement between Ni NPs and Pt(acac)2. Ni/C was dispersed in ethanol (50 mL), and the resulting suspension was heated at 50°C for 10 min. To the reaction mixture, a solution of Pt(acac)2 (6 mg, 0.015 mmol) dissolved in toluene (5.2 mL) was added slowly. After stirring for 6 h at 50°C, the solution was cooled to room temperature. Then, the sample was collected by centrifugation and washed with ethanol (3 × 10 mL) and hexane (3 × 10 mL). After drying in vacuum at 40°C for 24 h, the Pt1Ni/C catalyst was obtained as a black powder.C powder (Vulcan XC 72R, 1 g) was mixed with 50 mL 5 M nitric acid at 80°C and stirred for 18 h. The solids were separated by centrifugation and were washed with distilled water until constant pH was achieved. This pre-treated activated carbon (100 mg) was dispersed in water (20 mL) under magnetic stirring. K2PtCl4 aqueous solution (10 mL, 0.043 mg/mL) was slowly added to the resulting reaction mixture. After stirring at room temperature for 30 min, the solid was collected by centrifugation. The isolated solid was washed with water (3× 10 mL) and dried at 60°C in vacuum for 24 h. The Pt1/C was obtained as a black powder, and the HAADF-STEM images and elements mapping of Pt1/C are in Figure S6.Pt L3-edge XAFS data were recorded under fluorescence mode with a 32-element Ge solid-state detector at the SuperXAS beamline of Swiss Light Source (SLS) at the Paul Scherrer Institute (PSI, Villigen, Switzerland). The energy was calibrated according to the L3 absorption edge of pure Pt foil. Data analysis was performed using a standardized IFEFFIT package (including Athena and Artemis software). 53 The Pt1Ni/C powders were suspended in acetone and the resulting suspension was ultrasonicated for 1 h. Subsequently, the acetone suspension of NPs was deposited on a C film coated with Cu grid and then analyzed by TEM (FEI Talos, operated at 200 keV). XPS analysis were performed using a monochromatic Al Kα X-ray source of 24.8 W power with a beam size of 100 μm. XRD measurements were recorded in Bragg Brentano geometry on a Bruker D8 Discover diffractometer, equipped with a Lynx Eye XE detector, using non-monochromated Cu-Kα radiation.In a typical reaction, birch sawdust (0.1 g, size 0.25–0.5 mm) and Pt1Ni/C (0.02 g) in MeOH (5 mL) were added into a 100-mL stainless-steel batch reactor with a glass liner. The reactor was sealed, flushed with H2 3 times, and then pressurized to 5 MPa at room temperature. The reaction mixture was heated at 100°C for 3 h under stirring at 800 rpm, and then heated to 200°C. After 18 h, the autoclave was cooled in water and then depressurized.Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Paul Dyson (paul.dyson@epfl.ch).All materials generated in this study are available from the lead contact without restriction.All data generated in this study can be found in the article and supplemental information or is available from the lead contact upon request.The authors are grateful to the Swiss National Science Foundation and EPFL for financial support. L.P. acknowledges funding from the Swiss National Science Foundation under the Early Postdoc.Mobility Grant P2ELP2_195109. J.L. acknowledges funding from the European Union’s Horizon 2020 Research and Innovation program under the Marie Skłodowska-Curie Grant agreement no. 838686. The authors thank Dr. Maarten Nachtegaal and Mr. Urs Vogelsang for technical support at the SuperXAS beamline of SLS.All of the authors contributed to the design of the experiments and data analysis. L.C. performed the experiments, L.B. performed the HAADF-STEM tests, and J.L. performed the EXAFS measurements. L.C. and P.J.D. wrote the manuscript, and all of the authors discussed, commented on, and proofread the manuscript.The authors declare no competing interests.Supplemental information can be found online at https://doi.org/10.1016/j.xcrp.2021.100567. Document S1. Figures S1–S8 and Tables S1 and S2 Document S2. Article plus supplemental information

Due to the highly complex polyphenolic structure of lignin, depolymerization without a prior chemical treatment is challenging, and new catalysts are required. Atomically dispersed catalysts are able to maximize the atomic efficiency of noble metals, simultaneously providing an alternative strategy to tune the activity and selectivity by alloying with other abundant metal supports. Here, we report a highly active and selective catalyst comprising monodispersed (single) Pt atoms on Ni nanoparticles supported on carbon (denoted as Pt1Ni/C, where Pt1 represents single Pt atoms), designed for the reductive depolymerization of lignin. Selectivity toward 4-n-propylsyringol and 4-n-propylguaiacol exceeds 90%. The activity and selectivity of the Pt1Ni/C catalyst in the reductive depolymerization of lignin may be attributed to synergistic effects between the Ni nanoparticles and the single Pt atoms.

Triglyceride based biomass is an appealing alternative to produce transportation fuels as they are readily available, renewable [1] and its long chain usually contains carbons between 16 and 18 [2], which is within diesel range [3] and it results in high heating value [2]. However, vegetable oils cannot be directly used as fuels due to the high amount of oxygen and consequently engine incompatibility [3,4] and therefore, in order to be used as fuels, an upgrading process is needed [3,4]. The transformation of vegetable oils into transportation fuels can be achieved by cracking, transesterification and deoxygenation [4–7]. A main drawback of biodiesel produced by transesterification is its inferior quality due the high amount of oxygen, resulting in high viscosity, poor chemical stability, high pour point and high cloud point, which results in poor cold flow properties [1,4,5,7]. Even tough cracking of triglycerides yields a fuel similar to oil derived diesel [6], it is not a selective process due to the formation of a wide range of hydrocarbons [4] and there is reduction of energy content due to loss of carbon [8].An appealing alternative is the deoxygenation, producing a fuel similar to conventional diesel [5,7–9] and with an outstanding cetane number when compared to fossil-based diesel [1,4,10]. Moreover, this approach has superior removal of oxygen when compared to thermal cracking and it is less susceptible to coke formation, due to operation at high pressures [11]. Other advantages are absence of sulfur [12], the use of established structure for storage and distribution, and green diesel can be used in diesel engines pure or blended with petrol-diesel at any ratio [13]. The removal of oxygen in deoxygenation can be achieved by three main pathways: (i) hydrodeoxygenation, (ii) decarbonylation and (iii) decarboxylation [1,2,4,7,8,10,13–16]. In addition, one should note that there are not many studies of deoxygenation using solvent-free operating conditions [16]. Therefore, this factor is worth investigating and unlike previous work, the solvent-free approach results in a better understanding on how to design practical reactors for the production of sustainable hydrocarbons.Currently, many studies have investigated transition-metal phosphides as hydrotreating catalysts due to some advantages such as its high performance ascribed to ensemble and ligand P effects [7], affordable price, long lifetime [14,17], globular particles, enhancing site exposure [9] and resemblance with noble metals due to P incorporation into the MeOx framework [18]. Another advantage is its low activity for side reactions such as cracking and methanation [1]. Furthermore, it exhibits metallic and acidic properties [1], ascribed to Niδ+ (small positive charge functioning as Lewis acid site) and P-OH group (Brönsted acid sites, resulting from partial phosphate reduction) [10].Several studies have explored the use of nickel phosphides for the deoxygenation of biomass. From the reported supports, it is important to highlight the use of SiO2, Al2O3, TiO2 and CeO2 [9,19], from which it was concluded that SiO2 showed the higher activity and CeO2 the lower [9]. This has been attributed to the surface density of Ni sites, which result from interactions with the support, and differences in acidity and reducibility also impact the catalyst activity [9]. Even though the use of Al2O3 presents some advantages such as high mechanical strength, a drawback is the formation of AlPO4 and as consequence occurs the formation of the Ni12P5 phase [20,21]. Other relevant supports employed are zeolites such as H-Y [22], H-ZSM-5 [23] and H-β [24] and mesoporous materials such as MCM-41 [25] and SBA-15 [26]. The advantage of zeolites is their strong acidity, enhancing isomerization reactions, and as consequence the quality of the fuel produced [27]. On another hand, the use of mesostructured materials can enhance the hydrotreating activity [26]. Albeit there are many studies exploring the use of metallic phosphides, not many studies have reported the effect of the support on the deoxygenation of model molecules over nickel phosphide catalyst [7], therefore, this is worth investigating.Therefore, the aim of this work was the investigation of nickel phosphide catalysts supported on different structures (USY, H-ZSM-5 and Al-SBA-15) on solvent-free deoxygenation of oleic acid to produce diesel-like hydrocarbons. The performance of the catalysts was conducted on a batch reactor operated at 260–300 °C and 50 bar. The effects of the structure, reducibility, acidity and dispersion of the catalysts on the deoxygenation of oleic acid and the temperature were investigated.The commercial zeolites used were H-ZSM-5 (HCZP 90, Clariant, Si/Al = 50) and USY (HDT 9807, Cenpes/Petrobrás, Si/Al = 5.95). On the other hand, the Al-SBA-15 was synthetized as follows. The materials used were nickel nitrate hexahydrate (97%, Vetec), dibasic ammonium phosphate (99%, Acros Organics), oleic acid (90 wt%, Synth), nitric acid (65%, Synth), hydrochloric acid (37%, Anidrol), hydrofluoric acid (48%, Dinâmica), sulfuric acid (98%, Fmaia), isopropyl alcohol (99.5%, Synth), methanol (99.8%, Synth), hexane (98.5%, Anidrol), sodium hydroxide (97%, Anidrol), ammonium chloride (99%, Êxodo Científica), sodium chloride (99%, Nuclear), ammonium hydroxide (30%, Neon), aluminum isopropoxide (98%, Acros organics), tetraethyl orthosilicate TEOS (98%, Aldrich), triblock copolymer P123 (30 wt%, Aldrich), potassium bromide (>99%, Sigma-Aldrich) and pyridine (>99%, Synth).The synthesis of the support Al-SBA-15 (Si/Al = 10 at the gel) was performed as the literature [28,29]. Firstly, a solution was prepared with 0.85 g of aluminum isopropoxide, 8.5 g of TEOS, 10 mL of HCl 2 mol L−1 and it was stirred for 5 h. Another solution was prepared adding 4 g of P123 to 100 mL of HCl 2 mol L−1 and it was stirred for 5 h. Secondly, the first solution was added dropwise to the second one and the resulting gel was mixed for 20 h at 40 °C. Then, the pH was increased to 7.5 adding NH4OH dropwise. The mixture was placed into teflon-lined stainless steel autoclaves and heated at 100 °C for 48 h. Afterwards, the product was filtered, washed thoroughly with deionized water and dried at 100 °C overnight. Finally, the material was calcined at 550 °C for 6 h. The catalysts were prepared with 10 wt% Ni2P (the active phase was calculated to be 10% of the weight of the support) and with a Ni/P mole ratio of 1.25 by incipient wetness impregnation. Initially, an aqueous solution of nickel nitrate hexahydrate was prepared. To this solution, dibasic ammonium phosphate was added, followed by dropwise addition of nitric acid until no solids remained undissolved. After drying, the calcination was performed under nitrogen (50 mL min−1) with a heating rate of 2 °C min−1 until 500 °C and held for 3 h. Then, the ex situ reduction was conducted under hydrogen (50 mL min−1) with a rate of 5 °C min−1 until 350 °C followed by a rate of 2 °C min−1 until 650 °C and held for 3 h. Finally, the catalysts were passivated in 0.5% O2/N2 (50 mL min−1) for 1 h at room temperature.Atomic absorption spectroscopy (AAS) to determine the amount of Si, Al and Ni was conducted on a Spectra AA 50B (Varian) spectrometer. Previously, the samples were digested using nitric acid, hydrochloric acid and hydrofluoric acid.Temperature programmed reduction (TPR) of hydrogen was obtained on a ChemBet-3000 (Quantachrome) instrument using a heating rate of 10 °C min−1 up to 900 °C with a flow of 20 mL min−1 of 5% H2/N2. Before the reduction, the samples were treated with N2 at 300 °C under a flowrate of 20 mL min−1 for 30 min.X-ray diffraction (XRD) analysis was performed on a XRD 600 (Shimadzu) diffractometer, using Cu-Kα radiation (λ = 1.54 Å, V = 40 kV, i = 30 mA), a rate of 2° min−1 and a range of 2θ from 5° to 80°. The XRD analysis using small angle was conducted on a small angle X-ray scattering (SAXS) N8 Horizon (Bruker) diffractometer between the 2θ range from 0.1° to 5.5°.Nitrogen physisorption to evaluate the specific surface area, pore volume, and pore size was obtained on a ASAP 2020 (Micromeritics) instrument at liquid nitrogen temperature (77 K). The specific surface area was calculated using the Brunauer-Emmett-Teller (BET) method, while the average pore size was obtained from the Barrette-Joyner-Halenda (BJH) method.The temperature programmed desorption (TPD) of ammonia was carried on an Autochem II 2920 (Micromeritics) equipment. First, the samples were treated at 300 °C for 2 h under He flowrate of 20 mL min−1. Then, they were saturated with 10% NH3/He at 100 °C under a flow of 20 mL min−1 for 1 h, followed by a 50 mL min−1 He flow for 1 h. The desorption increased the temperature at a rate of 10 °C min−1 up to 700 °C. The acid sites were classified into weak ( ≤ 200 °C), medium (200° – 350 °C) and strong ( ≥ 350 °C) [30].Transmission electron microscopy (TEM) examinations were conducted on a JEM-1400 (JEOL) instrument. Previously, the samples were dispersed in isopropyl alcohol and then supported on a carbon grid and dried at room temperature overnight.Fourier-transform infrared spectroscopy (FTIR) was performed on a Vertex 70 V (Bruker) spectrometer using 4 cm−1 resolution and wavelength range from 4000 cm−1 to 400 cm−1. To prepare the samples, 1 mL of pyridine was dropped on 1 g of sample followed by heating at 150 °C overnight. Then, pellets were made blending 200 mg of KBr and 1 mg of sample.The deoxygenation tests were conducted in an autoclave batch reactor of 160 mL from Parr Instruments equipped with a sampling cylinder to allow the addition of the reactant free of oxygen. First, 1.5 g of passivated catalyst was reduced in situ at 5–10 bar and 250 °C for 12 h under hydrogen flow (50 mL min−1). After adding 50 g of oleic acid and reaching the desired temperature (260 °C, 280 °C and 300 °C), 50 bar of hydrogen was added in order to initiate the reaction and the catalytic tests were conducted for 6 h. A test with no catalyst was also performed in order to evaluate thermal effects. Afterwards, the liquid products were treated according to the literature [31]. Firstly, it was added to a test tube 100 μL of sample and 2 mL of 0.5 mol L−1 of NaOH in methanol and this was heated at 90 °C for 5 min. After cooling, it was added 3 mL of another solution (50 g of H2SO4 and 33.3 g of NH4Cl in 1 L of methanol) and heated at 90 °C for 5 min. Then, after cooling, it was added 3 mL of hexane and 2 mL of a saturated solution of NaCl. Finally, the material was centrifuged to separate the phases. The analyses to quantify the samples were conduct on a GC-10 Plus gas chromatograph (Shimadzu) with flame ionization detector (FID) equipped with a VA-5 (Varian) capillary column with dimensions 30 m × 0.25 mm × 0.25 μm, while the product determination was conducted using the NIST library. The conversion of oleic acid and degree of deoxygenation were determined by Eqs. (1) and (2), respectively: (1) X = C O A i - C O A C O A i . 100 (2) D O = C F A i - C F A C F A i . 100 where X is the oleic acid conversion (%), C O A i and C O A are the initial and final concentration of oleic acid (wt %), respectively, D O is the degree of deoxygenation (%), C F A i and C F A are the initial and final concentration of free fatty acids (wt %), respectively.The turnover frequency was determined by Eq. (3): (3) T O F = r H C L where T O F is the turnover frequency (h−1), r H C is the global reaction rate (mol gcat −1h−1) and L is the theoretical metal site concentration (mol gcat −1).An alternative to CO uptake is the determination of the theoretical metal site concentration, given by Eq. (4) [32]: (4) L = 6 n C ρ d n A where n is the average surface metal atom density of Ni2P (1.01.1015 cm−2), C is the fractional weight loading (gNi2P gcat −1), ρ is the density of Ni2P (7.09 g cm−3), d is the crystallite size calculated from Scherrer equation (cm) and n A is the Avogadro’s constant (mol−1).Assuming the behavior of an ideal batch reactor with constant volume, the reaction rate may be calculated as shown in Eq. (5) [33]: (5) r H C = m O A W d C H C d t 1 M M where m O A is the mass of oleic acid (50 g), W is the weight of catalyst (1.5 gcat), C H C is the total amount of hydrocarbons C10-C18 (wt %), t is time (h) and M M is the average molar weight of the products (g mol−1).The TPR profiles of the calcined catalysts, denoted NixPyOz/support, are illustrated in Fig. 1 . The nickel phosphide formation occurs by reduction of Ni+2 to metallic Ni followed by phosphate species reduction [34]. The NixPyOz/H-ZSM-5 catalyst started to reduce at 600 °C and was complete at 950 °C. On the other hand, the catalysts supported on USY and Al-SBA-15 reduced at lower temperatures, both starting at 400 °C and ending at 800 °C and 870 °C, respectively. The first reduction peak at 723 °C (NixPyOz/H-ZSM-5), 509 °C (NixPyOz/USY), 640 °C (NixPyOz/Al-SBA-15) can be attributed to reduction of NiO species, while the second peak at 835 °C (NixPyOz/H-ZSM-5), 594 °C (NixPyOz/USY), 783 °C (NixPyOz/Al-SBA-15) can be ascribed to reduction of P on the P-OH bond [7,9,35].As the reduction of the nickel phosphide supported on H-ZSM-5 occurs at higher temperature, there is a stronger interaction between the support and the metals [36], which can be assigned to the strong interaction between Ni and the acid OH of the zeolite [36]. The strong interaction between the USY zeolite and PO4 −3 can reduce the interaction between nickel and phosphate species, what explains its reduction at lower temperatures [9]. In addition, the results from the TPR and TPD analyses suggest that catalysts with stronger weak acid sites (desorption peaks of temperature < 200 °C) are more difficult to reduce, thus requiring higher temperatures at the reduction. A previous study has concluded that a stronger Brönsted acidity (e.g. weaker O–H bonds) hinders the reduction of the catalysts, as it facilitates the mobility of H+ species [37]. Indeed, the TPD result (Section 3.1.4) shows that the order of strength of weak acid sites is: USY < Al-SBA-15 < H-ZSM-5, which is consistent with the reducibility order given by the TPR.The wide angle XRD patterns of the supports and nickel phosphide reduced catalysts are shown in Fig. 2 . Comparing the patterns of the zeolitic supports with patterns found in the literature, it is possible to confirm the MFI structure of the H-ZSM-5 [23,24,25] and the FAU framework of the Y zeolite [12,25,26,27]. On the other hand, the peak at around 22.5° at the Al-SBA-15 and its reduced catalyst can be assigned to amorphous silica found on the pore wall of this ordered mesoporous material [7,26].The pattern of Ni2P/H-ZSM-5 shows peaks at 40.8°, 44.7°, 47.4°, 54.2° and 74.8° ascribed to the Ni2P nickel phosphide phase [35]. On the other hand, the pattern of Ni2P/USY exhibits only a minor peak 44.7° attributed to Ni2P [35]. This difference can be explained by the high specific surface area of the USY zeolite, which enhances the crystallites dispersion [40]. The nickel phosphide supported on Al-SBA-15 exhibits peaks at 40.7°, 44.5° and 54.2° attributed Ni2P, while the peaks at 32.6°, 38.4°, 41.8°, 46.9°, 48.9°, 56.1° and 60.1° are ascribed to Ni12P5 phase [35].It is also important to highlight that the Ni2P phase is more active in hydrotreating reactions than the Ni12P5 phase [39]. Moreover, an excess of P is necessary to form the Ni2P phase [41,42]. However, even though the synthesis used excess P, this amount was not enough to form only Ni2P on the Al-SBA-15.The low angle SAXS patterns of the support Al-SBA-15 and its reduced catalyst Ni2P/Al–SBA–15 are illustrated in Fig. 3 . These patterns are typical of ordered mesoporous materials, exhibiting a main peak at the (100) lattice plane and two peaks of lower intensity at (110) and (200) planes, or peaks at 0.85°, 1.45° and 1.68°, respectively [43]. Furthermore, the pattern of the reduced catalyst confirms that the structure of Al-SBA-15 is kept even at a high reduction temperature (650 °C). The TEM of this sample (Section 3.1.5), also corroborates this, showing the typical structure of the Al-SBA-15 molecular sieve.Physicochemical properties of the supports and the reduced catalysts are detailed on Table 1 . The nominal Si/Al ratios are close to the experimental Si/Al for all catalysts. Furthermore, the Ni loading of the reduced catalysts is also close to the nominal value of 7.91 wt% Ni (that is equal to 10 wt% Ni2P). After nickel phosphide impregnation, the specific surface area and pore volume decreased. The reduction is more intense for Ni2P/Al-SBA-15, suggesting accumulation of the active phase on its channels [26,44]. The catalyst with the higher surface area is Ni2P/USY, followed by Ni2P/H–ZSM-5 and Ni2P/Al-SBA-15. Although USY zeolite has a significant mesoporous contribution, the zeolite in this study did not suffer an intense dealumination process, due to the small mesoporous area compared to the total area and the low Si/Al ratio, close to 6, the value found to be a limitation of the synthesis process of the zeolite Y [39]. The Ni2P/Al-SBA-15 exhibits the higher pore volume, followed by Ni2P/USY and Ni2P/H-ZSM-5.The TPD results of the supports and reduced catalysts are given on Table 2 and the acid profiles are shown in Fig. 4 . The acid strength of the supports, evaluated by the temperature [45] follows the order: H-ZSM-5 > USY > Al-SBA-15. The ordered mesoporous material Al-SBA-15 showed the lower acid strength when compared with the zeolites, due to the fact that the walls of its pore are amorphous [46]. Moreover, the H-ZSM-5 acidity is stronger than the USY zeolite, because the latter has a lower Si/Al, therefore, there are more Al atoms on its structure and more acid sites and its framework suffers less unbalanced during the creation of acid sites [47]. Furthermore, the amount of acid sites is related with the Si/Al ratio. The Si/Al ratio increases with the order: USY < Al-SBA-15 < H-ZSM-5, while the total amount of acid sites decreases with the following order: USY > Al-SBA-15 > H-ZSM-5. This occurs because a lower Si/Al means a higher amount of Al atoms and as a consequence a higher amount of acid sites [47].Regarding the acid strength of the nickel phosphide catalysts, they follow the same trend as the bare supports. Comparing the nickel phosphide catalysts and their respective supports, they show a reduction of the total amount of acid sites and this is in accordance with previous studies [22,36,48]. Comparing the USY zeolite and its catalyst, there is an increase of the amount of weak and medium sites, while there is a decrease of the quantity of strong sites, probably because the nickel phosphide creates weak acid sites and at the same time it covers strong acid sites [9,22]. In addition, the surplus acidity can be ascribed to the P-OH group on Ni2P [38], since nickel phosphide has both Brönsted and Lewis acidity, attributed to P-OH and to electron deficiency of Ni, respectively [9,22]. Regarding the nickel phosphide supported on Al-SBA-15 and its support, a decrease of weak and strong sites takes place and we suggest that even tough occurs creation of acid sites, the destruction of them is more intense. Moreover, the Ni2P supported on H-ZSM-5 presents a reduction of the amount of weak and strong sites, which can be ascribed to the strong interaction between Ni and the acid OH on the zeolite [36].The TEM images of the supports and the reduced catalysts are illustrated in Fig. 5 and the nickel phosphide particle size distribution of the catalysts are given in Fig. 6 . These figures show that the average diameter of the phosphide particles are 18.84 nm, 10.25 nm and 13.10 nm for Ni2P/USY, Ni2P/H-ZSM-5 and Ni2P/Al-SBA-15, respectively. It seems that the particles have a better dispersion on USY, due to its high surface area and its mesopores, that can enhance the metallic dispersion. On another hand, the particles supported on H-ZSM-5 are not well dispersed, due to the presence of micropores contained on large crystals [48]. Moreover, the image of nickel phosphide supported on Al-SBA-15 also shows a good dispersion and it confirms that the channels and the ordered structure are maintained, even the at a high reduction temperature (650 °C) [7]. In addition, the catalysts supported on USY and Al–SBA–15 can have nickel phosphide inside their pores because their mesopores present an average diameter of 14.29 nm and 7.23 nm, respectively, obtained by the BJH method (Section 3.1.3). On the other hand, due to its small micropores (0.53 nm × 0.56 nm), the metallic phosphide is formed on the external surface of H-ZSM-5 [38].The FTIR spectra of pyridine adsorbed on the supports and nickel phosphide catalysts are illustrated in Fig. 7 . The bands at 1540 cm−1 and at 1450 cm−1 are ascribed to pyridine protonated by Brönsted sites and pyridine coordinated with Lewis acid sites, respectively [49], while the band at 1490 cm−1 is attributed to combination of both acid sites [49,50]. Albeit some bands are not very defined on some samples, all supports and all catalysts exhibit bands assigned to Lewis sites, to Brönsted sites and combination of both of them.Regarding zeolites, Brönsted sites can be ascribed to OH acid, such as terminal silanol groups (Si-OH) [51] and bridged hydroxyl group (Si-OH-Al) [47], while Lewis acid sites are due to extraframework aluminum [51]. In addition, nickel phosphides catalysts present Brönsted sites due to P-OH group, resulting of non-reduced P [9,22] and Lewis acid sites are ascribed to non-reduced and partially reduced Ni species [27]. Moreover, it is important to highlight that Brönsted sites are required in reaction involving transformation of hydrocarbons, while Lewis sites do not present catalytic activity by themselves, however, they enhance the strength and activity of sites when associated with them [52].The product distribution correspond to diesel fuel range (C10-C25) (Fig. 8 ) [13]. The main products are C17 and C18 hydrocarbons. C17 hydrocarbons derived from decarbonylation/decarboxylation of oleic acid, while C18 hydrocarbons derived from hydrodeoxygenation of this fatty acid [3]. The gas chromatography analysis of the oleic acid showed that this reactant contained 90% of oleic acid, 1% of stearic acid and 5% of palmitic acid and the latter explains the presence of C15 and C16 hydrocarbons, derived from decarbonylation/decarboxylation and hydrodeoxygenation, respectively. Furthermore, C15 and C16 can result from scission β and α of oleic acid, respectively, however, this approach is hardly considered on previous studies [53].The yield of the total amount of hydrocarbons (C10-C18) at 300 °C and at 6 h follows the order: Ni2P/Al-SBA-15 (42%) > Ni2P/H-ZSM-5 (29%) > Ni2P/USY (24%) (Fig. 8). Without catalyst, the production was c.a. 0.5% of C17 and 0.6% of C18 hydrocarbons, an insignificant production when compared to the catalytic tests, therefore proving the activity of the prepared catalysts. The highest yield over the catalyst supported on the Al–SBA-15 can be ascribed to its mesoporous nature, which implies less diffusional resistance. Between the zeolites, the catalyst supported on H-ZSM-5 achieved more hydrocarbons, probably due to its strong acidity and although the USY has a mesoporous contribution, it has a small amount of them. Furthermore, albeit the major phase on the Ni2P/Al-SBA-15 was the Ni12P5, which is least active than the Ni2P phase on hydrotreating reactions [39], it yielded more hydrocarbons, confirming its impressive activity (i.e. production of hydrocarbons). Comparing with previous studies, Jeon et al. (2019) obtained 54% of selectivity of C9-C17 hydrocarbons with a Ni catalyst promoted with Pt and supported on Ce0.6Zr0.4O2 on the deoxygenation of oleic acid [5]. In addition, the work of Silva et al. (2016) resulted in 43% of hydrocarbons on the deoxygenation of macauba pulp oil using Pd/C [13]. As on the present work it was obtained an amount of hydrocarbons close to the previous results, this shows that the Ni2P/Al-SBA-15 has a remarkable activity on hydrocarbons production, comparable to noble metals. As can be seen in Table 3 , the reaction conditions of previous studies, considering the temperature, the pressure and hydrocarbon yield is similar to the ones used for this study, which permits an adequate comparison between the results. Although nickel phosphide has been tested as an active phase, no previous studies were found that compare the effects of the supports presented in this paper.Concerning the surface specific area of the catalysts, it follows the order: Ni2P/USY > Ni2P/H–ZSM-5 > Ni2P/Al-SBA-15, suggesting that the catalyst area is not related with the reaction pathway [54]. In addition, as the oleic acid kinetic diameter is 0.55 nm [55], which is bigger than the micropores of the zeolites USY and H-ZSM-5, thus the reaction occurs on the external crystal surface [54]. The difference between the zeolites can be attributed to the strong acidity of H-ZSM-5, resulting in lighter products (C10-C16) (Section 3.2.2), due to cracking reactions [56].Another factor that can impact the production of hydrocarbons is the crystallite size [5,36]. On the present work, the nickel phosphide crystallites sizes follows the order: Ni2P/H-ZSM-5 (10.25 nm) < Ni2P/Al–SBA-15 (13.10 nm) < Ni2P/USY (18.84 nm). Indeed, as this is related with production of hydrocarbons, the catalysts with smaller crystallites produce more hydrocarbons, as shown by fact that Ni2P/Al–SBA–15 and Ni2P/H-ZSM-5 presented a higher degree of deoxygenation (Section 3.2.2). This is in agreement with the work of Zhang et al. (2017), in which the authors investigated the effect of citric acid addition on the size of nickel phosphide nanoparticles supported on mesoporous H–ZSM-5 and they concluded that the hydrodesulfurization activity of 4,6– dimethyldibenzothiophene increased with smaller Ni2P particles [36]. Moreover, it is known that Ni2P has two types of Ni structures: tetrahedral (Ni (1)) and pyramidal (Ni (2)). The fraction of the latter increases with a decrease on the particle size. As this site is responsible for the hydrodeoxygenation pathway while Ni (1) is responsible for decarbonylation/decarboxylation reactions [57]. As the main product is C17 hydrocarbons for all catalysts, we can state that the size of the Ni2P crystallites are such that the majority of sites is in the form Ni (1), thus decarbonylation/decarboxylation reactions prevail, which is an advantage due to the lower consumption of H2 when compared to hydrodeoxygenation [1].The results of the turnover frequency are summarized in Table 4 and they are in agreement with results reported previously [7,58]. The TOF follows the order: Ni2P/Al-SBA-15 > Ni2P/H-ZSM-5 > Ni2P/USY, which is the same order of the production of hydrocarbons. Albeit Ni2P/USY achieved the highest metal site concentration, it had the lowest TOF, due to the lowest production of hydrocarbons.The catalytic tests were performed at 260 °C, 280 °C and 300 °C as this range of temperature is appropriate for the deoxygenation of fatty acids [59]. Fig. 9 shows that the total amount of hydrocarbons (C10-C18) increased from 260 °C to 280 °C however, there is a decrease from 280 °C to 300 °C. Fig. 10 exhibits the degree of deoxygenation according to the temperature. At 300 °C the highest degree of deoxygenation is ascribed to Ni2P/H-ZSM-5, followed by Ni2P/Al–SBA-15 and Ni2P/USY. At 280 °C Ni2P/Al-SBA-15 achieves the highest oxygen removal and at 260 °C this catalyst has the lowest degree of deoxygenation. Fig. 11 (c) shows that the yield of lighter hydrocarbons (C10-C16) increases with temperature, because cracking reactions are favored are higher temperatures [1]. Regarding C17 hydrocarbons (Fig. 11 (a)), there is a peak of production at 280 °C. The oleic acid conversion (Fig. 11 (d)) also exhibits a maximum hydrocarbon production at 280 °C except for the H-ZSM-5 supported catalyst. We suggest that the decrease of production at 300 °C can be ascribed to catalyst deactivation [2] and probably H–ZSM-5 does not deactivate due to small pores, preventing coke deposition [47]. Regarding C18 hydrocarbons (Fig. 11 (b)), more hydrocarbons are achieved at a higher temperature, except with the catalyst supported on USY. Another interesting outcome is that Ni2P/Al-SBA-15 achieves more hydrocarbons (C10-C18) at 300 °C and 280 °C but it achieves less than the other catalysts at 260 °C. This shows that this catalysts leads to a reaction mechanism with higher activation energy, and consequently, changes in temperature have a larger impact on reaction rate.The literature shows that long reaction runs decreases the diesel yield, as higher temperatures and long reaction times lead to cracking of fatty acids and hydrocarbons into smaller molecules [1], supporting the activity decrease from 280 °C to 300 °C. For instance, Horáčcek et al. (2019) tested Mo carbide, nitride and phosphide catalysts on the deoxygenation on rapeseed oil and it was reported a decrease on the catalytic activity as temperature increased due to an increase on cracking reactions [60]. In addition to that, Yang et al. (2013) noticed a decrease on the yield of C15-C18 hydrocarbons as temperature increased from 300 °C to 325 °C and to 350 °C on the hydrotreating of oleic acid [61].According to the product distribution obtained and based on previous works, a possible reaction pathway was proposed for the deoxygenation of oleic acid over nickel phosphide catalysts, as illustrated in Fig. 12 . Firstly, oleic acid is hydrogenated to stearic acid (1) [2,5] or it can be decarboxylated [12] and hydrogenated to heptadecane (2). Stearic acid can then form heptadecane from decarboxylation (3) [2,12,17] or it can be reduced to form an aldehyde (4) [17,33]. The aldehyde can form heptadecane from decarbonylation (5) [17] or it can reduce and result in an alcohol (6) [17,33]. Finally, hydrodeoxygenation of alcohol result in octadecane (7) [33].The catalysts present different behaviors regarding reducibility, acidity and dispersion, which influences that deoxygenation activity and the diesel yield. The total amount of hydrocarbons (C10-C18) reached at 300 °C followed the order: Ni2P/Al-SBA-15 (42%) > Ni2P/H-ZSM-5 (29%) > Ni2P/USY (24%). The catalyst Ni2P/Al-SBA-15 achieved more hydrocarbons than the other catalysts due to its mesoporous nature and its small nickel phosphide particles size. Between the zeolites, the Ni2P/H-ZSM-5 produced more hydrocarbons, on account of its strong acidity and small particle sizes. All catalysts exhibited more C17 than C18 hydrocarbons resulting from decarbonylation/decarboxylation reactions. Regarding the temperature effect, the amount of hydrocarbons increased from 260 °C to 280 °C but it decreased from 280 °C to 300 °C, probably due to catalyst deactivation. Mariana de Oliveira Camargo: Conceptualization, Methodology, Formal analysis, Investigation, Writing - original draft. João Lourenço Castagnari Willimann Pimenta: Writing - review & editing, Formal analysis, Investigation. Marília de Oliveira Camargo: Writing - review & editing, Investigation. Pedro Augusto Arroyo: Supervision, Project administration, Funding acquisition, Writing - review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.We would like to acknowledge CAPES, Brazil (Coordination for the Improvement of Higher Education) for the student scholarship. We are also grateful to COMCAP at State University of Maringá by the analyses conducted on its facilities such as SAXS, TEM and FTIR.

An alternative to non-renewable fuels is the production of green diesel from deoxygenation of vegetable oils. In this study, the effect of the supports on nickel phosphide catalysts for deoxygenation of oleic acid to produce hydrocarbons was investigated. Nickel phosphide catalysts supported on USY, H-ZSM-5 and Al-SBA-15 were synthetized by temperature programmed reduction (TPR) of metal phosphate precursors and the prepared catalysts were characterized by atomic absorption spectroscopy (AAS), X-ray diffraction (XRD), small angle X-ray scattering (SAXS), temperature programmed reduction of hydrogen (TPR), nitrogen physisorption, temperature programmed desorption of ammonia (TPD), transmission electron microscopy (TEM) and Fourier-transform infrared spectroscopy (FTIR). The catalytic tests were performed in an autoclave batch reactor at 260 °C, 280 °C and 300 °C and 50 bar of hydrogen. The production of hydrocarbons (C10-C18) at 300 °C followed the order: Ni2P/Al-SBA-15 > Ni2P/H-ZSM-5 > Ni2P/USY. The Ni2P/Al-SBA-15 exhibited a remarkable deoxygenation activity, compared to noble metals. All the catalysts achieved more C17 than C18 hydrocarbons, therefore, decarbonylation/decarboxylation reactions prevail. With regard the temperature effect, the amount of hydrocarbons increased from 260 °C to 280 °C but it decreased from 280 °C to 300 °C for all catalysts, which can be ascribed to catalyst deactivation.

Serious environmental and health challenges that are caused by air pollution such as global warming, weather changes, damage to plants, and spread of many diseases such as chronic asthma, pulmonary insufficiency and cardiovascular diseases, which is responsible for about 9 million deaths per year, have become a common problem in the world [1,2]. For example, the combustion of fuel is the main cause of emission of most CO pollutant, which has been called as the silent killer for the 21st century [3,4]. However, the World Health Organization has reported that, stricter guidelines for air quality standards not to exceed 4 mg/m3 for CO [5]. Therefore, many researchers have reported that the removal of pollutants from exhaust gases at ambient conditions is carried out using heterogeneous catalytic methods [6]. For example, Dey et al. study on gases of a laboratory composition containing CO and O2 and argon or nitrogen as balance, showed that silver (Ag) is the best catalyst for many catalytic oxidation reactions and is highly active for carbon monoxide (CO) oxidation at low temperature [7]. Whereas, the study concluded that the performance of silver catalysts depends on their structure, surface active sites, and different AgO reactions. So the small size silver metal (Ag0) particle is the main factor to improve the catalytic performance of the supported catalyst [7]. While Hernandez et al. reported that Ag0 is formed on the catalysts prepared by impregnation even without a reduction treatment [8]. Among the catalysts used in catalytic oxidation reactions are also transition metal oxide catalysts, for example nickel oxide catalyst supported on Al2O3 or TiO2 [9]. Mixed silver‑nickel oxide AgNiO2 have high activity in CO oxidation in gases of a laboratory composition containing CO and O2, and helium as balance at low temperature [10]. It was noticed that above 150 °C the removal of CO becomes complete because the structure of AgNiO2 undergoes complete and irreversible destruction with the formation of individual Ag0/AgOX and NiOX particles [10]. Among the cheap transition metal oxide catalysts, Iron oxide and their composite oxides are also used as catalyst and catalyst carriers on CO oxidation [11]. furthermore, it has been found that introduction of silver to the transition metal oxide systems strongly increased activity of the catalysts in the oxidation of CO [12,13,14], combustion of methane [15,16] or volatile organic [11,17]. The increased activity is often attributed to the donation of oxygen from surface transition metal oxide sites to silver species [18] and increased dispersion of silver [19,20]. On the other hand, Zhang et al. reported that silver catalyst supported on iron oxide, alumina and montmorillonite, as a clay binder, has a high catalytic activity for propylene oxidation with a complete removal 100% at 350 °C [21]. Clay bonding materials of zeolites or bentonite or a mixture of them are usually used, which increase the dispersion of the catalyst due to the large specific surface of these materials [22]. In general, Eqs. (1),(2) and (3) show the air oxidation of CO and HC over a catalyst [23]. (1) CO + O 2 → CO 2 (2) HC + O 2 → CO X + H 2 O (3) CO + H 2 O → CO 2 + H 2 The catalytic efficiency of the oxidation reaction of CO and HC increases with the increase in the concentration of oxygen in the exhaust gas. Therefore, a phase of air (secondary air injection) is added to the exhaust gas after remove nitrogen oxides in a reduction catalyst [24,25]. The main goal of the present study is to prepare a new catalyst from relatively cheap raw materials(a mixed fabrication of NiO, Ag, Fe2O3, Al2O3, and Syrian natural bentonite) as a catalyst for removing CO and CH instead of the triple catalyst that contains rare and expensive materials such as platinum, rhodium and palladium.Silver nitrate (98%) was purchased from Sigma- Aldrich, iron oxide (98%) from BDH. Nickel nitrate hexahydrate (98%) and alumina were bought from Merck. Syrian natural bentonite (63 μm), Aleppo Pylon. Commercial quartz tubes are 30 cm long and 1 cm average in diameter.X-ray fluorescence (XRF) analysis was performed on a Sequential ARL 8410 spectrometer. X-ray diffraction (XRD) patterns were recorded using a Philips (PW1830) diffractometer with Cu Kα radiation operating at 40 kV and 50 mA. the patterns were collected in a 2θ range from 3° to 60°, with a sampling width of 0.01° and a scanning step of 2°/min. The differential thermal analysis (DTA) was conducted with Shimadzu DTA & DTG-60H instrument. BET surface areas were determined from the nitrogen adsorption curve by the conventional multipoint technique with a Micromeritics Gemini 3. The samples for BET measurements were pretreated at 300 °C for 3 h at high vacuum. Surface morphology of catalysts was examined by scanning electron microscopy (SEM) (VEGA II xmu, TESCAN). The energy dispersive X-ray spectra (EDX) of samples were also acquired during their conventional scanning electron microscopic investigations. Gas Analyzer (Kane AUTO 5–1) was used for CO, CO2, O2, NO and HC determination.(NiO, Ag/ Fe2O3, Al2O3, Bentonite) catalyst was synthesized in our laboratory according to the literature [26] with some modification. Briefly, the mixed silver and nickel oxide was prepared by impregnation from a solution of AgNO3 (0.3 N) and Ni(NO3)2·6H2O (3 N) on a pre-calcined base composed of iron oxide, alumina and bentonite (0.57:1:1 wt) respectively. The base was suspended on etched quartz tubes (Fig. 1a). The length of the suspended part of the base on the quartz tube was 25 cm, and the overall thickness of the base layer inside and outside was 1 ± 0.15 mm. The whole catalyst was dried for 24 h in air and calcined at 400 °C for 5 h.The catalytic measurements were performed by pulse method using temperature-programmed reactor (an automatic setup) equipped with two flowmeters and gas analyzer. The reactor and blank tube were placed vertically in a split-open furnace as shown in Fig. 1b. Applying the same temperature conditions to both the catalytic reactor and the blank tube allows eliminating the effect of gas expansion in the gas phase. First, the engine is running for five minutes, then the exhaust gas was collected in a vacuum air bag in order to control the exhaust gas flow velocity precisely and obtain a stable exhaust gas composition for each single experiment (one temperature and one flow velocity). The exhaust gas was mixed with air at the same flow space velocity as it entered the reactor or blank tube. No difference in pressure was observed at the reactor inlet and outlet, as well as the blank tube due to the relatively large diameter of the quartz tube that suspends the catalyst and space between each other. As a result, a rather homogeneous diffusion of the mixture gas occurs over the catalyst surface, so the mass and temperature gradients resulting from the oxidation reaction are minimal. In addition, the relatively short pulse time and the expansion of the catalyst positively affect the kinetic study, (small amount of heat released from oxidation reactions and its dissipation). The weight flow per hour of CO or HC was calculated using flow space velocity of mixture gas (F = 1000 to 1800 ml.min−1), CO or HC blank concentration, and ideal gas equation at laboratory temperature and pressure using Eqs. (4),(5). The weight hourly space velocity was then calculated by the Eq. (6). The catalytic oxidation of CO and HC over catalyst (NiO, Ag /Fe2O3-Al2O3-Bentonite) was carried out at atmospheric pressure in a fixed bed tubular reactor containing 3 tubes of catalyst (net weight of catalyst 5.81 g (at different temperatures (300°, 320°, 340°, 360°, 380° and 400 °C) and different weight hourly space velocity (WHSVCO 0.193 to 0.529 h−1, WHSVHC 0.014 to 0.031 h−1). The initial concentrations of the components of the gas mixture measured in the blank tube (CO 1.71 to 2.8%, HC 394 to 525 ppm, CO2 1.1 to 2.1%, O2 13.02 to 16.46%) NO gas was not observed. The conversion of CO or HC was calculated by the Eq. (7). (4) PV = nRT (5) weight x flow per hour = F × P × Mw x × C x blank × 60 R × T (6) WHSV x = weight x flow per hour catalsyt weight (7) X x = C x blank − C x catalyst C x blank Where x is either CO or HC in Eqs. (5), (6) and (7).Characterization of bentonite was carried out using XRF which aimed to analyze and determine the elemental composition. Table 1 shows the elements present in the bentonite sample.DTA analysis of the natural bentonite in Fig. 2a demonstrates endo-thermal actions at 107.34 °C, 224.36 °C and 747.25 °C. Önal and Sarikaya reported that endo-thermal actions of the natural bentonite between 25 °C to 400 °C are due to dehydration of inter particle water, adsorbed water and inter layer water, and endo-thermal actions between 400 °C to 800 °C with the maximum rate at 668–672 °C was due to the formation of dehydroxylation water [27]. Dehydroxylation is defined as the “loss of a water molecule from two adjacent hydroxyls molecules” [28]. In addition, the third endothermic action may be due to the decay of carbonates, as XRD spectrum in Fig. 2b shows that bentonite contains calcite (CaCO3) and dolomite (MgCa(CO3)2) which is decomposed to (MgO + CO2 + CaCO3) by an endothermic action at 750 °C to 800 °C [29]. XRD also showed that bentonite contains quartz, palygorskite, Kaolinite and montmorillonite. It has been found that the essential beams of an X-ray diffraction spectrum for the minerals in clay appear within 2° to 37° (2θ) [30].Thus, the calcining temperature did not increase during the calcining of the base (Fe2O3, Al2O3, Bentonite) or the catalyst (NiO, Ag/ Fe2O3, Al2O3, Bentonite) over 400 °C. The TGA curve in Fig. 2a indicates a weight loss of about 5% in the adsorbed dehydration region (<400 °C) and a weight loss of 17% in the dehydroxylation region and carbonates decay (400–800 °C).SEM images of catalyst are shown in Fig. 3 , which shows pronounced differences in the particle shapes. The catalyst particle sizes in Fig. 3a, ranged from 2 μm to 4 μm with aggregation. Small particles of different sizes are observed at 10,000 times zoom as shown in Fig. 3b. Relatively large nanoparticles, with diameters ranging from 100 to 500 nm, have been observed at 30,000 times zoom as shown in Fig. 3c. A clearer picture was obtained by enhancing the Fig. 3c using ImageJ software which is shown in Fig. 3d. The later was used to calculate the distribution of particles according to their diameters as shown in Fig. 3e. The elemental analysis of full area in Fig. 3b were confirmed by the EDX analysis in Table 1, revealing that it consisted mainly of O, Fe, Al, Si, Ni and Ag in addition to C. This is due to the presence of dolomite and calcite in bentonite as shown in Fig. 2B.BET surface area and pore volumes of bentonite, alumina, the base (Fe2O3, Al2O3, Bentonite) and the catalyst (NiO, Ag/ Fe2O3, Al2O3, Bentonite) were determined as shown in Table 1. The decrease in the BET surface area and the pore volumes was observed when the base had been impregnated with silver and nickel to prepare the catalyst. The decrease is likely due to agglomeration of Ag on the surface of the catalyst and occlusion of small-sized pores [31], which corresponds to what is shown in Fig. 3C, d.The concentrations of CO, CO2 and HC in the gas mixture from the catalytic reactor or from the blank tube were monitored using infrared radiation in gas analyzer (Kane). Fig. 4A & B show the conversion ratios of CO and HC, respectively, during 7:30 min at all tested temperatures and WHSVs. Relative stability of the conversions was observed along the time as well.Highest conversion ratios for CO were about 99% at 400 °C with all tested WHSVCO, and at 380 °C when (WHSVCO is 0.216 h−1) (Fig. 4a). Then CO conversion ratio decreases with the increase of WHSVCO (96–72)% at 360 °C Fig. 4a, (91–48)% at 340 °C Fig. 4a, (74–22)% at 320 °C Fig. 4a and (62–13)% at 300 °C Fig. 4a.HC conversion ratios decrease with the increase of WHSVHC (77–62)% at 400 °C Fig. 4b, (63–39)% at 380 °C Fig. 4b, (45–18)% at 360 °C Fig. 4b, (26 – 8)% at 340 °C Fig. 4b, (11 – 4)% at 320 °C Fig. 4b and (8 – 4)% at 300 °C Fig. 4b.The values of the apparent reaction rate constant are calculated when the kinematic equation is applicable to the residual concentration of CO and HC. the reaction of CO oxidation on NiO is of the first order of CO concentration at high temperature [32]. The applicability of the kinetic equation for the conversions of CO and HC was observed at all temperatures, except the temperatures in which the conversion rate of CO exceeded 98%. The shift of these points from linearity is observed in Fig. 5a & b. The catalyzed sites are likely to be occupied by CO oxidation when the conversion ratio is above 98%. Chin et al. found that “O reacts much faster with CO than with CH4, causing any CO that forms and desorbs from metal cluster surfaces to react along the reactor bed with other O to produce CO2 at any residence time required for detectable extents of CH4 conversion” [33–37]. Moreover, the initial concentration of CO is much more than the initial concentration of HC, ([CO]/[HC] = 40 to 65).New catalyst was prepared by impregnation method from base of Syrian bentonite, alumina and iron oxide for use as supports of nickel oxide and silver catalyst. It was used to oxidize CO and HC from exhaust gas emissions of gasoline electricity generator. The analysis results showed that the catalyst contained varied nanoparticles size 100–500 nm, and contained 10.01w% nickel and 3.22 w% silver on its surface. The surface area was 31.01 m2.g−1 by BET equation. The activity catalytic results showed that the catalyst has high activation for oxidation CO ̴ 99% at 400 °C with all tested WHSVCO, and it had medium activity for oxidation HC 77% at 400 °C when WHSVHC was 0.014 h−1. A. Fawaz: Conceptualization, Methodology, Investigation, Visualization, Writing – review & editing. Y.W. Bizreh: Funding acquisition, Project administration. L. Al-Hamoud: Visualization, Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This research was funded by the Scientific Affairs in Damascus University. The authors would like to thank Prof. Wail Al-Zoubi, Dr. Maysoon Alhafez and Dr. Hanan Al-Chaghouri for their fruitful help.

The catalytic oxidation of carbon monoxide (CO) and hydrocarbons (HC) is a very important process for maintaining health protection systems. So nickel and silver are widely used in heterogeneous catalysis. In this work, the activate of (NiO, Ag/ Fe2O3, Al2O3, Bentonite) based gasoline Oxidation Catalyst on the exhaust emissions of gasoline electricity generator was tested. Catalyst was confirmed by scanning electron microscope (SEM), Energy Dispersive X-ray (EDX) and nitrogen adsorption for determination of BET surface area. This catalyst (Ni, Ag /Fe2O3-Al2O3- Bentonite) exhibits superior catalytic performance in the oxidation of CO and HC, with air as an oxidizer. Maximum conversion in CO and HC emissions were obtained at the rate of 99% and 77%.

With the increasing harsh of the legislation on a sulfur content, as the conventional method of reducing sulfur, the hydrodesulfurization (HDS) could not meet the increasingly severe requirements of deep desulfurization [1]. Recent years, the oxidative desulfurization (ODS) has been developed because of its high activity to aromatic sulfur compounds under mild reaction conditions [2–6]. In the ODS process, the aromatic sulfur compounds in oils can be oxidized into the corresponding sulfones by oxidants. Therefore, it is crucial to seek suitable oxidants and efficient catalysts for the ODS process. O2 in air has been regarded as a desirable oxidant due to its advantages of low cost, and source abundant [3,7]. However, O2 is difficult to be activated effectively under mild reaction conditions. Thus, the O2 activation is vital for the ODS catalyst performance [2–4].Recently, noble metal-based catalysts (W, Mo, V, Ti, Co, Cu, and Pt etc.) have been developed for the aerobic oxidation [2,8–12]. However, the high cost of noble metals limits their applications. Theoretical and experimental results demonstrated that the reduction of metal size into nano size or even a single atom would significantly improve the metal catalytic activity and selectivity due to more low-coordination atoms available for catalysis [13–15]. Therefore, dispersing the metal particles into single atom is highly desirable way not only to maintain a high density of active sites but also to reduce the cost of the catalyst [16,17]. Two dimensional hexagonal boron nitride (h-BN), possesses many inherent advantages such as high thermal and chemical stability [18]. Additionally, some defects such as boron and nitrogen vacancies are often identified during the synthesis of the h-BN nanosheets [19]. Such defects make it an excellent support to anchor single metal atom for catalytic reactions [20–24]. Recently, Cu and Pt nanoparticles supported on the h-BN have been synthesized and applied into the ODS process [2,11]. However, the single atom catalysts (SACs) have not been realized in the ODS field. In this work, we theoretically investigated the potential of Cu and Ni atoms supported on h-BN pristine and defective surfaces as the ODS catalysts. We hope this work will shed a light for the design of the SACs for ODS reactions.In this work, all calculations were carried out by using B3LYP functional with a dispersion-corrected term (B3LYP + D3) implemented in Gaussian 16 program [25]. For all the atoms including the metal atoms, a triple-zeta basis set with polarization functions (6-311G(d,p)) was used to describe the electronic wave function. Such a method has been proved successful in previous studies [26]. All the coordinates were fully optimized and no imaginary frequencies were found for the reactants, intermediates, and products. The transition states (TSs) were identified by confirming the existence of an imaginary vibration mode and intrinsic reaction coordinate calculations. The temperature is 298 K used for the free energy calculations.Three cluster models are chosen to anchor the single metal atom on the h-BN monolayer, as shown in Fig. 1 . Specifically, h-BN model containing 27 B and 27 N atoms is used to simulate the pristine h-BN with the truncation boundary closed by H atoms. Two typical monovacancies defects are considered, one is boron vacancy (VB/h-BN), which is constructed from the h-BN by removing a boron atom. Similarly, a nitrogen vacancy (VN/h-BN) is built by removing a nitrogen atom from h-BN. All the displayed geometries have been optimized. The h-BN model maintains a planar configuration with a BN distance of 1.448 Å, consistent with the experiment value of 1.44 Å [27]. Both vacancies experience obvious lattice deformations. In the case of VB, the values of NN distance near the vacancy are 2.739, 2.739, and 2.878 Å, consistent with the corresponding value of 2.64, 2.65 and 2.70 Å in previous report [28]. For the VN, the distances between BB near the vacancy are 2.042, 2.043, and 2.458 Å, which are comparable with the values of 2.06, 2.10 and 2.10 Å in the previous report [28]. Thus, the accuracy of our models and theory is further confirmed, and will satisfy the purpose of our work. The adsorption Gibbs free energy was defined as: ΔG ad = G adsorbate + model − G free molecule or atom − G free model where G (adsorbate+model) denotes the total Gbbis free energies of the adsorption system, while G (free molecule or atom) and G (free model) are the total Gbbis free energies of the separate metal atom and h-BN models, separately. As the definition, positive values of ΔGad indicate an endothermic process, whereas negative values an exothermic process.Three types of structures for Cu and Ni atom anchored on h-BN are considered here, as shown in Fig.S1. In the case of the pristine h-BN surface, both TM atoms prefer to be adsorbed on the top of N atom of h-BN, and the adsorption height for Cu and Ni on the h-BN sheet is 2.019 Å, 1.835 Å, respectively, in good agreement with previous reports [21,29,30]. When the TM atom is introduced into the VB site, it is observed an outward displacement from the plane of h-BN due to the larger size of TM atoms as compared with B atoms, and three TM − N bonds are formed. Specifically, in the case of Cu/VB, three CuN bond lengths are 1.763, 1.763, and 1.801 Å, respectively, comparable with previous report of 1.83 Å [29]. In the case of Ni/VB, three NiN bonds are observed with the average bond length of 1.776 Å, in line with previous report of 1.81 Å [31]. In the case of N-vacancy, the TM atom is located near the center of the defect with three TM − B bonds formation. For example, for Cu/VN, the average Cu − B bond length is 2.102 Å, close to the value of 2.14 Å in previous report [29]. For Ni/VN, three Ni − B bond lengths are equivalent with the distance of 1.873 Å, in accordance with previous results of 1.91 Å [31]. Therefore, when TM atom is embedded in the VB or VN, the TMN3 or TMB3 moiety is formed and probably exhibit high catalytic activity for ODS reactions. Fig. 2 demonstrates the binding Gibbs free energy (ΔGb) of Cu and Ni on the pristine and defective h-BN surface. Table S1 summarizes the ΔGb and NPA charges for the adsorption system. Our calculations illustrate that Cu and Ni could not bind strongly on the pristine h-BN due to the slightly positive binding energy, In the previous report, although the calculated Eb was negative for Cu on h-BN, the small value of −5.1 ~ −5.5 kcal/mol also indicates the weak interaction between Cu and h-BN [29,32]. The interactions become stronger when TM atom is trapped on the VB or VN. Specifically, Cu atom is adsorbed at VB and VN with the binding energy of −95.7 and − 43.8 kcal/mol, respectively. In the case of Ni, our calculated binding energy is − 145.9 and − 69.5 kcal/mol on VB and VN, respectively. Thus, both TM atoms are more stable on VB site than on VN site, this is in good agreement with previous reports [29,31]. Besides, it is vital for a single atom catalyst (SAC) that the metal atom exists as a single atom rather than as diffused or aggregated atoms, the binding energy of the TM supported on h-BN is compared with the corresponding experimental cohesive energies of metals. For example, the experimental cohesive energy per atom of the bulk metal for Cu and Ni is − 80.5 kcal/mol and − 102.4 kcal/mol respectively [33]. Therefore, Cu and Ni atoms may prefer not to be clustered when they are dispersed on the VB site of h-BN. Overall, the binding energy is largely enhanced at the vacancies as compared with that on pristine h-BN, which indicates the monovacancy site can be a good anchoring site for the single metal atom.In previous study, O2 was found to be inert on the defect free h-BN surface, but activated on defected h-BN [26], and metal atom supported on h-BNNSs [20,21,29]. We performed the O2 adsorptions on the supported TM atoms, and the optimized geometries and corresponding binding energies are shown in Fig. 3 . In the case of h-BN surface, both TM atoms have the exceptional ability for O2 activation. On Cu/h-BN and Ni/h-BN, O2 is adsorbed on top of TM atom with the OO bond length largely stretched to 1.356 and 1.418 Å, respectively. And the corresponding adsorption energies are −29.4, −38.2 kcal/mol, respectively. However, the activation of O2 is quite different on TM/VB and TM/VN. Specifically, O2 is readily activated on TM/VN with the OO bond largely enlarged, and the corresponding binding energies are −20.0, −30.4 kcal/mol respectively. While in the case of TM/VB system, O2 is not easily adsorbed on both TM/VB due to positive binding energies. Therefore, the activation ability for O2 decreases in the order TM/h-BN, TM/VN and TM/VB. It is interesting to note TM/VB has the most stability but the least activity for O2 activation. With the exception on the TM/VB, O2 will be activated to the superoxide state (O2 ¯), which will be reflected from the NPA charges of adsorbed O2, see Table S2. As we all known, the O2 activation is a key step for the next ODS reactions. Therefore, the TM atoms supported on the pristine and the N-vacancy of h-BN will probably be an active ODS catalysts.As discussed above, O2 is not activated on the Cu/VB, the overall ODS process is further carried out on Cu/h-BN and Cu/VN surfaces. Fig. 4a presents the optimized geometries of each elementary step on the surface of Cu/h-BN. The co-adsorptions of O2 and dibenzophene (DBT) are both nearby the Cu atom, and the OO bond is stretched to 1.348 Å (ISa). Subsequently, the activated O2 will transfer one of its O to the DBT, forming transition state (TS1a). In TS1a, the OO bond is further elongated to 1.705 Å, and the dissociated O is gradually approaching the S atom of DBT with the SO bond shortened to 1.721 Å. As a result, the DBTO species is formed in the intermediate (IM1a), and the remaining O atom is settled on Cu atom with a CuO bond length of 1.752 Å. In the next step, Cu-O⁎ species will act as the active species for DBTO further oxidation. At last, the DBTOO is obtained in FSa by overcoming TS2a, and the single Cu atom is now free on top of N atom of h-BN surface, which will continue to play an important role for the ODS reactions. Fig. 4b displays the geometries of each step on Cu/VN surface. In the co-adsorption configuration, O2 is bridged adsorbed between Cu and B atom of the h-BN surface. Notably, the OO bond is largely stretched into 1.431 Å (ISb). Overcoming TS1b, one of the O atoms is transferred to DBT, resulting in the formation of DBTO in IM1b. Note that in IM1b the residual O atom is now bridged bound between the B and Cu, leading to the formation of a stable five-membered ring with the BO and CuO bond distance of 1.461 Å, 1.887 Å, respectively. As mentioned in our previous report [26], the bridge-bound O will lose its activity on metal-free surface. Fortunately, the bridged B-O⁎ species is transformed into Cu-O⁎ species as shown in IM2b, then Cu-O⁎ species as the active species is attacking the S atom of DBTO. Finally, DBTOO is produced as shown in FSb. Fig. 5 displays the corresponding free energy profiles on the Cu embedded surfaces. In the case of Cu/VN, firstly, the co-adsorption of O2 and DBT is thermodynamically beneficial by an energy release of 33.9 kcal/mol. Then the first O transfer occurs, which can be written as DBT + O2 ⁎ → DBTO +O⁎. This step must overcome the energy barrier of 37.5 kcal/mol. The free energy of DBT oxidation to DBTO is slightly uphill by 5.2 kcal/mol. In the subsequent step, the residual O atom experiences the transformation from the bridged adsorption into the terminal adsorption on the Cu atom, and this step requires an energy cost of 13.3 kcal/mol. At last, the DBTO is oxidized by the Cu-O⁎ species with the Gibbs free energy decreased by 7.8 kcal/mol. In this pathway, the ODS process could take place smoothly, and the first O transfer becomes the potential-limiting step. Finally, DBTOO can be easily to desorb and the Cu/VN catalyst thus be resumed for a new round of DBT oxidation. Likewise, in the case of Cu/h-BN, the first O transfer process is still the rate-limiting step with the energy barrier of 42.4 kcal/mol, and the second O transfer process takes place easily with a slight barrier of 2.2 kcal/mol. The overall reaction is exothermic by 0.4 kcal/mol on Cu/h-BN surface. Therefore, Cu/VN and Cu/h-BN could become the efficient catalysts for ODS reactions. The single Cu atom plays an important role for O2 activation and DBT oxidation. Specifically, O2 activation on CuN or CuB3 moiety ignites the oxidation, and the second O transfer process becomes easier due to the participation of the Cu-O⁎ species.The ODS process is also investigated on Ni/h-BN and Ni/VN surfaces as shown in Fig.S2. The whole process on Ni/VN is similar to that on Cu/VN surface. At first, O2 is largely activated on NiB3 moiety with a stretched bond length of 1.326 Å, which acts as the oxidant for the formation of the sulfoxide. Note that the residual O dissociated from O2 is now bridged adsorbed between the Ni and B atom. At the presence of NiB3 moiety, the bridged O could be transformed into the terminal O on Ni atom, and Ni-O⁎ active species is for the final production of sulfone. Likewise, O2 is firstly activated on Ni/h-BN and initiates the ODS reaction.The free energy profiles on Ni embedded surfaces are shown in Fig.S3. The first step of ODS is still the rate-limiting step with the intrinsic energy barrier of 46.9 and 50.2 kcal/mol on Ni/h-BN, Ni/VN surface, respectively. Although the intrinsic energy barrier is relatively high, considering the large co-adsorption energies of O2 and DBT, which are −50.4 and − 47.7 kcal/mol on Ni/h-BN and Ni/VN, respectively, the apparent activation energy is relatively low, which is −3.5, 2.5 kcal/mol on Ni/h-BN and Ni/VN, respectively. Therefore, the co-adsorption energies of O2 and DBT provide significant driving force for the ODS reactions. In the second step, Ni-O⁎ will account for the sulfone formation as the active species on both surfaces. Thus, both Ni/h-BN and Ni/VN surfaces will be appropriate catalysts for the ODS reaction.In summary, we systematically calculated the possibility of the single TM atoms (Ni and Cu) embedded on the pristine and defective h-BN sheet as the ODS catalyst. Some useful conclusions can be drawn from our calculations:TM atoms anchored on the VB or VN site are more favorable than on the pristine h-BN. Except for that on TM/VB surface, O2 could binds strongly and be activated to O2 − species on both TM/h-BN and TM/VN. The first step of DBT oxidation into DBTO by the activated O2 is the rate-limiting step, and then DBTO is finally oxidized at the presence of TM-O⁎ species. Considering the excellent stability of TM/VN, both Cu/VN and Ni/VN surfaces are expected to the good catalysts with high activity toward O2 oxidation and subsequent ODS process. Naixia Lv: Investigation, Formal analysis, Writing – original draft. Jinrui Zhang: Formal analysis, Data curation. Jie Yin: Resources, Methodology. Hongshun Ran: Resources, Methodology. Yuan Zhang: Formal analysis, Data curation. Tianxiao Zhu: Resources, Methodology. Hongping Li: Conceptualization, Writing – review & editing, Supervision, Funding acquisition.The authors declare no competing financial interest.This work was financially supported by the Guizhou Basic Research Project (ZK[2022]561), the Qianxinan Prefecture Science and Technology Project(2021−2−32), and the National Natural Science Foundation of China (Nos. 22078135). Supplementary material Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.catcom.2022.106492.

The rational design of the single transition metal atoms (Cu and Ni) anchoring on the boron nitride (TM-BN) pristine and defective monolayer has been investigated by means of density functional theory. Our calculations revealed that both B-vacancy (VB) and N-vacancy (VN) on BN monolayer are good sites for trapping Cu and Ni atoms. TM/VN and TM/h-BN systems exhibit high catalytic activity toward O2 activation and subsequent oxidative desulfurization (ODS) reactions. Considering the stability, the TM/VN is expected to be a good catalyst for ODS process.

CO2 capture and utilization is an important option for a future carbon-neutral economy as excessive CO2 emissions affect the climate and the environment [1,2]. The catalytic conversion of CO2 to value-added chemicals is one of the most promising and researched approaches. In combination with renewable hydrogen, the conversion of CO2 into fuels can help close the carbon cycle. Various products such as alcohols, acids, aldehydes, and olefins have been synthesized through thermal, electrical and photochemical CO2 conversion [3–8]. Among such products, CO2 hydrogenation to CO, also known as the reverse water-gas shift reaction (RWGS, Eq. 1), has attracted attention since CO is more active than CO2 and can be further hydrogenated to variety of products [9–13]. (1) C O 2 + H 2   ↔ C O + H 2 O Δ H   =   41.2   k J . m o l - 1 After conversion to CO, CO2 can be converted into light olefins as well as jet fuel through further hydrogenation in the Fischer-Tropsch reaction [14]. In addition, the RWGS is part of a pathway to form methanol as a CO2 hydrogenation product [15–19]. Many catalysts have been used for CO2 conversion to CO. In general, transition metals of the groups 8–10 such as Ni, Pd, Ru, and Rh have been shown to form both CO and CH4, while group 11 metals such as Cu, Ag and Au produce CO more selectively [20]. In particular, Cu-based and Mo-based catalysts were found to catalyze the almost exclusive formation of CO [9,21–25]. Metal-carbides, nitrides, and phosphides can also form active structures for CO2 hydrogenation reactions [26–28]. Among various catalysts, certain patterns have been observed. For instance, although nanoparticles of group 8–10 transition metals form both CO and CH4, atomic dispersion of the metals shifts their selectivity towards CO formation [29–31]. However, the oxidation state of the metal sites can also affect the product selectivity [32]. The addition of promoters to improve the catalyst activity and stability was also studied in detail [33–42]. Reina’s group investigated the addition of Cu and Cs promoters to Mo2C to increase the catalyst activity for the RWGS reaction [12,13]. The addition of Cu added more active sites, whereas the addition of electropositive Cs shifted the electron density favorably. The addition of alkali metals as electropositive elements has the same positive shifting effect on the electron density in the case of the RWGS reaction [43]. Other transition metals were used to form bimetallic catalysts, which promote the CO selectivity and/or catalyst activity for CO2 hydrogenation [33,44]. Ni-based catalysts have been studied extensively for hydrogenation reactions, since they are relatively inexpensive and active materials. They tend to form both CO and CH4, but are mostly used for catalyzing the CO2 conversion to CH4 (Sabatier reaction). If Ni-based catalysts could be tuned to selectively produce CO, they would be a favorable RWGS catalysts since they are more robust at higher temperatures compared to Cu-based catalysts [45]. This is an important property since the endothermic RWGS reaction requires high temperatures, at which Cu-based catalysts suffer from low stability [9]. There have been successful attempts to shift the selectivity of CO2 hydrogenation toward CO on modified Ni-based catalysts. Le Saché et al. synthesized a Ru-Ni/CeO2-ZrO2 catalyst with which they achieved 91 % CO selectivity, but only at high temperatures (750 °C) [46]. Braga et al. also promoted Ni by addition of Pd. They managed to reach up to 45 % CO selectivity at 600 °C [47].In this context, Cammarota et al. noticed that the addition of Ga to Ni can stabilize formate on Ni during the CO2 hydrogenation reaction [48]. Formate has been reported to be one of the intermediates in the RWGS. Although this study was carried out with homogeneous catalysts and under high pressures (34 atm), it motivated us to consider the addition of Ga to Ni to form a Ni-based catalyst with potentially high selectivity towards CO. Following this concept, we prepared a Ni-Ga catalyst on an alumina support that produces CO almost completely selectively (98 % selectivity) at a relatively low temperature (400 °C). We also studied the reaction mechanism on both Ni/Al2O3 and Ni-Ga/Al2O3 using in-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), from which we concluded that the presence of hydroxyl groups on the surface strongly influences the activity of the catalyst in CO2 conversion.The supported Ni-Ga catalysts as well as the control catalyst (Ni/Al2O3) were synthesized by means of the incipient wetness impregnation method based on a previously published study [49]. The desired amounts of Ni(NO3)2∙6H2O (Sigma-Aldrich) and Ga(NO3)3∙9H2O (Sigma-Aldrich) were used as metal precursors and γ-alumina (Merck) was used as the support. For instance, 2.48 g of nickel precursor and 3 g of gallium precursor was used for the preparation of 1Ni-1Ga/Al2O3. The amount of Ni was constant for all catalysts while the amount of Ga was varied. The procedure for the synthesis of Ga/Al2O3 was the same as for the other catalysts except that no Ni was added to this catalyst. The actual loadings are presented in Table S1 in the Supporting information. The catalysts were dried at 120 °C overnight and calcined at 700 °C for 6 h with a temperature ramp of 5 °C min−1.The catalytic activity tests were performed using a laboratory test bench with a fixed-bed quartz tube reactor with heating wires and a temperature controller. The outlet stream was analyzed with an online MATRIX-MG01 FTIR spectrometer (Bruker) with a 10 cm gas cell, heated at 120 °C. The spectrometer was operated with the OPUS-GA software to evaluate the spectra. For each test, 100 mg of catalyst was fixed in the reactor using quartz wool. The catalyst was reduced at the desired temperature (heating ramp 10 °C min−1) in-situ for 1 h under 20 % H2 flow diluted in Ar prior to each test. After reduction, the temperature was set to 400 °C and a mixture of CO2 and H2 diluted in Ar was dosed into the reactor. The CO2:H2 ratio was set to 1:4 with a weight hourly space velocity (WHSV) of 30,000 mL∙gcat −1 h−1.For the OH passivation test on Ni/Al2O3, the same procedure as above was used. However, after reduction at 700 °C, the catalyst was exposed to 10 mL min−1 of CO diluted in Ar. The temperature was set to 350 °C to avoid the formation of nickel tetracarbonyl (Ni(CO)4), which forms at temperatures lower than 230 °C [50]. During CO introduction, the formation of CO2 was monitored and when no more CO2 was detected in the in-line FTIR, the reactor was purged with Ar for 30 min. The temperature was then set to 700 °C with a 10 °C min−1 ramp to desorb potentially adsorbed CO from the surface of the catalyst. The temperature was set again to 400 °C and the same reaction mixture of CO2 and H2 was used at the above mentioned conditions. At the end of the test, the Ni content of the sample was checked with ICP-OES to ensure no Ni was lost in form of Ni(CO)4 during the passivation process (Table S1).Barrett-Joyner-Halenda (BJH) mesoporous volumes and Brunauer-Emmett-Teller (BET) surface areas were calculated from N2 physisorption isotherms collected with a Micromeritics 3Flex instrument. Before measurements, all materials were degassed overnight under vacuum (<10−3 mbar) at 120 °C (10 °C min−1 ramp rate).Hydrogen Temperature Programmed Reduction (H2-TPR) profiles was recorded on a Micromeritics Autochem 2920 II instrument. Typically, the samples (∼400 mg) were diluted with silicon carbide (∼250 mg), loaded into a quartz U-shaped quartz cell, and dried under He (50 mL min−1) at 150 °C. After cooling to 40 °C under He, the samples were heated to 900 °C at a ramp of 10 °C min−1 under a flow of 10 % H2 with balance Ar (50 mL min−1). The effluent gasses were passed through a cold trap with a mixture of liquid N2 and ethanol, and H2 consumption was monitored with a TCD.Additional H2-TPR experiments were carried out using a slightly modified procedure. Fresh samples (∼100 mg diluted in 500 mg silicon carbide) were dried and cooled under He as before. Then, the samples were heated to 700 °C (10 °C min−1) and held at temperature for 1 h under a flow of dilute H2 (10 % H2, bal. Ar, 50 mL min−1). The samples were cooled under He, and reheated again at 10 °C min−1 to 900 °C under dilute H2.Diffuse reflectance infrared Fourier transform (DRIFT) spectra were collected using a Bruker Vertex70 spectrometer equipped with a liquid nitrogen-cooled HgCdTe detector. The spectra were recorded from 4000 to 1000 cm−1 at a resolution of 4 cm−1 and scanner velocity of 80 kHz. The sample and background spectra resulted from averaging 10 and 100 scans, respectively. Approximately 30−40 mg of the catalyst sample were placed in a custom-built spectroscopic cell with a low void volume [51]. The cell was equipped with a 2-mm-thick CaF2 window (Crystran) and attached to a Praying Mantis™ accessory (Harrick Scientific) in the compartment of the IR spectrometer. Prior to the in-situ experiments, the sample was activated under 80 vol% H2/Ar flow for 30 min. The subsequent steps were as follows: (1) introduction of 7 vol% CO2 at 250 °C; (2) removal of 7 vol% CO2 at 250 °C with Ar; (3) ramp-up to 350 °C under Ar; (4) introduction of 7 vol% CO2 and 30 vol% H2 at 350 °C; (5) removal of 7 vol% CO2 and 30 vol% H2 at 350 °C.The X-ray photoelectron spectroscopy (XPS) measurements were carried out on an Axis Supra (Kratos Analytical) using the monochromated Kα X-ray line of an aluminum anode. The pass energy was set to 40 eV with a step size of 0.15 eV. The samples were electrically insulated from the sample holder and charges were compensated. Spectra were referenced at 284.8 eV using the C 1s orbital of the CC bond. Before XPS measurements, the samples were reduced for 1 h at 700 °C with the same flowrates as indicated in the catalytic experiments. The samples delivered to the instrument were prepared using a glovebox to ensure no air exposure occurred.Data for X-ray diffraction (XRD) were acquired using a D8 Bruker Discover diffractometer, which was equipped with a LynxEye XE detector as well as a non-monochromated Cu-source. The XRD patterns were measured for 2θ between 10 ° and 80 ° with the step size of 0.01.High-Angle Annular Dark-Field Scanning Transmission Electron Microscopy (HAADF-STEM) was conducted on a FEI Talos with 200 kV acceleration voltage in the Z contrast mode. Samples were dispersed in ethanol and placed on a carbon coated copper grid. Energy-Dispersive X-ray Spectroscopy (EDXS) analysis was performed using Bruker Esprit software.While Ni nanoparticles tend to dissociate C and O on their surface, addition of Ga to Ni has been shown to suppress this effect [49]. In order to study the effect of Ga addition to Ni on its catalytic activity and product selectivity, we synthesized three alumina-supported catalysts with various Ni to Ga molar ratios, i.e. 1Ni–1Ga/Al2O3 (Ni:Ga = 1:1), 2Ni–1Ga/Al2O3 (Ni:Ga = 2:1), and 1Ni–2Ga/Al2O3 (Ni:Ga = 1:2), and tested them for CO2 hydrogenation. The measured CO2 conversions and the selectivities to CO and CH4 are presented in Fig. S1 in the Supporting information. The CO selectivity of these catalysts followed the order: 1Ni–2Ga/Al2O3 ≅ 1Ni–1Ga/Al2O3 > 2Ni–1Ga/Al2O3 (selectivities are reported in Table S2) and the CO2 conversion was almost the same for all three, but was slightly higher for 1Ni–2Ga/Al2O3 (9 %) and 2Ni–1Ga/Al2O3 (9.5 %) compared to 1Ni–1Ga/Al2O3 (7 %). This means that the overall best performing candidate in terms of both activity and selectivity was 1Ni–2Ga/Al2O3, which was selected for further studies. The CO2 conversion rate of these studied catalysts are presented in Table S2 for comparison. The structure of these catalysts as well as the reference Ni/Al2O3 catalysts were studied by X-ray Diffraction (XRD) and the patterns are presented in Fig. S2. No clear peaks of Ni or NiO were detected, which supports the high dispersion of Ni on all catalysts. Fig. 1 shows the CO2 conversion and CO selectivity for 1Ni–2Ga/Al2O3 and Ni/Al2O3 as the reference catalyst. Ga/Al2O3 was also tested as a reference but only insignificant CO2 conversion (<2 %) was detected on this catalyst. The low conversion of CO2 on Ga/Al2O3 is caused by the lack of active nickel sites for H2 adsorption and dissociation. While the CO2 conversion was clearly higher using Ni/Al2O3 compared to 1Ni–2Ga/Al2O3 (∼30 % as opposed to 10 %), its CO selectivity was significantly lower compared to 1Ni–2Ga/Al2O3 (∼40 % as opposed to 98 %). The addition of Ga caused the shift of selectivity to CO, resulting in the observed decreased conversion level since the methanation reaction almost ceased. To compare the CO selectivity using various catalysts, they should be tested in the same range of CO2 conversion. Therefore, we have decreased the WHSV for 1Ni-2Ga/Al2O3 (Fig. S3), to bring its level of CO2 conversion close to the one reported for Ni/Al2O3 in Fig. 1. We observed that the CO selectivity on this catalyst did not change compared to Fig. S1 despite its higher CO2 conversion.When using alumina-supported transition metals as catalysts, the RWGS reaction mechanism is known to proceed through CO2 adsorption (forming carbonate or bicarbonate) followed by its reaction with dissociated H atoms to form oxygenated intermediate products (formate or carboxylate). The formed intermediate product then decomposes to form CO and H2O [52–55]. In order to understand the reasons behind the differences in the performance of the presented catalysts, we studied the reaction mechanisms by means of in-situ DRIFTS experiments. The IR spectra were collected when introducing and purging CO2 as well as during the CO2 hydrogenation reaction and reactant cut-off. Fig. 2 a and b shows the IR spectra obtained during the adsorption of CO2 on 1Ni-2Ga/Al2O3 and Ni/Al2O3. The persistence of the peaks at 2344 and 2358 cm−1 reflects the continuous presence of gas-phase CO2. For both samples, we observed two peaks at 1649 cm−1 and 1442 cm−1, which together with the weak signal at 1220 cm−1, are characteristic for the formation of bicarbonates [54] formed through the reaction of CO2 with surface hydroxyl groups. However, the area of the bicarbonate peak at 1649 cm−1 over Ni/Al2O3 was up to 23 % larger than that over 1Ni-2Ga/Al2O3, suggesting that Ga incorporation results in less CO2 uptake and adsorption (Fig. S4). Since bicarbonates are formed due to the interaction of CO2 with surface hydroxyl groups, this observation points out that there are fewer hydroxyl species in the Ga-containing sample. Indeed, the single-beam spectrum of 1Ni-2Ga/Al2O3 showed comparatively less hydroxyl groups than on Ni/Al2O3 (Fig. S5). Therefore, it is reasonable to conclude that the surface hydroxyl groups control the adsorption CO2 and the formation of bicarbonate as the intermediate product. This observation explains the lower activity of 1Ni-2Ga/Al2O3 for CO2 hydrogenation. Fig. 2c and d shows the spectra obtained during desorption of CO2 from 1Ni-2Ga/Al2O3 and Ni/Al2O3. The abrupt disappearance of the gas-phase CO2 peaks was accompanied by a much slower extinction of the peaks at 1649 cm−1 and 1442 cm−1, which suggests the desorption of bicarbonate species. The bicarbonates almost fully desorbed from 1Ni-2Ga/Al2O3 after 10 min of purging while a significant portion remained bound to Ni/Al2O3. According to literature, stronger adsorption of the intermediate products results in their further hydrogenation to CH4 [56]. This explains the increased CH4 formation on Ni/Al2O3, where residual bicarbonate species persisted after CO2 removal. In contrast, only weakly adsorbed CO2 participates in the formation of CO through the RWGS [24]. Hence, the weaker interaction of 1Ni-2Ga/Al2O3 with the bicarbonate species proved to be beneficial in driving the selectivity towards CO.The weak adsorption of CO2 on 1Ni-2Ga/Al2O3 can be explained by the structure of this catalyst. Fig. 3 shows the STEM images of the 1Ni-2Ga/Al2O3 catalyst. While both small and relatively large Ni nanoparticles were formed on Ni/Al2O3 (Fig. 4 ), a uniform dispersion of Ni and Ga was observed on 1Ni-2Ga/Al2O3. The STEM images in Fig. 3 suggest that Ni and Ga have mostly covered the surface of alumina since the blue color representing Al is not prominent on the edge of the STEM images on Fig. 3. Highly dispersed transition metals can only weakly adsorb the intermediate products of CO2 hydrogenation and therefore, selectively form CO [31].The aforementioned alumina surface coverage is also supported by N2 physisorption results (Figs. S6 and S7 as well as Table 1 ). Through comparison of the BET surface area (SBET) and pore volume (Vp), we noted that impregnation of Ni on alumina did not change the initial surface area (120 m2 g−1for Al2O3 and 117 m2 g−1 for Ni/Al2O3) and pore volume (0.232 cm3 g−1 for Al2O3 and 0.226 cm3 g−1 for Ni/Al2O3) of alumina. However, for 1Ni-2Ga/Al2O3, both SBET and Vp decreased to 102 m2 g−1 and 0.1798 cm3 g−1, respectively. Although this decrease in surface area as well as pore volume is not very pronounced, it may be due to the formation of a Ni-Ga entities on the catalyst surface, which blocked some of the pores and consequently decreased the surface area. Therefore, we hypothesize that the formation of the Ni-Ga layer may have decreased the availability of the Al2O3 surface and caused the aforementioned decrease in hydroxyl group concentration. This hypothesis was also checked using Al 2p XPS spectra (Fig. S8). It can be noted that the signal intensity for Al 2p decreased for 1Ni-2Ga/Al2O3 compared to Ni/Al2O3, which supports the formation of the Ni-Ga layer.The CO2 hydrogenation reaction was studied with IR spectroscopy and the results are shown below (Fig. 5 ). After introduction of H2 on both catalysts, the peaks at 1649 cm−1 and 1442 cm-1 started to disappear and peaks at 1595 cm−1, 1393 cm−1 and 1375 cm−1 formed, which indicate the presence of formate species [57]. We concluded that bicarbonate was hydrogenated to formate upon H2 introduction, in line with previously reported CO2 hydrogenation studies on Au/Al2O3 [52]. A higher formate concentration was detected on 1Ni-2Ga/Al2O3 compared to Ni/Al2O3, as evidenced by the higher ratio of the peak areas for formate (1595 cm−1) and bicarbonate (1649 cm−1). This means that nickel alone is more capable of activating adsorbed formate species than Ni-Ga, resulting in the observed higher concentration of surface formate on the Ga-treated catalyst. This explanation is also confirmed by the spectral response upon gas cut-off (Fig. 5c and d). When the gas flows were cut off, unreacted formate on both catalyst surfaces were observed, which suggests that at least some formate species remains on the catalyst surface as spectator species, most probably those formate species far distant from Ni. Formate is also known to be a potential intermediate for the methanation reaction [56]. Therefore, we propose that the formates that remain unreacted on the 1Ni-2Ga/Al2O3 surface, poisoned some of the active sites, which resulted in the lower activity of this catalyst. The same species react further to form methane on the Ni/Al2O3 catalyst, which lowers the relative selectivity of this catalyst compared to those containing Ga.Based on the results obtained from the in-situ investigation of the catalysts, the concentration of hydroxyl groups were found to be correlated with the adsorption of CO2. To further test the importance of these surface groups for CO2 hydrogenation, we passivated the hydroxyl groups on the surface of Ni/Al2O3 using an approach described by Yang et al. [58]. Their method consists of reacting the OH groups with CO to remove them from the catalyst surface through the water-gas shift reaction (WGS) (details are described in Section 2.2). Fig. 6 shows the activity of Ni/Al2O3 for CO2 hydrogenation with and without passivation with CO. Importantly, the CO2 conversion on the CO-passivated catalyst decreased compared to the non-passivated catalyst. This further confirms the effect of the hydroxyl groups on adsorption of CO2. Both the IR spectroscopy results and the CO passivation results indicate that the presence of hydroxyl groups facilitates the CO2 adsorption through formation of bicarbonate, which in turn results in higher CO2 conversion. Al2O3 not only provides a large surface area for the dispersion of metallic active sites for the adsorption and dissociation of H2, but also supports the presence of hydroxyl groups on the catalyst surface that can actively react with CO2 and thus participate in the reaction.To better understand the relationship of structure and activity, the surfaces of 1Ni-2Ga/Al2O3 and Ni/Al2O3 were investigated by XPS. Since the catalysts were reduced in-situ before the catalytic tests, we reduced the catalysts under the same conditions before the XPS measurements. The Ni 2p spectra for both catalysts are presented in Fig. 7 . The presence of Ga on the surface of 1Ni-2Ga/Al2O3 is evident from the Ga 2p spectra (Fig. S9). Ni on the surface of 1Ni-2Ga/Al2O3 appears to be mostly reduced and present in metallic form. However, the Ni 2p spectrum for Ni/Al2O3, shows that Ni was not fully reduced on this catalyst, where it was predominantly present as NiOOH and Ni(OH)2. Ni 2p fitting was performed based on the approach of Biesinger et al. [59]. While CO2 adsorption on Ni/Al2O3 could be facilitated due to the presence of the hydroxyl groups, H2 adsorption and activation is less favorable in the absence of metallic Ni. Considering the similar Ni loading and the same reduction conditions, this observation demonstrates that Ga promotes the reducibility of Ni.The reducibility of these catalysts were further studied in H2-TPR experiments. In our first experiment, the calcined catalysts were reduced up to 900 °C. It is clearly visible from Fig. 8 that the reduction of 1Ni-2Ga/Al2O3 starts at lower temperatures compared to Ni/Al2O3 (∼520 °C as opposed to ∼600 °C). This confirms the promoting role of Ga for the reducibility of Ni. In the second experiment, we conducted H2-TPR on both catalysts after reducing them under the same conditions used during the catalytic tests (20 vol% H2/He flow at 700 °C for 1 h). The results showed that both catalysts were partly reduced during this treatment (Fig. S10). However, by comparison with the H2-TPR profiles for the calcined catalysts, we observed that most of the Ni on 1Ni-2Ga/Al2O3 was already reduced during the pre-treatment, while a notable portion of the Ni sites of Ni/Al2O3 catalyst remained unreduced after the pre-treatment. This is also in agreement with the aforementioned XPS results (Fig. 7).As observed in the H2-TPR profiles, the peak of H2 consumption for both catalysts occurred at 830 °C, which ensured the full reduction of Ni sites. Since, based on XPS results, the Ni sites on the surface of 1Ni-2Ga/Al2O3 were mostly reduced, we did not expect much variation in catalyst activity after reduction at 830 °C. For Ni/Al2O3, however, a higher catalyst activity would be expected compared to the treatment at 700 °C, since metallic Ni has higher H2 adsorption and dissociation activity. To study these effects, we performed the same catalytic test but with in-situ catalyst reduction at 830 °C for 1 h prior to the test (Fig. 9 ). The Ni/Al2O3 catalyst activity increased after this treatment (initial CO2 conversion was 33 % as opposed to 26 % in Fig. 1). As expected, for 1Ni-2Ga/Al2O3 catalyst, little change was observed. Our results and corresponding discussions in the literature confirm that two groups of active sites are required for the RWGS reaction. One serves for adsorption and dissociation of H2 molecules, the other for adsorption of CO2. The metal sites (e.g., on Ni/Al2O3 and 1Ni-2Ga/Al2O3) are responsible for the adsorption and dissociation of H2. A lack of these sites is the reason for the low activity of Al2O3 or Ga2O3/Al2O3. However, for high selectivity of Ni-based catalysts in CO formation, high dispersion of the metallic sites is required. Otherwise, accumulation of the metallic Ni sites may lead to the formation of CH4. On the other hand, the presence of hydroxyl groups on the metal oxide (Al2O3 in this case) is required to adsorb CO2, as shown by the DRIFTS study.We did not observe any deactivation in any of the catalytic tests we performed. However, we investigated the spent catalysts with XRD for comparison. No diffraction peak for Ni or NiO was observed for both the fresh and spent catalysts (Fig. S11). For 1Ni-2Ga/Al2O3, no sign of Ga was observed in the diffraction patterns. This suggests that no large crystalline nanoparticles (> 3 nm) of Ni, NiO or Ga2O3 were formed in the fresh or spent catalysts either during the calcination, reduction, or reaction. To confirm these observations, we also took STEM images of the 1Ni-2Ga/Al2O3 after 20 h of reaction (Fig. S12). No sign of Ni and/or Ga agglomeration and sintering could be detected, confirming that 1Ni-2Ga/Al2O3 did not sinter during the reaction.Ni-based catalysts are known for their tendency to form CH4 during the CO2 hydrogenation reaction. Although they are usually more thermally robust compared to Cu-based catalysts, they are not used for the RWGS due to their low selectivity for CO. This was also confirmed by our experiments using a Ni/Al2O3 catalyst. However, our present study showed that addition of Ga can restructure Ni particles on the surface of alumina in a way, which increased the selectivity for CO. Although this shift in selectivity came at the expense of lower catalyst activity, selective formation of CO eliminates the need for downstream separation processing in industrial applications. The lower CO2 conversion could be compensated by recycling the unreacted gases to increase the overall conversion. Using in-situ DRIFTS studies as well as other catalyst characterization methods showed that the addition of Ga led to high dispersion of Ni and high surface coverage of alumina, which limited the availability of hydroxyl groups on the surface. Our results support the view that hydroxyl groups are crucial species in the CO2 hydrogenation mechanism since they are the active sites for CO2 adsorption and formation of bicarbonates and their presence contributed to the higher activity of Ni/Al2O3 compared to 1Ni-2Ga/Al2O3. Although Ga is more expensive compared to Ni and the addition of Ga does not seem economically promising, this disadvantage is compensated by the better temperature stability of these catalysts, which might help the process in the long run. Nevertheless, further studies will be required to increase their activity. Ali M. Bahmanpour: Conceptualization, Methodology, Formal analysis, Investigation, Validation, Writing – original draft, Data curation. Rob Jeremiah G. Nuguid: Investigation, Data curation, Writing – original draft. Louisa M. Savereide: Investigation, Data curation, Validation, Writing – original draft. Mounir D. Mensi: Investigation, Data curation, Writing – original draft. Davide Ferri: Supervision, Resources, Writing – review & editing. Jeremy S. Luterbacher: Supervision, Resources, Writing – review & editing. Oliver Kröcher: Supervision, Resources, Project administration, Writing – review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.JSL, and LMS acknowledge funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (Starting grant: CATACOAT, No. 758653). RJGN, DF, and O.K. acknowledge funding from the Swiss National Science Foundation (SNF, #172669). The authors would like to acknowledge Dr. Pascal Schouwink for XRD analysis, and Mr Sylvain Coudret for ICP-OES analysis.Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jcou.2021.101881.The following is Supplementary data to this article:

Ni/Al2O3 is an active catalysts for CO2 hydrogenation to both CH4 and CO. By doping with Ga, we succeeded in shifting the selectivity of these catalysts almost completely toward CO. In-situ IR spectroscopy studies showed that the catalyst activity is directly related to the concentration of surface hydroxyl groups that are responsible for the adsorption of CO2 and the formation of intermediate bicarbonates and formates on the catalyst surface. The addition of Ga improved the Ni dispersion which was concomitant with the formation of a Ni-Ga layer on the surface of alumina, thereby reducing the surface hydroxyl concentration. The reduced and weakened interaction between the intermediate products, i.e. bicarbonates and formates, and the catalyst surface, increased the CO selectivity from ∼40 % to 98 %.

Data will be made available on request.Ethylene is mainly produced at industrial scale by steam cracking of naphtha, LPG or natural gas condensates at high temperature (between 750 and 875 °C) and short contact time, resulting in products that are difficult to separate [1,2]. This process presents a low energy efficiency and it is considered as one of the most energy-consuming process in the chemical industry [3,4], with a massive CO2 formation of ca. 1.8 kgCO2/kgEthylene [4].The oxidative dehydrogenation (ODH) of ethane could be a low demanding energy alternative, that can operate at temperatures up to 400 °C, with high selectivity to ethylene [1,5–7]. The most promising catalytic systems for the ODH of ethane are multicomponent MoVTeNbO catalysts [8–10] and NiO based catalysts [11–25].In the last 15 years, Me-doped NiO catalysts, especially Nb5+- [11–17], but also W6+- [18], Sn4+- [19–21], Zr4+- [22], Ti4+-doped catalysts [20,21], or supported NiO catalysts [22–25], have been proposed as highly selective in ethane ODH when working at moderate ethane conversions. In this way, it has been reported that the incorporation of high valence state cations, especially Nb5+, Ta5+, Sn4+ and W6+ into the NiO lattice, decreases the amount of electrophilic O- in NiO and improves the selectivity to ethylene during the ethane ODH [21,26–29]. In an opposite trend, the incorporation of low valence state cations (such as Li+ [30] or K+ [26]) favors the deep oxidation of ethane. Recently, it has been proposed that, by tuning the Ni3+ quantity in NiO, the reaction pathway can be modulated: low Ni3+ concentration favors the ethane ODH whereas high Ni3+ concentrations lead to the deep oxidation of hydrocarbons [30]. Furthermore, other authors suggest that a high-valence dopant in an irreducible oxide will increase the bonding of neighboring surface oxygen atoms to the oxide (which would hinder oxidation reactions through a redox mechanism), however, this behavior could change depending on the atmosphere (presence/absence of oxygen in the feed) [31].By using density functional theory, the dissociative adsorption of ethane on the surface of a Nb-doped nickel oxide catalyst has been studied [32]. Thus, it has been suggested that Nb species (or NbO2 groups) substituting Ni atoms in the surface layer of NiO could favor the strong adsorption of the O2 fed in the gas phase. In this way, the presence of Nb favors a low oxidant capacity of oxygen atoms and could act as a lower-valence dopant, activating also the surface oxygen atoms surrounding these Nb-sites [32].In addition to this, NiO is a p-type semiconductor [33–35] with high catalytic activity for the deep oxidation of hydrocarbons and CO [35,36]. Lemonidou et al. characterized pure and doped nickel oxide catalysts with Li, Mg, Al, Ga, Ti, and Nb determining their electrical conductivity. This study showed that all undoped and doped materials were p-type semiconductors (before and after reaction), with positive holes as the main charge carriers, where a correlation between the p-type semiconductivity and the catalytic performance was observed [34]: the lower the p-type semiconductivity the higher was the selectivity to ethylene. Additionally, these authors concluded that the reaction mechanism involves surface lattice O- species operating through a Mars-van Krevelen mechanism. On the other hand, in the case of Nb-doped NiO catalysts prepared by evaporation, the electrical conductivity decreased when the Nb-content increased. This correlates well with their intrinsic rates of ethane consumption [35].Furthermore, and as it has been proposed in the case of pure NiO [37–42], the catalytic performance of doped catalysts [11–22,43], strongly depends on the catalyst preparation method.Recently, it has been proposed that the presence of oxalic acid in the synthesis gel changes the textural and the catalytic properties of NiO [41]. This fact is explained as a consequence of changes in the catalyst precursors, the formation of Nickel (II) oxalate dihydrate being paramount to achieve high selectivity to ethylene.ODH of ethane on Nb-doped NiO catalysts proceeds through a redox mechanism. Therefore, the study of the electrochemical properties of these samples is highly interesting. However, there are just a couple of articles about NiO based catalysts that have dealt with the conductivity of p-type semiconductors, and they have related this fact with the catalytic behavior in the ODH of ethane [34,35]. Then, we have decided to undertake a deeper electrochemical study of Nb-doped NiO catalysts in order to check if a correlation between electrochemical properties and catalytic performance does exist. Thus, the technique of Electrochemical Impedance Spectroscopy (EIS) has been applied, which is a powerful non-destructive technique typically used in electrochemistry and from which the resistance values of the catalysts can be properly determined. Moreover, we have also undertaken capacitance analysis (Mott-Schottky analysis) to verify the type and extent of the semiconductor behavior. Finally, cyclic voltammetries have been carried out in order to study the electrochemical activity of the different synthesized catalysts.The obtained results show that the presence or absence of oxalic acid in the synthesis gel as well as the calcination temperature have a strong influence on both the incorporation of Nb5+ atoms in the framework of NiO and on the catalytic performance in ethane ODH. In addition, a change in the semiconducting character from p- to n-type is observed in the Nb-doped NiO catalysts prepared in specific conditions (i.e. samples synthesized in the presence of oxalic acid in the synthesis gel and calcined at 500 °C), being these catalysts the ones that presented the best catalytic performance.Nb-promoted nickel oxides catalysts (with a Ni/Nb molar ratio of 9/1) were prepared through evaporation at 90 °C of aqueous solutions of nickel nitrate (Ni(NO3)2·6H2O, Sigma-Aldrich) and ammonium niobate (V) oxalate hydrate (C4H4NNbO9·xH2O, Sigma-Aldrich) with different amount of oxalic acid (C2H2O4, Sigma-Aldrich) in the synthesis gel. The solids were dried overnight in a furnace at 120 °C and, finally, they were calcined in static air for 2 h at 350 or 500 °C. Unpromoted nickel oxide catalysts have been also prepared by using the same procedure, for comparison [41]. Catalysts are named as NiNb/x-T, where × is the oxalic acid/nickel (OxA/Ni) ratio in the synthesis gel of 0.0, 1.0 or 3.0, and T is the calcination temperature (350 or 500 °C). The as-synthesized samples (before calcination) are named as NiNb/x-as. Nomenclature and physicochemical properties of Nb-promoted and unpromoted catalysts are shown in Table 1 and Table S1, respectively.The specific surface areas were estimated by the Brunauer-Emmet-Teller (BET) method from N2 adsorption isotherms at 77 K measured in a Micromeritics TriStar 3000 instrument.X-ray diffraction patterns were recorded with a PANalytical CUBIX instrument equipped with a graphite monochromator, by using Cu Kα radiation (λ = 0.1542 mm) and operating at 45 kV and 4 mA.High Resolution Transmission Electron Microscopy (HRTEM) was performed on a JEOL JEM300F electron microscope by working at 300 kV (point resolution of 0.17 nm). Crystal-by-crystal chemical microanalysis were performed by energy-dispersive X-ray spectroscopy (XEDS) in the same microscope equipped with an ISIS 300 X-ray microanalysis system (Oxford Instruments) with a detector model LINK “Pentafet” (resolution 135 eV). Samples for transmission electron microscopy (TEM) were ultrasonically dispersed in n-butanol and transferred to carbon coated copper grids.Raman spectra were collected with an “in via” Renishaw spectrometer equipped with an Olympus microscope. The samples were excited by the 514.5 nm line of an Ar+ laser (Spectra Physics model 171) with a laser power of 2.5 mW or by the 325 nm line (UV-Raman) generated with a Renishaw HPNIR laser with a power of approximately 15 mW.Temperature-programmed reduction experiments (TPR-H2) were carried out in an Autochem 2910 (Micromeritics) equipped with a TCD detector. The reducing gas composition was 10 % H2 in Ar (total flow rate of 50 mL min−1), and using a heating rate of 10 °C min−1 until 800 °C.X-ray photoelectron spectroscopy (XPS) measurements were performed on a SPECS spectrometer equipped with a Phoibos 150 MCD-9 detector using a monochromatic Al Kα (1486.6 eV) X-ray source. Spectra were recorded using an analyzer pass energy of 50 eV, an X-ray power of 100 W, and an operating pressure of 10-9 mbar. Spectra treatment was performed using CASA software. Binding energies (BE) were referenced to C 1s at 284.5 eV.Electrochemical interfacial properties of the catalysts were studied in a three-electrode electrochemical cell with an Ag/AgCl 3 M KCl reference electrode and a platinum wire as counter electrode. The catalysts were connected to the potentiostat as working electrodes with 0.5 cm2 of exposed area to the electrolyte. Mott-Schottky (MS) measurements, Electrochemical Impedance Spectroscopy (EIS) tests and cyclic voltammetries were carried out to characterize the catalysts, specifically their electrochemical properties at the interface between the catalysts (electrodes) and the electrolyte. Mott-Schottky plots were performed at a frequency of 5000 Hz, scanning the potential from 1 to −0.5 VAg/AgCl at 0.05 V/s using an amplitude signal of 0.01 V. Before EIS measurements, catalysts were immersed in the solution for 1800 s at 0.5 VAg/AgCl. After this pre-treatment, EIS tests were carried out applying a potential of 0.5 VAg/AgCl using an amplitude of 0.01 V and scanning frequencies from 100 kHz to 0.01 Hz. Cyclic voltammetries were performed in the potential range of −0.1 to 0.6 VAg/AgCl at 0.01 V/s. The electrolyte used for MS and EIS tests was 0.1 M of Na2(SO)4, while for the cyclic voltammetries, a solution of 10 mM of Fe(CN)6K4 with 0.1 M Na2SO4 was employed.The oxidative dehydrogenation of ethane was carried out in an isothermal fixed-bed quartz reactor, at atmospheric pressure, in the range of 300–350 °C. The feed consisted of an ethane/O2/He mixture with 3/1/26 M ratio. The total flow and the catalyst’s weight were varied (25–100 mL min−1 and 0.1–1.0 g of catalyst) in order to achieve several contact times. Reactants and products were analyzed by gas chromatography using two packed columns [25]: (i) molecular sieve 5A (2.5 m); and (ii) Porapak Q (3 m). Fig. 1 A shows the XRD patterns of Nb-doped nickel oxide catalysts calcined at 350 or 500 °C. In all cases, the presence of diffraction maxima at 2θ = 37.10°, 43.30°, 62.86°, 76.50° and 79.22° can be indexed to a face-centered cubic NiO phase (JCPDS-ICDD pattern number 47–1049, space group Fm3m). No additional maxima related to Nb-compounds were detected. XRD patterns of samples calcined at 500 °C present peaks with higher intensities, which suggest a higher degree of crystallinity and/or bigger size of NiO crystals (Table 1) [38].This is in agreement to previous results, in which the size of NiO particles in Nb-doped nickel oxide catalysts is smaller than those of the corresponding pure NiO catalysts [11,15]. This lower crystal size is attributed to the incorporation of Nb5+ into the NiO network which hinders crystallization thus leading to smaller crystallites. In addition, the presence of additional oxalic acid in the synthesis gel promotes a decrease in crystal size, in a way that increasing the concentration of oxalic acid in the synthesis gel gradually decreases the intensity of the NiO diffraction maxima. Thus, the average crystallite size estimated by the Scherrer equation shows a variation from 20 nm in the catalyst prepared without oxalic acid in the synthetic gel to approximately 10 nm in the catalysts prepared with oxalic acid (Table 1).X-ray diffraction patterns of as-synthesized catalysts (Fig. S1A, in Supporting Information) confirm the presence of Nickel (II) oxalate dihydrate (NiC2O4⋅2H2O) and Nb-oxalate in samples prepared with oxalic acid in the synthesis gel (NiNb/1-as and NiNb/3-as), whereas Nickel(II) nitrate hexahydrate (Ni(NO3)2⋅6 H2O and Nb-oxalate are only observed in the sample prepared in the absence of additional oxalic acid in the synthesis gel (i.e. NiNb/0-as). As a consequence of changes in the nature of reactants during the synthesis procedure, changes in the thermogravimetric analysis were also observed in the decomposition of precursors (Fig. S1, B and C): i) the decomposition of Ni-nitrate corresponds to a low intense broad peak at ca. 300 °C for sample NiNb/0-as; ii) the decomposition of Ni-oxalate shows a narrow peak (very intense) at ca. 350 °C for samples prepared with oxalic acid in the synthesis gel (i.e. NiNb/1-as and NiNb/3-as); iii) changes in the nature of exothermic peak depending on the concentration of oxalic acid in the synthesis gel. These results are similar to those reported for pure NiO [41], except that a decomposition peak at ca. 250 °C, related to the decomposition of Nb-oxalate, is observed for sample NiNb/0-as. This is in agreement with previous results observed during the preparation of unpromoted nickel oxide in the presence/absence of oxalic acid in the synthesis gel [41]. Fig. 1B shows the UV Raman spectra of Nb-doped nickel oxide catalysts calcined at 350 or 500 °C. Several bands at 571, 724, 901 and 1128 cm−1 [44–46] have been observed (see Supporting Information for the assignment of the bands). In general, the intensities of these bands change depending on the preparation method and/or the calcination temperature of catalysts. According to the spectra, the catalyst prepared in the absence of oxalic acid in the synthesis gel and calcined at 500 °C (NiNb/0–500) likely presents large NiO crystal size [46]. However, the samples prepared with oxalic acid and calcined at 350 °C (NiNb/1–350 and NiNb/3–350) probably presents a lower crystal size and/or a higher concentration of defects [47]. Accordingly, the different crystal size and/or the concentration of defects are mainly related to both the amount of oxalic acid in the synthesis gel and/or the calcination temperature.Raman spectra of all catalysts, using a 514 nm laser (Fig. S2) present a strong band at ca. 497 cm−1 (Ni-O stretching mode) [11,15,38], with a shoulder at 410 cm−1 (which could be related to the nonstoichiometry of catalysts [11,12]. In the case of NiNb/0–350 and NiNb/0–500 samples, the width of the band at 497 cm−1 increases with the calcination temperature. However, no significant changes are found for catalysts prepared with higher oxalic acid contents in the synthesis gel. On the other hand, bands at 790, 850 and 1071 cm−1 are also observed for sample NiNb/0–500. The band at 1071 cm−1 is related to the v1 vibration mode of carbonate groups on the NiO phase [16], whereas the band at ca. 840 cm−1 has been related to stretching vibrations of slightly distorted NbO6 octahedra in amorphous Nb2O5 [12]. The band at 790 cm−1, detected by other authors in Nb-doped NiO catalysts prepared in a similar way than that of sample NiNb/0–350 [15], has been related to vibration modes of bridging Ni-O-Nb bonds, suggesting the formation of a Ni–Nb solid solution.A detailed study of these catalysts was performed by High Resolution Transmission Electron Microscopy. Fig. 2 shows characteristic high-resolution images of the catalysts under study. The largest NiO crystals were obtained when oxalic acid was not added to the synthesis gel (Fig. 2a). For the oxalic free NiNb/0–500 catalyst the size of the crystallites ranges between 20 and 50 nm. Smaller particles (size ≤ 5 nm) of amorphous material were also visible in small amount. EDS nanoanalysis showed that the large NiO crystals present some Nb incorporated (typically 6 at. % of Nb) whereas the small particles present higher concentration of niobium (55 at. % of Nb). Corresponding spectra (Fig. 2a) include both Ni and Nb atomic percentages as determined from semiquantitative analysis. Note that the limited spatial resolution of the nanoprobe prevents to determine with absolute certainty if the segregation of niobium with respect to the NiO crystals is total in the selected zones, but the percentages indicate a very clear trend.The addition of oxalic acid to the synthesis gel implied a diminishing of the NiO crystal size, as previously observed by XRD. Fig. 2b shows a group of NiO crystals of NiNb/1–350 and, for that sample, crystallite size ranges between 5 and 10 nm. The amorphous particles are barely observed in this catalyst. Moreover, it must be noted that increasing the calcination temperature increases the particle size to a small extent. This fact can be seen in Fig. 2c, which shows a low magnification micrograph of several particles of NiNb/1–500 catalyst. As observed, this catalyst is formed by rod like particles longer than 1 μm which in fact are constituted by nanocrystals of about 20 nm. This appreciation probably suggests that a sintering process occurs when calcined at 500 °C. On the other hand, the high magnification high resolution micrograph for NiNb/1–500 sample and the enlarged detail observed in Fig. 2d show an average crystal size of ∼ 20 nm. The included EDS spectra show the atomic percentages of Ni and Nb in the two types of particles observed, revealing that the amorphous particles are mostly composed of niobium (81 at. %) whereas the large NiO crystals present a very low Nb-loading (0.6 at. %). This suggests that the presence of oxalic acid in the synthesis gel decreases the Nb-incorporation in the NiO lattice and leads to the formation of more small particles very rich in Nb on the surface of the NiO crystals.At this point it is important to mention that no significant differences were observed in the microstructure of the catalysts prepared with different concentrations of oxalic acid (Oxal/Ni ratios of 1 and 3).Nb-doped nickel oxide catalysts have been also characterized by diffuse reflectance UV–vis spectroscopy (DRS). Fig. 3 A groups together the DRS spectra of the Nb-doped catalysts calcined at 350 (black) or at 500 °C (red). According to the literature, all samples show typical NiO absorption bands [22,48] at 380, 416 and 722 nm (assigned to the octahedral Ni2+ in the NiO lattice [22]), 510 nm (attributed to charge transfer in NiO [49]) and in the 400–600 nm range, which have been linked to the presence of non-stoichiometric oxygen.Catalysts treated at 350 °C exhibit higher absorbance in the 400–600 nm range than catalysts treated at 500 °C. The high background absorbance in this region has been associated with an increase in non-stoichiometric oxygen concentration [37]. Therefore, samples calcined at 350 °C exhibit a higher concentration of non-stoichiometric oxygen species. Fig. 3B presents the H2-TPR profiles of catalysts calcined at 350 or 500 °C. The maximum of the reduction peak slightly shifts to lower temperatures when increasing the OxA/Ni ratio in the synthesis gel, especially for samples calcined at 500 °C. These patterns have reduction maxima centered between 277 and 366 °C, the relative intensity of which depends on the preparation method. The presence of several reduction peaks has been related to steps proposed for the reduction of NiO [50] but also to the presence of mixed phases linked to the Nb-promoter [12]. Indeed, the high temperature signals in supported NiO catalysts have been attributed to a strong influence of the promoter. In addition, the reduction peaks at low temperature (around 200 °C) observed in samples calcined at 350 °C could be related to the presence of undoped NiO particles [51], as suggested in the microscopy characterization. It has been proposed that non-stoichiometric nickel oxide exhibits a lower initial reduction activation energy than stoichiometric NiO, thus giving reduction signals at lower temperature [51]. The enlarged part of the area around 200 °C shows that the samples calcined at 350 °C have higher intensity. The use of oxalic acid in the synthetic gel leads to a decrease in its intensity. In addition, the reducibility of Ni-O bonds in Ni-Nb-O catalysts calcined at 500 °C (NiNb/x-500 series) is slightly lower than those of Ni-Nb catalysts calcined at 350 °C (NiNb/x-350 series). The reducibility of the catalysts with oxalic acid slightly increases compared to that without oxalic acid, being more evident for catalysts treated at 500 °C. When comparing these results with those achieved for undoped NiO catalysts [41], it can be concluded that the incorporation of Nb5+ promotes a lower reducibility of Ni2+ species.XPS experiments were undertaken for Nb-doped catalysts and the quantitative results of the analysis are presented in Table 2 . Fig. 4 shows Ni 2p3/2 spectra for Nb-doped NiO catalysts, whereas Fig. S3 presents the Nb 3d spectra of catalysts.In the case of the Ni 2p3/2 core level, spectra for all the samples can be deconvoluted into two signals at 853.2 and 855.3 eV, with a satellite peak at 860.2 eV. The peak at about 853.2 eV is related to the presence of structural Ni2+ species within the lattice, while the second signal (named Satellite I) at 1.5–2.0 eV above the main peak has been associated with the presence of structural defects (such as Ni3+, Ni2+–OH species or Ni2+ vacancies in the network [20,26,29]). In addition, a second larger satellite (named Satellite II), at about 860.2 eV, is related to metal–ligand charge transfer [20,26,29]. Although these three types of signals are clearly identified, notable differences can be observed among the catalysts. Thus, for catalysts calcined at 350 °C, the intensity of the signals seems to be lower than those obtained for the catalysts treated at 500 °C regardless of the preparation method.In addition, the catalysts prepared with oxalic acid in the synthetic gel, in any quantity, show a shift towards higher binding energies, which could be associated with different electrochemical properties of the catalysts.On the other hand, spectra at the Nb 3d level are shown in Fig. S3. Typically, a niobium spectrum consists of a doublet with a spin–orbit separation of 2.72 eV between the 3d5/2 and 3d3/2 components. A single well-defined doublet is observed for the NiNb/0–350 catalyst, with peaks at 206.7 (3d5/2 ) and at 209.4 eV (3d3/2 ), respectively, which corresponds to the Nb5+ species [52].For catalysts calcined at 500 °C, a well-defined doublet corresponding to the presence of Nb5+ species is also observed, regardless of the amount of oxalic acid in the synthesis gel (Fig. S3). Certain authors [53] suggest the presence of Nb4+ species when reducing agents (such as oxalic acid) are present, especially in higher amounts, however, this is not our case. As evidenced in Fig. S3, no noticeable differences can be ascertained from Nb 3d XPS analysis regarding the different synthesis method and thermal treatment in terms of chemical state. Nevertheless, quantitative differences are suggested since dissimilar intensities were observed depending on the sample.In this sense, the results in Table 1 suggest that the use of oxalic acid in the synthesis gel and high activation temperatures maximizes the presence of niobium species on the surface of the catalyst, finding the highest value for the NiNb/1–500 catalyst. Then, it is observed that the concentration of Nb on the surface of NiNb/1–500 is remarkably higher (4-fold) than that of the oxalic acid-free sample (NiNb/0–500). These results are in agreement with that observed by TEM. Additionally, the concentration of Nb on the surface is also determined by the calcination temperature, as the calcination at 500 °C leads to a concentration of niobium double than that catalyst activated at 350 °C (NiNb/1–350).The catalytic results for the ODH of ethane using Ni-Nb-O catalysts prepared using different approaches are shown in Table 2. All catalysts resulted to be highly selective to ethylene, although some catalytic differences were observed. The catalysts were tested in a temperature range of 300–350 °C in order to avoid changes in the chemical nature of the samples treated at low calcination temperature and to have a better comparison of the catalytic performance of these mixed oxides, i.e. minimizing the effect of reaction temperature on ethane conversion and selectivity to ethylene. During all catalytic tests, ethylene and CO2 were the only reaction products observed. Fig. 5 shows the variation of the catalytic activity per mass of catalyst (Fig. 5A), the catalytic activity per surface area of catalyst (Fig. 5B) and the selectivity to ethylene under isoconversion conditions (Fig. 5C) for Nb-doped samples, calcined at 350 °C or at 500 °C, with several OxA/Ni ratio in the synthesis gel. It is observed that the addition of oxalic acid in the synthesis gel leads to an increase in both the catalytic activity and the selectivity to ethylene regardless of the calcination temperature, with a maximum in the selectivity to ethylene for catalyst NiNb/1–500. Thus, the ethylene selectivity for catalysts prepared with oxalic acid ranges between 80 and 90 %, whereas the catalysts prepared in the absence of oxalic acid exhibit an ethylene selectivity of about 75 %. In addition, samples calcined at 500 °C show higher selectivity to ethylene than those calcined at 350 °C. Note that the optimal performance was observed for an OxA/Ni molar ratio of 1.We must remark that the activity per gram of catalyst differs in an important manner compared with the activity per surface area. The presence of oxalic acid in the synthesis gel highly increases the activity per mass of catalyst. However, if considered the catalytic activity per surface area the presence of oxalic acid in the synthesis gel follows a completely different trend, which depends on the calcination temperature as well as on the amount of oxalic acid in the synthesis gel.We must inform that, in all cases, the selectivity to ethylene barely varies with the conversion of ethane under the reaction conditions employed (Fig. S4). This means that carbon oxides are mainly formed directly from ethane [54]. However, variations in the selectivity to ethylene in the ODH of ethane are observed for the different catalysts tested, the NiNb/1–500 catalyst exhibiting the highest ethylene selectivity while NiNb/0–350 is the least selective one. Fig. S5 shows the variation of STY (space time yield for ethylene formation) with OxA/Ni ratio in the synthesis gel for the Nb-doped catalysts. This parameter takes into account both the catalytic activity and the selectivity to ethylene. It can be observed that the formation of ethylene is increased by the incorporation of oxalic acid in the synthesis gel, this effect being clearer in the case of the NiNb/x-350 set of catalysts. In addition, STY values for catalysts calcined at 350 °C are superior to those of the corresponding catalysts calcined at 500 °C. These results are in agreement with those of Fig. 5, which show a higher reactivity of the catalysts calcined at 350 °C.Consequently, the optimal catalyst in the Nb-doped NiO catalysts requires the presence of a certain amount of oxalic acid in the synthesis gel and a final calcination at 500 °C, leading to excellent ethylene selectivity (about 90 %). Moreover, and in agreement with the catalyst characterization results, the catalytic performance of Ni-Nb-O oxides can be explained in terms of the different physicochemical properties of the catalysts.On the other hand, the influence of the calcination temperature is not entirely clear at this time. Nevertheless, the positive influence of the calcination temperature in the Nb-containing NiO catalysts could be related to the higher presence of Nb5+ species on the surface of the NiO particles. Thus, the existence of isolated Nb5+ may be a key factor in increasing vacancies in the NiO lattice [32] which turn out to have a positive effect in ethylene selectivity in ethane ODH. Moreover, it is well known that the incorporation of Nb5+ in mixed metal oxides requires relatively high calcination temperatures [56], as observed in our case.Up to this point a conventional catalyst characterization has been conducted. However, little is known about electrochemical properties of the catalysts’ surface and their possible relationship with the catalytic performance. Fig. 6 shows the Mott-Schottky (MS) plots for Nb-doped catalysts synthesized with OxA/Ni molar ratios of 0, 1 or 3, and calcined at 350 °C and 500 °C. For comparison, the corresponding undoped nickel oxides catalysts have been also tested (Fig. S6). It can be seen that most samples (except for NiNb/1–500 and NiNb/3–500 catalysts) show a linear region with negative slope in the MS plots, which is indicative of p-type semiconductivity.Indeed, NiO is a well-known p-type semiconductor which contains an excess of cationic vacancies or, in other words, an excess over the lattice O2– anions [1,11,34,35,42], resulting from a non-stoichiometric composition (Ni1-xO). Both species (cationic vacancies and lattice oxygen anions) are associated with positive holes, which are the main charge carriers in p-type semiconductor materials. Concerning cationic vacancies, their linkage with positive electron holes is given by the following reaction (using the Kröger-Vink notation): (1) V M X → V M ″ + 2 h · where VM X is a neuter cationic vacancy (i.e. with a null electric charge after losing electrons from its surroundings), which can take two electrons from the valence band of a neuter lattice oxygen in a regular position (OO X, equivalent to an O2– anion), resulting in an ionized cationic vacancy (V’’M) and two electron holes positively charged (h•).On the other hand, electron holes are also related to lattice oxygen anions [35,57]: (2) O O X + h · → O O · where the neuter lattice oxygen anion (OO X) losses an electron (reacts with an electron hole) and becomes a lattice anion with positive effective charge (OO •), which is equivalent to the O- species [35]. Therefore, a positive hole corresponds to an electron vacancy in the valence band of a lattice OO X anion, which is to say that the “chemical site” of an electron hole corresponds to a non-stoichiometric lattice OO • (O-) anion [35,58]. Several studies have indicated that oxygen species on the NiO surface and within its structure can be of two types, their nature and reactivity being determinant for the catalytic properties of the oxide. Electrophilic oxygen species (O-), the non-stoichiometric oxygen (NSO), have been found to be related with the deep oxidation reactions, while nucleophilic species (O2–) are usually involved in more selective oxidation reactions [1,11,42,58,59].For our catalysts, it can be seen that for the samples showing p-type semiconductivity, the negative slopes increased for the catalysts calcined at 500 °C with respect to those obtained at 350 °C (Fig. 6 and Fig. S6). Besides, as observed for undoped NiO catalysts [41], the slopes of Nb-doped NiO catalysts prepared with oxalic acid in the synthesis gel were higher than for those prepared in the absence of oxalic acid in the synthesis gel. The value of the slope is inversely related to the density of acceptor defects (cationic vacancies) in the space-charge region developed at the oxide surface (NA ), according to the following equation [60,61]: (3) N A = - 2 ∊ ∊ 0 e σ where ε is the dielectric constant of the oxide (a value of 12 has been assumed for NiO) [33,62], ε0 is the vacuum permittivity (8.85·10-14F cm−1), e is the electron charge (1.60·10-19C) and σ is the negative slope of each straight line in MS plots for samples showing p-type semiconducting behavior.The density of cationic vacancies, NA , which corresponds to the density of holes (main charge carriers), for the samples with p-type semiconductivity is shown in Table 1. In general, NA values were higher in catalysts calcined at 350 °C than at 500 °C. Moreover, and regardless of the temperature of the thermal treatment, NA decreased between 1 and 2 orders of magnitude upon treating NiO with oxalic acid and after Nb doping [41]. This decrease indicates that the density of positive holes was lower in those cases than for simple NiO. Consequently, and due to the correspondence between electron holes and non-stoichiometric oxygen (NSO) species, i.e. O-, given by equation (Eq. (2)), it can be suggested that the higher the value of NA , the higher the predominance of NSO in the lattice and on the catalyst surface. Therefore, the selectivity of the catalyst towards specific oxidation reactions, such as the oxidative dehydrogenation of ethane to ethylene [1,11,42,59] should be lower. Our results (Table 2) are in accordance with those previously presented, in which the selectivity to ethylene was higher, in general, for Nb-doped NiO catalyst, but also for undoped NiO prepared with oxalic acid [41], than for the reference NiO catalyst.A drastic change occurred for the Nb-doped NiO catalysts prepared in the presence of oxalic acid in the synthesis gel and calcined at 500 °C (samples NiNb/1–500 and NiNb/3–500), in which a linear region with a positive slope could be clearly observed (Fig. 6). For these catalysts (NiNb/1–500 and NiNb/3–500), there was an evident modification in the semiconducting nature of the Nb-doped catalysts, evolving from p-type to n-type semiconductivity. This change can be attributed, on the one hand, to the Nb-doping, given that niobium oxide is an n-type semiconductor with oxygen vacancies and free electrons as main charge carriers [11]. Niobium, together with other high valence metals such as W (W6+) or Sn (Sn4+), is known for its ability to insert into the NiO lattice, filling Ni2+ vacancies with Nb5+ cations, whose ion sizes are compatible [1,11,34,35,55], and acting as electron donors. This substitution, as a consequence of changes in the catalyst preparation procedure, resulted in a decrease in the main charge carriers (hole) concentration, as observed for the Nb-doped NiO catalysts (Table 1), or even in a modification in the semiconducting nature of NiO from p-type to n-type semiconductivity, as observed for the NiNb/1–500 and NiNb/3–500 catalysts. Hence, the presence of oxalic acid when doping NiO with Nb5+ played a fundamental role, since in that case Nb-rich nanoparticles (Nb2O5, n-type) cover the surface of the NiO rich crystals. Thus, the doping procedure not only decreased the concentration of positive holes (and the concentration of electrophilic non-selective O- species), but it also changed the semiconducting character of the base NiO oxide, this way affecting its catalytic properties in terms of selectivity. Certainly, the selectivity towards ethylene formation for the NiNb/1–500 and NiNb/3–500 samples was the highest, reaching values of 90 % and 86 %, respectively (Table 2).Mott Schottky analyses of two representative catalysts after ODH reaction (i.e. NiNb/0-500R and NiNb/1-500R, R refers to reused) were carried out (Fig. S7A). Results show that semiconductive behavior of the samples is maintained after the ODH reaction, i.e. p-type semiconductivity remains for the catalyst formulated without oxalic acid and calcinated at 500 °C, showing a slight decrease in the acceptor density value (from 4.74·1020 to 4.38·1020 cm−3) for the sample after the ODH reaction. This could be explained considering that after the reaction, some electrophilic oxygens (reactive sites) might be consumed. In any case, the catalysts prepared with oxalic acid in the synthesis gel and calcinated at 500 °C the n-type semiconductivity is maintained after ODH reaction. Fig. 7 shows the Bode-module plots of the catalysts, calcined at 350 °C (Fig. 7A) and 500 °C (Fig. 7B), at an applied potential of 0.5 VAg/AgCl, where the impedance associated with the total resistance of the system corresponds to the impedance at low frequencies. The total electrical resistance of the catalysts surface has been related to their catalytic performance for the ethane oxidative dehydrogenation into ethylene [11,34,35,63]. The total resistance obtained from Fig. 7 was higher for the samples calcined at 500 °C. Additionally, the resistance increased for the samples doped with Nb5+ and prepared with oxalic acid. This fact can be explained taking into account that the total electrical resistance is inversely proportional to the density of charge carriers, NA , and the effect of Nb5+ to the NiO based catalysts is to partially (or totally, if oxalic acid is added to the electrolyte for catalysts calcined at 500 °C) remove the electrophilic NSO species (see Table 1). We must indicate that an undoped NiO catalyst, prepared without the addition of oxalic acid in the synthesis gel (Ni/0-500), also presented the lowest resistances [41]. Additionally, EIS tests were carried out after ODH for representative samples (Fig. S7B), showing similar impedance profiles for the catalysts after the ODH reaction even though total resistances after the reaction are somewhat higher. This is in agreement with the decrease of the acceptor densities after the ODH reaction presented for the capacitance measurements.Cyclic voltammetries were registered in order to study the electrochemical activity of the different catalysts. Fig. S6 shows, as an example, the cyclic voltammograms for the catalysts at 350 and 500 °C using the Ferro/Ferri redox couple. Note that 10 cycles were performed for each catalyst and no considerable differences were observed between the first and the tenth cycle, hence, Fig. 7C and 7D shows the different cyclic voltammograms of the catalysts for the tenth cycle. The cyclic voltammetries of Fig. 7 and Fig. S6 clearly show two peaks at ∼ 0.18 VAg/AgCl (cathodic) and at ∼ 0.30 V Ag/AgCl (anodic), typical response of the Ferro/Ferri couple. In all cases, the anodic peaks were the highest for the catalysts annealed at 350 °C, which is consistent with the increase of the catalysts active area for samples calcined at low temperatures. Besides, for a given temperature, the anodic peak values are higher for the samples doped with Nb5 + in the presence of oxalic acid. This behavior might be also attributed to the increase of the surface area in those catalysts, specifically to the electrochemical active area, due to the Nb5+ and oxalic acid contents.Nb-doped nickel oxide catalysts are found to be highly selective catalytic materials in the ODH of ethane, in agreement with previous results [11–17,21,25–29]. However, as presented here, the addition of oxalic acid to the synthesis gel and the selection of a suitable calcination temperature strongly influence the physicochemical characteristics and, consequently, the catalytic behavior of Nb-doped NiO catalysts. Both the addition of oxalic acid and a lower calcination temperature (350 ºC) resulted in an increase in the surface area of the catalysts with an influence on the catalytic activity. However, the activity normalized per surface area (Table 2) highly varies depending on the sample, so that other factors also have a strong influence on the catalytic activity. Moreover, it has been observed by Raman spectroscopy that, depending on the amount of oxalic acid in the synthesis gel and/or the calcination temperature, samples with different crystal size and/or concentration of defects have been synthesized. Electrochemical characterization has been undertaken to further refine any conclusion to be drawn.The catalytic behavior of these catalysts has been linked to several parameters, in particular: the high concentration of Ni defects and the minimum concentration of electrophilic oxygen species, as well as the highest presence of Nb5+ species on the surface of the catalyst. Although changes in physicochemical and catalytic properties have been recently proposed for undoped NiO [41], the changes presented here for Nb-doped catalysts show notable differences to those observed for undoped NiO.The influence of the presence of oxalate in the synthesis gel and the calcination temperature on the catalytic properties of bulk NiO catalysts has been previously studied [41]. Then, it would be interesting to study, in a comparative manner, the possible differences in the influence of the two synthesis parameters on the catalytic properties of the NiO and Ni-Nb-O catalysts. Fig. S7 shows the change in selectivity to ethylene under isoconversion (10 %) conditions of NiO [41] and those achieved over Ni-Nb-O catalysts calcined at 350 or 500 °C. In the case of undoped NiO catalysts, the selectivity to ethylene initially increases with the incorporation of oxalate anions, presenting a maximum selectivity for the catalyst prepared using oxalic acid (OxA/Ni of 1) and calcined at 350 °C. In any case, catalysts calcined at 350 °C showed a greater selectivity to ethylene than those calcined at 500 °C regardless of the oxalic acid amount employed in the synthesis.In the case of Nb-doped NiO catalysts, the selectivity to ethylene increases with the incorporation of oxalic acid into the gel. However, unlike NiO catalysts, ethylene selectivity is higher for catalysts calcined at 500 °C than for those at 350 °C. Thus, the calcination step at 500 °C has a positive effect on the selectivity towards ethylene for Nb-doped NiO catalyst prepared with oxalic acid in the synthesis gel. Fig. S8 shows the variation of the catalytic activity for the NiO and Ni-Nb-O catalysts with the OxA/Ni ratio in the synthetic gel. A similar trend is observed in the catalytic activity for the NiO and Ni-Nb-O catalysts. Thus, catalysts calcined at 350 °C are more active than those calcined at 500 °C. In addition, an influence of the presence of oxalic acid in the synthetic gel on the catalytic activity is observed, so that the catalysts prepared with an OxA/Ni ratio of 1 are the most active regardless of the presence or the absence of Nb5+ in the catalyst. The presence of Nb5+ has a significant positive effect on the selectivity to ethylene (Fig. S7) but a weak effect on the catalytic activity (Fig. S8), whereas the incorporation of oxalic acid in the synthesis gel has a positive influence on the catalytic activity for both NiO and Ni-Nb-O catalysts.Accordingly, it can be concluded that the incorporation of Nb5+ increases the selectivity to ethylene and the rate of formation of ethylene, this effect being greater if oxalic acid is incorporated into the synthesis gel. As indicated in the characterization of the catalysts, the presence of Nb5+ in Ni-Nb-O promotes a low reducibility of Ni-O bonds, as determined by TPR-H2 (Fig. 3B). This aspect is more evident in catalysts calcined at 500 °C, which explains the high selectivity to ethylene, in particular for the sample NiNb/1–500 (Fig. 5). Fig. S9 shows the reaction rates for the formation of ethylene and CO2 to the product kgcat -1h−1 (the sum corresponds to the catalytic activity for the conversion of ethane) with the oxalic acid/Ni molar ratio in the synthesis gel for catalysts calcined at 350 °C (Ni/x-350 and NiNb/x-350 series) and at 500 °C (Ni/x-500 and NiNb/x-500 series). In all cases, catalysts prepared with an oxalic acid/Ni ratio of 1 exhibit the highest reaction rate for the formation of ethylene. In addition, catalysts calcined at 350 °C exhibit the highest reaction rates, while catalysts calcined at 500 °C, especially those containing Nb, exhibit the highest ratio between the rate of ethylene formation and the rate of CO2 formation, i.e. the greatest selectivity to ethylene.Thus, the combined use of a relatively high calcination temperature (500 °C) and the inclusion of an appropriate load of oxalic acid in the synthetic gel during the preparation step resulted in excellent selectivity to ethylene (approx. 90 %). In addition, and in accordance with the characterization results, the catalytic performance of Ni-Nb-O catalysts can be explained in terms of the different physicochemical properties of the catalysts, including changes in the number of vacancies and in the size and concentration of electrophilic oxygen species.Lemonidou et al. [55] demonstrated the existence of a strong kinetic isotopic effect (KIE) on NiO and Nb-doped NiO catalysts, suggesting that: i) breaking the CH bond is the determining step in the speed of the reaction in the ODH of ethane over both catalysts; ii) these two catalysts should have similar active sites, although the abundance or surface concentration of selective and non-selective sites changes with the incorporation of Nb5+ into the NiO lattice. In this way, it has been proposed that the nature of the surface sites is strongly influenced by the valence and acid-base characteristics of the metal oxide promoters, which have a great impact on the selectivity to ethylene [21,25–28]. Thus, these authors have proposed a good correlation between the selectivity to ethylene and the valence of the promoter. In conclusion, the authors found that Nb5+ was the best promoter.On the other hand, the influence of the calcination temperature is not entirely clear at this time. However, the positive influence of calcination temperature in the catalytic performance in ethane ODH of Nb-doped NiO catalysts and the change in semiconductivity nature (from p- to n-type) in samples calcined at 500 °C could be related to the higher or lower incorporation into the structure of NiO particles and/or the dispersion of Nb5+ on the surface of the NiO particles, as evidenced by the XPS results (Table 1).The main promotional mechanism to explain the improvement in selectivity to ethylene in NiO catalysts doped by transition metals, and especially those doped with niobium, is related to the elimination of cationic vacancies in the NiO particles by the incorporation of Nb5+ into the nickel oxide structure [34].A second promotional mechanism for improving the selectivity to ethylene in NiO-based catalysts has been proposed for supported NiO catalysts, in which the interaction of nickel oxide with supports, determines the catalytic performance [64,65]. In this case, by using an appropriate metal oxide support, the enhanced catalytic performance has been related to the high dispersion of nickel oxide particles on the support, which leads to a lower reducibility of the nickel oxide, hindering the oxidation of ethane into carbon oxides.A third promotional mechanism could be related to the lower crystallization of NiO crystals and the interaction between NiO and promoter oxides due to the presence of highly dispersed oxides on the catalyst surface. This is the case of SnO2-promoted NiO catalysts, in which the nature of Ni species has been related to changes in the size of NiO crystallites and the presence of SnOx crystals highly dispersed on the surface of NiO [19]. Thus, the presence of oxalic acid in the synthesis gel of our optimal Ni-Nb-O catalysts seems to favor the formation of Nb-rich nanoparticles on the surface of the NiO large crystals. In fact, the characterization results suggest the presence of agglomerates of nanoparticles of an amorphous nature that concentrate niobium, while the visible isolated platelets of NiO contain a low amount of Nb.Then, in the optimal catalyst of the present article, it seems that both the first and the third promotional mechanisms occur: a little amount of Nb incorporates to the NiO lattice (first mechanism) and Nb-rich nanoparticles cover the surface of the NiO rich crystals (third mechanism). This way, the simultaneous occurrence of both promotional mechanisms leads to an enhanced ethylene formation.Moreover, a deep electrochemical study of the catalysts has been also carried out. Previous studies on promoted NiO catalysts showed that these catalysts present, similarly to unpromoted NiO, p-type semiconductivity (before and after reaction) [33–35]. Thus, it was observed that the addition of suitable promoters decreases the p-type semiconductivity, leading to an increase of the selectivity to ethylene. In this work, this trend has been observed but, additionally, the most selective catalysts present n-type semiconducting character. Moreover, other interesting correlations between electrochemical and catalytic properties have been also found.Then, Fig. 8 A plots the relationship between the selectivity to ethylene and the acceptor density for these catalysts with p-type semiconductivity. These results, consequently, reveal a correlation between NSO density (directly associated with electron holes concentration) and the selectivity towards ethylene. Hence, in general, the catalysts with the highest selectivity correspond to those with the lowest NA values, i.e. fewer electrophilic oxygens.The electrochemical impedance (resistance) of these catalysts could be related to catalytic performance since a high resistance could hinder non-selective reactions. Accordingly, Fig. 8B shows a relationship between the selectivity to ethylene with the total electrical impedance. Since cationic vacancies and electron holes are related to the presence of electrophilic oxygen species, which in turn are associated with the total ethane oxidation to CO2, as explained before, the general trend presented in Fig. 8B is consistent. That is, higher total electrical resistances (where cationic vacancies and, therefore, O- species are partially or totally eliminated) correspond to higher selectivity values. Therefore, there is a clear inverse correlation between the total conductivity of catalysts (and the concentration of charge carriers within their structure) and the selectivity towards ethylene formation.Above, some correlations between the electrochemical properties and the selectivity to ethylene have been found. At this point, it could be interesting to find a representative electrochemical parameter linked to the catalytic activity. In this way, it makes sense that the anodic peak values could be related with the activation of the ethane both selectively and non-selectively. Fig. 9 A shows a clear correlation between the ethane conversion at fixed conditions (conversions lower than 15 %) and the intensity of the anodic peak values, which is related to the electrochemical activity of the samples. Conversion of ethane depends on the number of active sites capable of activating ethane and also on the reactivity of these active sites. In these bulk NiO catalysts the amount of active sites highly depends on the surface area and the amount of Nb, which is supposed to be inactive in this reaction conditions. Then, we have calculated the turnover frequency (TOF), which is the parameter that considers the catalytic activity of the active sites. The determination of TOF requires the estimation of the amount of exposed surface sites. In the case of these catalysts, we have considered the surface area, the surface composition and taking into account that the amount of molecules that cover the monolayer of nickel oxide is 9.7.1014 molecules of NiO per cm2 [66]. Interestingly, Fig. 9B shows that there is also a certain relationship between the anodic peak and the catalytic activity per surface site of the catalysts (TOF). Thus, the higher anodic peak values correspond to catalysts with an enhanced activity per surface site (molecules reacted per surface site per unit of time) but, especially, with the conversion of ethane (and also with the activity per gram of catalyst, which also takes into account the enhanced surface area, see Table 1). Therefore, there is a clear link between surface area and electrochemical active area, but also between the last one and catalytic activity towards ODH of ethane.The n-type semiconductivity observed for selected NiO based catalysts in the present article, requires simultaneously: i) the presence of Nb; ii) the use of oxalic acid in the preparation method; and iii) a calcination temperature of 500 °C. If one of these factors are absent, that n-type character is not observed.Accordingly, the presence of Nb5+ decreases the number of cationic vacancies [35] whereas high calcination temperatures favor the Nb5+ incorporation into the NiO lattice (according to our TEM data). The role of the oxalic is not straightforward to explain but its presence favors the formation of small Nb-rich particles and minimizes the Nb-insertion into the NiO lattice. Then, we can hypothesize that n-type catalysts (prepared with oxalic acid) are formed by Nb-containing NiO crystallites whose p-type character has been notoriously reduced together with many Nb2O5 nanoparticles with a clear n-type semiconductivity [67,68] located on the surface of the NiO particles, leading to an overall n-type semiconducting character. The p-type character of the catalyst without oxalic acid (NiNb-0/500) could be the result of a higher incorporation of Nb and a lower amount of small particles that, additionally, present a relatively high concentration of Ni, then maintaining the overall p-type character.Finally, we want to mention that the optimal catalyst is quite stable after 32 h on-line. Using a high contact time, W/F, of 80 gcat h (molC2)-1 and a feed rich in oxygen (the initial conversion was as high as 57–57.5 % with an ethylene selectivity of 70–71 % (yield of 40–41 %). After 2 h, the ethane conversion decreased until 55–55.5 % which remained almost stable after the rest of the experiment (Fig. S11). This slight drop in the catalytic activity could be related to the subtle fall of the surface area (32.9 m2/g for the fresh catalyst whereas 31.5 m2/g for the used sample), since no apparent variations in the electrochemical properties, in the near surface or in the crystalline phases (by XPS or XRD) have been observed in the post-mortem catalyst compared to the fresh catalyst (Fig. S12).Controlling the preparation conditions (calcination temperature of 500 °C and an appropriate amount of oxalic acid in the synthesis gel) a high and stable selectivity to ethylene of ca. 90 % can be obtained. By using these synthetic conditions two mechanisms that promote ethylene selectivity (Nb5+ incorporated into the NiO lattice and the interaction of large Nb-containing NiO crystals with tiny Nb-rich particles) take place simultaneously. Unlike the other samples, the optimal catalysts present n-type semiconductivity, in spite of the fact that the composition (Ni/Nb ratio) is the same for all the catalysts studied. In the present article, we have also shown that the behavior in the oxidative dehydrogenation of ethane of NiO catalysts can be estimated knowing their electrochemical properties. Interestingly, an inverse relationship between the density of non-stoichiometric oxygen and the selectivity towards ethylene has been clearly found in the catalysts presenting p-type semiconductivity. Thus, the catalysts with high selectivity to the olefin present low NA values, i.e. fewer electrophilic oxygens. Additionally, a correlation between the selectivity to ethylene and the total electrical resistance has been observed. Then, the optimal catalysts present high total electrical resistance as a result of the removal of cationic vacancies and electrophilic O- species, which are selective towards the CO2 formation. Overall, the most selective catalysts are those having low concentration of cationic vacancies and electrophilic oxygen species as well as a high amount of Nb5+ species on the surface of the catalyst.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to thank the the Ministerio de Ciencia e Innovación of Spain, MINECO/FEDER (Projects: TED2021-129555B-I00, PID2021-126235OB-C31, PID2021-126235OB-C33, TED2021-130756B-C32 and MFA/2022/016). Y.A. thanks the Ministry of Higher Education and Scientific Research of Algeria for the National Exceptional Program for the fellowships. A.A. acknowledges Severo Ochoa Excellence Program for his fellowship (BES-2017-080329).Supplementary data to this article can be found online at https://doi.org/10.1016/j.jcat.2023.02.009.The following are the Supplementary data to this article: Supplementary data 1

In the present article, a relationship between the catalytic performance in the oxidative dehydrogenation of ethane (ODHE) for Nb-doped NiO catalysts and their electrochemical properties has been proposed. To do so, highly stable and selective Nb-doped NiO catalysts for the ODHE to ethylene have been synthesized by optimizing synthesis parameters such as the amount of oxalic acid in the synthesis gel and the calcination temperature. These catalysts have been characterized by means of XRD, HRTEM, Raman and UV–vis diffuse reflectance spectroscopies, TPR and XPS. Moreover, Electrochemical Impedance Spectroscopy (EIS), capacitance measurements (Mott-Schottky analysis) and cyclic voltammetries studies were also carried out. Electrochemical characterization indicates changes in the type of semiconductivity: p-type for samples calcined at 350 °C and the sample prepared in the absence of oxalic acid and calcined at 500 °C, to n-type for samples prepared in the presence of oxalic acid in the synthesis gel and calcined at 500 °C. According to the obtained results, the most selective catalysts present a low Nb-incorporation on the NiO lattice with a large amount of tiny Nb2O5 nanoparticles covering NiO crystallites. This work presents, for the first time, a complete electrochemical characterization of Nb-doped NiO catalysts showing a correlation between electrochemical properties and catalytic performance.

Data will be made available on request.Due to economic, social, and environmental reasons, research focusing on alternative fuel and chemical sources has gained interest. As reserves of fossil fuel are declining worldwide, combined with an increasing demand for petroleum fuels by emerging economies, the price of conventional fossil fuel shall continue to rise, increasing the risk in energy supply around the world. Furthermore, the environmental damage resulting from the combustion of fossil fuel, by releasing atmospheric pollutants and CO2, is a contributing factor to global warming [1]. As the only natural-occurring source of renewable organic carbon, biomass represents an essential feedstock to produce chemicals and liquid transportation fuels. The recalcitrant C–C and C–O bonds found in biomass requires a bulk depolymerisation technique to produce bio-oil, a viscous liquid that can be easily processed, stored, and safely ‘dropped-in’ into the supply chain for large scale chemical conversions in existent refineries [2]. The bio-oil produced from common thermochemical methods consists of a complex mixture of oxygenated compounds (∼50% oxygen), which can be upgraded to hydrocarbon fuels and specialty chemicals through proper refining methods such as hydrodeoxygenation, decarboxylation, decarbonylation, isomerisation, hydrogenation, dehydration, etc. [3–7].In general, bio-oil can be obtained from biomass by pyrolysis and/or hydrothermal liquefaction (HTL) methods. The bio-oil from pyrolysis is highly unstable as a result of its high oxygen content and as a result of vigorous reaction condition which can destroys the natural structure of phenolic compounds [8]. Our earlier investigation on using pyrolysis for valorization of PJ biomass has yielded only 25% bio-oil after removing the aqueous phase [9]. On the other hand, conventional HTL takes place under subcritical water conditions (250 – 373 °C, 4–22 MPa) eliminating the need of a pre-drying step. Under these conditions, water acts both as solvent and as acid-base catalyst, improving the solvation and deoxygenation of intermediate compounds, yielding higher-quality bio-oils compared to pyrolysis [10]. However, supercritical conditions, with temperatures above 380 °C, have been shown to improve glucose conversion, reducing char production compared to subcritical HTL [11]. Besides the carbon rich bio-oil, HTL also yields solid hydrochar, gas products (mostly CO2) and an aqueous phase, which can be easily separated from the desired bio-oil product.A potential HTL catalyst must be water-tolerant, display high selectivity towards bio-oil, minimizing char and gas formation. Homogeneous catalysts consisting of base or basic salts such as NaOH, KOH, Na2CO3, K2CO3 have been intensively utilized for biomass HTL, decreasing the biochar formation and increasing bio-oil yield, however, presenting challenges in separation, extraction, and reusability of the catalyst [12,13]. Heterogeneous catalysts have the advantage of easy separation from the liquid products, improved process economics and energy efficiency. Xu et. al. utilized solid alkaline earth metal catalysts such as hydrotalcite, MgO, and colemanite for woody biomass HTL which improved bio-oil yield and quality [14]. Noble metal catalysts such as Pd/C, and transition metal catalysts based on Ni, W, Co, Mo, and Fe such as Raney nickel, Fe ore, FeS, Ni and Fe metals, CoMo/γ–Al2O3, etc. has been also explored for biomass HTL [15–17]. In particular, Ni catalysts produced bio-oil with improved yield and quality due to its hydrogenating property [18,19]. We have previously reported the deoxygenating behavior of Nb2O5 catalyst due to its oxophilic nature that can strongly bind with the oxygen groups helping to cleave the C–O bond [9].As known, biomass tends to be rich in carbon but hydrogen deficient. Hence, incorporating a H-rich co–reactant during the biomass HTL could has the ability to increase the bio-oil yield and quality [20,21]. As such, plastic waste composed of polyolefins (polyethylene, polypropylene, their copolymers and olefinic rubbers) could be a potential co-reactant as contains hydrogen-rich polymers [22–24]. Utilising plastic as a co–reactant not only benefit increasing the bio-oil yield but also benefit mitigating the waste plastic landfilling environmental issue, that could result in an effective waste management strategy. Polypropylene (PP), a non–oxygenated light weight polymer made of long chain molecules (C3H6)n, [25] is one of the most utilized commodity plastics and is present as the largest fraction in the waste-stream [26]. PP at the subcritical hydrothermal liquefaction conditions (350 °C, 20 min, non-catalytic) produced mainly solid (83%) [27]. However, under supercritical water liquefaction has been reported to increase the PP degradation to 91 wt.% oil (80% range naphtha hydrocarbons) at 425 °C and 2−4 h [28]. The same oil yield was achieved at 0.5 – 1 h reaction time when the temperature was increased to 450 °C [28]. This hydrocarbon oil produced from PP can then synergistically improve the biomass conversion and bio-oil quality when is co-liquefied together with biomass [29]. However, until now there has been limited investigation on catalytic liquefaction of biomass with PP and their synergetic interactions on bio-oil yields obtained.In this study, we aim to investigate the production of renewable hydrocarbons from abundantly available non-food biomass such as PJ using the hydrothermal co-liquefaction route. Prosopis juliflora (PJ) with a growth rate of 25 km2/year in India, is an abundant biomass which is resistant to drought and adaptable to different soil types and therefore, is a promising biomass source for biofuel production used in India and other tropical countries [30]. Our target also seeks to elucidate the synergetic effect occurring when polypropylene (PP) wastes are added to PJ in terms of improving the bio-oil yield. The co-liquefaction studies of PJ and PP were conducted over a broad temperature range of 340 °C to 440 °C, using different PJ/PP ratios at 60 min reaction time, where the synergy percentage effect was calculated at each condition.The first part of this study consisted of the non-catalytic co-liquefaction reactions to optimise the temperature and percentages of PP added to PJ in terms of high bio-oil yield. On the second part, a series of alumina supported metal oxide catalysts were tested for optimum conversion of biomass-plastic mixture. On this account, we firstly aimed to synthesise, characterize, and evaluate the catalytic activity of a series of transition metal oxides (Ni, Mo, W, Nb) supported on γ–Al2O3 for the individual and co-liquefaction of PJ and PP. The temperature, percentage of PP added to PJ, effect of catalyst, catalyst: feed ratio, were all optimized in terms of high bio-oil yield. The reaction products from the HTL process (i.e. bio–oil, aqueous phase, gas, and bio– char) are all characterised and optimal process conditions are reported. The regeneration and reusability of the best performed catalyst at the best reaction condition was also studied.Ammonium molybdate tetrahydrate (81.0 – 83.0%), nickel (II) nitrate hexahydrate (>98.5%), ammonium metatungstate hydrate (≥85%), niobium pentachloride (99%) were purchased from Merck, India. γ–Al2O3 was purchased from BASF chemicals company. Ethanol (99.9%) was purchased from Changshu Hongsheng fine chemicals. Prosopis juliflora (PJ) and single-use polypropylene (PP) (polypropylene packaging bags) were collected in and around SSN College of Engineering campus, Chennai, Tamil Nadu, India. Both the PP and PJ waste were cut into small pieces using a blade shredder and sieved to a size of 1mm.PJ and PP were characterized to understand their composition that plays a vital role in product formation. PJ is a hardwood biomass with 37.9% cellulose, 19% hemicellulose, and 37% lignin [30]. The C, H, N, S and O content of PJ was 48%, 7%, 0%, 2% and 43%, respectively, whereas PP contains 86% C and 14 % H (Table 1 ). The (H/C)eff of PJ and PP were 0.375 and 1.95, respectively. It is explicit that PP is H-rich whereas PJ is H- deficient. PP contained high amount of volatile matter (96.2%) with less amount of fixed carbon (3%) and negligible ash content. On the other hand, 79% volatile matter, 15.2% fixed carbon (non-volatile carbon) and 5.8% ash content were found in PJ. The high heating value (HHV) of a biomass is highly influenced by the composition of lignocellulosic components, extractives and detrimentally by moisture and ash contents [31]. The HHV of PJ and PP are 20 MJ/Kg and 41 MJ/Kg, respectively. The high HHV of PP is attributed to its high H content and the absence of heteroatoms.The moisture content, volatile matter, ash content and fixed carbon were analyzed according to ASTM standards E871–82 and E1755–01. Ultimate analysis was conducted using an ELEMENTAR Vario EL III elemental analyzer. Thermogravimetric analysis (TGA) was performed to determine the waste degradation rate using a Shimadzu TGA 50H thermogravimetric analyzer. TGA was performed using 10 mg of waste at a temperature of 30 to 800 °C under 20 °C/min heating rates and held at final temperature for 10 min.Alumina (γ–Al2O3) supported metal catalysts (Mo, Ni, Nb, W) were prepared by simple wetness impregnation method with a nominal metal content of 7 wt.%. The desired amount of aqueous solution of the metal precursor was added to the alumina support and mixed in a rotary evaporator at room temperature for 12 h (In the case of niobium pentachloride, ethanol was used as the solvent due to its decomposition in water). Water was then removed by the rotary evaporator at 50 °C, followed by drying the catalyst overnight at 100 °C, and subsequent calcination at 550 °C for 5 h in a muffle furnace. The catalysts were then labelled as Mo/alumina, Ni/alumina, Nb/alumina, and W/alumina.The catalyst structural analysis was elucidated by X –ray diffraction (XRD) using Bruker D8 advance with monochromatic Cu Kα radiation (λ = 1.542 Å) at 30 kV and 15 mA with a step size of 0.1 °, for the range of 10 ° ≤ 2θ ≤ 80 °. Nitrogen adsorption-desorption data were obtained at -196 °C using a Micromeritics TriStar II 3020 surface area and porosity analyser. Prior to physisorption measurements, all samples were degassed under vacuum at 200 °C overnight. The specific surface area was determined by applying Brunauer– Emmett– Teller (BET) method and pore volume were calculated from the amount of N2 adsorbed at P/P o of 0.99. An Inductively coupled plasma mass spectrometry (ICP–MS) from Thermo Fisher iCAP RQ ICP–MS was used for the bulk elemental analysis. The amount and strength of the catalyst acid sites were characterized using a Micromeritics Autochem II 2920 chemisorption analyzer (TPD-ammonia), fitted with a TCD detector for monitoring NH3 desorption profile. About 50 mg of sample was preheated for 2 h under the flow of helium gas at 400 °C. Then the sample was saturated by passing 15 vol% NH3 in He for 1 h at 100 °C. Afterward, was heated from 100 °C to 800 °C at a heating rate of 10 °C/min. In parallel, Pyridine Fourier- Transform Infrared Spectroscopy (FT-IR) was used to determine the nature of the catalyst acid sites. A known amount of pyridine was adsorbed on the 50 mg catalyst at 150 °C. The excess and physiosorbed pyridine were removed by passing N2 at 150 °C for 30 min and FT-IR was recorded using a Perkin Elmer 200 FT–IR, USA spectrometer at 128 scans and 4 cm-1 resolution. A Field Emission Scanning Electron Microscope (FE–SEM) – JOEL 6390LA microscope operated at 30 kV with backscattering (BSE) and Energy Dispersive X–ray Spectroscope (EDAX) detectors was used for characterising the morphology of catalysts. A high-resolution transmission electron microscope (HR-TEM, JOEL/JEM 2100) operated at 200 kV, fitted with an energy dispersive X-ray (EDS) detector was used to find the particle size distribution and metal dispersion over the alumina support. X-ray photoelectron spectrometer (XPS) by Scienta O micron was used to find the oxidation states of Nb2O5. The peaks were calibrated by using C 1 s line in the carbon spectra at 284.0 eV as a reference.Hydrothermal liquefaction reactions (HTL) were carried out in a 250 ml capacity stainless steel closed high–pressure batch auto–reactor. The reactor was loaded with 15g of feed (biomass and/or PP), a feed: water ratio of 1:8 and pressurized to 5 MPa with nitrogen. After heating to the desired reaction temperature (320 °C to 440 °C, heating rate 10 °C/min), temperatures were maintained for 60 min under constant stirring at 740 rpm. The effect of Mo/alumina, Ni/alumina, Nb/alumina, and W/alumina catalysts on bio-oil yield was studied by varying the catalyst wt.% with respect to feed (1 wt.%, 2 wt.%, 3 wt.%, 4 wt.% and 5 wt.%). It must be noted that the catalysts were not reduced before the reaction. Before dismantling the reactor, the reaction was quenched by removing the heating jackets and immersing the autoclave in an ice water bath. The pressure was released by collecting gases using a Tedlar gas bag. Bio-oil produced from the HTL crude was separated through solvent extraction process using hexane [32]. The contents of the reactor were extracted using hexane as the solvent and transferred into a 250 ml separating funnel where the organic phase was recovered. The organic phase (bio-oil) was subjected to vacuum separation to remove excess hexane. The solid along with the catalyst was collected by filtration and washed with ethanol and dried overnight at 100 °C and analyzed for the coke deposition using an ELEMENTAR Vario EL III elemental analyzer. For the reusability tests, before conducting each test, the catalyst was regenerated by burning off the deposited coke at 400 °C in a muffle furnace [9]. The yield of bio-oil, gas, aqueous phase, and solids, higher heating value (HHV), percentages of deoxygenation in bio-oil, and carbon recovery in bio-oil are evaluated using the formulae given in the electronic supplementary information (ESI). % Synergy and % calculated yields are evaluated using the formulae: % Synergy = Experimental yield − Calculated yield Calculated yield ∗ 100 Calculated Yield = x PJ ∗ y PJ + x PP ∗ y PP / 100 where x is the mass fraction, y is the % yield, PJ is Prosopis juliflora, and PP is polypropylene.Gas chromatography–mass spectrometry (GC–MS) was used to analyze the bio-oil, obtained from hydrothermal liquefaction process. An Agilent 7890 GC equipped with an Agilent 7683B auto–injector, a HP–5 column and flame ionization detector (FID) was used. The injector temperature was 250 °C. The column temperature was set at 100 °C and held for 1 min, followed by ramping at 10 °C/min to 200 °C and held for 10 min. A volume of 0.5 μL liquid product was injected in a split mode ratio of 100:0. The average molecular weight of bio-oil was analyzed by gel permeation chromatography (GPC) using a Water GPC 1515 pump system provided with Styragel HT–6E and HT–3 columns linked in series. UV (Waters 2489) and RI (Waters 2414) detectors were used for finding the average molecular weight of bio-oil products. The bio-oil samples were dissolved in 1 mg/ml THF (used as an eluent with a flow rate of 1 ml/min) and filtered using a 0.45 micron filter before analysis. The system was calibrated using the narrow polystyrene standards in the range of Mw 1.3 million Da to 1350 Da.Four catalysts (Mo/alumina, Ni/alumina, W/alumina, and Nb/alumina) were studied for the co–liquefaction of PJ and PP. The alumina support had a surface area of 192 m2 g−1 with a pore volume of 0.49 cm3 g−1 (Table 2 ). The metal impregnation over alumina support decreased the surface area to 127, 137, 124, and 139 m2 g−1, for Mo, Ni, W and Nb catalysts, respectively. Similarly, the deposition of the metal particles in the surrounding pore mouth of alumina decreased the pore volume as expected. The elemental percentage as measured by ICP– MS (Table 2) was 7.3, 6.5, 7.2, and 6.9 for Mo, Ni, W and Nb catalysts, respectively which is in good agreement with the theoretical values (standard deviation = 0.3317). Table 2 also reports the catalyst acidity measurement by TPD-ammonia. The acid strength is categorized as weak, moderate, and strong depending on the temperature at which ammonia was desorbed from the catalyst [33]. From the table, alumina support exhibits a total acidity of 0.76 mmol/g where about 55% are weak acid sites and 45% moderate acid sites. The metal loading to alumina support increased the total acidity where the highest acidity was found to be for Nb/alumina (1.23 mmol/g). Alumina supported Mo, Ni, and W exhibited 0.92, 0.87 and 0.98 mmol/g total acid sites, respectively. The nature of acid sites (Brønsted/ Lewis) was examined using pyridine FT-IR spectroscopy (Fig. 1a). The characteristic absorption bands at 1425 and 1630 cm-1 represented the surface coordinated pyridine molecules with the Lewis (PyL) acid sites whereas, the absorption peak at 1540 cm-1 represents the pyridine ion adsorbed on the catalyst Brønsted (PyB) acid sites [33]. The adsorption band at 1480 cm-1 is characteristic for a combination of Brønsted and Lewis acid sites (PyL + B). Alumina support exhibits a strong Lewis acidity with a negligible Brønsted acid site. The deposition of metals increased the Lewis and Brønsted acid sites where the highest PyL, PyB and PyL + B was observed with Nb/alumina catalyst in agreement with TPD-ammonia results in Table 1.The XRD spectra of alumina support in Fig. 1b displayed three main peaks at 2θ = 37.2°, 45.5° and 66.7° corresponding to the d311, d400, d440 reflections of γ–Al2O3 (PDF 00–050–0741) [34]. All the supported metal catalysts, apart from exhibiting the signals related to alumina, had additional peaks corresponding to the respective metal oxides (Fig. 1b). The SEM coupled with EDAX images given in Fig. S1 (a-d) in ESI indicated the absence of other elemental impurities.Firstly, non-catalytic co-liquefaction of PJ with different percentages of PP added was investigated in a temperature range of 340–440 °C and the bio-oil yields are depicted in Fig. 2a. In the absence of PP, increasing the temperature from 340 °C to 420 °C increased the bio-oil yield from 13.5% to 42.5% with a concomitant decrease in solid residues (Fig. S2 in ESI). Increasing temperature stimulates the conversion of organic compounds into bio-oil, gaseous products, and other water-soluble compounds. While further increasing the temperature to 440 °C, the yield of bio-oil slightly dropped (from 42.5% to 41.2%), because of thermal cracking of bio-oil compounds following in an increase of gaseous product from 24.4% at 420 °C to 28.9% at 440 °C (Fig. S2 in ESI).On the other hand, HTL of PP alone, at the subcritical conditions (below 380 °C) yielded mainly 58.6%, 50.2% and 47.4% solid residue products, with an oil yield of 15.2, 16.5 and 17.2% at 340 °C, 360 °C, and 380 °C, respectively which is comparable to the findings by Savage. et. al and Biller. et. al. [20,27], where it was claimed that at subcritical reaction condition, PP degradation was low due to the insufficient number of reactive active sites for dehydration [27]. The appreciable oil yields started at the supercritical condition, where oil yield of 30.2 % was obtained at 400 °C, reaching a maximum yield at 420 °C (37.5%) with further decrease to 31.3% at 440 °C, similar behavior as observed with PJ. Concurrently, the solid products decreased from 58.6% (at 340 °C) to 20.2% at 420 °C, owed to the supercritical water that stabilizes the radicals minimizing coke formation [28].When adding 25% PP during liquefaction of PJ, a substantial increase in bio-oil yield at 340 °C (13.5% to 27.1%) was obtained. This yield is 97.6% higher than the calculated yield based on a weighted average of PJ and PP yields (Fig. S3a in ESI) and suggests a significant synergy occur during the co-conversion of the two materials. Similarly, the solids decreased to 30.2%, compared to 47.3% when PJ alone was used at 340 °C. Subsequently, a gradual increase in bio-oil yields up to 46.5% at 420 °C was observed, representing about a 12.8% bio-oil yield improvement for all the non-catalytic HTL reaction conducted in this study. It is known that during HTL, decomposition through free-radical formation is more prevalent and the PP is known for the rapid formation of (more stable tertiary) free radicals upon C-H cleavage during thermal decomposition [35]. These free-radicals from PP are expected to bond with the oxygen radicals from biomass, thereby promoting the cleavage of the oxygenated groups from biomass enhancing the oil fraction formation [36]. As such, the bio-oil yield dropped to 45.1% when the temperature was further increased to 440 °C.As can be seen in Fig. 2a, a further increase in PP substitution to 33%, 50%, 67% and 75% respectively, although indicated good synergy (Fig. 2b) and bio-oil yield improvement, when compared to the anticipated calculated value, the amount of solid products formation increased resulting in poor bio-oil yield when compared to the yield obtained with 25% PP added to PJ. The maximum bio-oil yield at 25% PP implies that only a small amount of PP is sufficient to be added to generate enough radicals to break down biomass. Based on the data obtained from the non–catalytic HTL tests performed, 420 °C was selected as an optimum HTL reaction condition for further studies under the presence of a catalyst.The alumina supported transition metal oxide catalysts were screened for the co-liquefaction of PJ and PP at 420 °C, then compared to the non-catalytic reaction results (Fig. 3a). Initially, to distinguish the role of metals on the bio-oil composition and production, a blank experiment for the co-liquefaction reaction was carried out using only alumina support as catalyst.At 420 °C, HTL of PJ on alumina as a catalyst, resulted in 40.2% bio-oil yield, found to be 5.4% lower than that obtained from the non-catalytic reaction. Even with the addition of 25% and 33% PP to PJ, similar decrease in the bio-oil yield was observed (7.4%, 4.8% decrease in bio-oil yield at 25%, 33% PP addition, respectively). On the contrary, further increase in the PP % to 50%, 67% and 75% improved the bio-oil yield by 9.1%, 13.1%, and 11.7%, respectively.In an interesting approach, where HTL reaction was carried out using PP only, a 32% improvement in the bio-oil yield was observed, confirming a positive effect when alumina is present compared to the non-catalytic reaction. As PP comes in contact with alumina support Lewis acid sites, the degradation of PP to lower molecular weight compounds increases sharply [37]. As is known, the catalytic degradation of PP occurs via an ionic mechanism through two steps. The first step is the abstraction of hydride ion from the hydrocarbon polymer which is promoted by the Lewis acid sites of the alumina, where the second step is the formation of a variety of hydrocarbon isomers due to isomerization reaction and β–scission [35,38]. From these results, it can be inferred that alumina as a catalyst is very promising for PP conversion in terms of high oil yield as compared to PJ.With the presence of transition metal oxides, the conversion of PJ alone, contrastingly, showed an increase in the bio-oil yield for all the catalysts tested in the order of Nb > Ni > Mo > W (22.6%- Nb, 3.8%- Ni, 1.7 %- Mo, and %- W improvement when compared to non-catalytic conversion). Similarly, when PP alone was used, the performance of the supported metal catalysts was exceptional increasing oil yield to 73.3%, 65.3%, 57.3%, and 54.6% for Nb/alumina, Ni/alumina, Mo/alumina, and W/alumina, respectively when compared to the non-catalytic conversion.The bio-oil yields obtained for the co-liquefaction experiments with 25%, 33%, 50%, 67%, and 75% PP substitution and at 420 °C, are then compared with the calculated anticipated value based on the corresponding individual conversion from the weight fractions of PP and PJ as shown in Fig. S4 (a-e) in ESI. At 25% PP addition, an excellent synergy between PJ and PP was observed producing a high bio-oil yield of 59.4% for Nb/alumina catalyst. The bio-oil yields obtained when using the other catalysts were 47.8%, 48.7% and 49.6% for Mo/alumina, W/alumina, and Ni/alumina, respectively. Evidently, the presence of a metal oxide catalyst improves the overall conversion of solid feed, suppressing gas product formation, thereby increasing the liquid hydrocarbons yield (Fig. 3b) [21]. An increase in the aqueous phase yield was noted, indicating the extraction of organic compounds into the aqueous phase and due to increased deoxygenation reactions, such as demethoxylation in the presence of catalyst. These results suggest that adding 25% PP to PJ is an optimum value in terms of high bio-oil yield (both in the case of catalytic and non– catalytic conversion).As established Nb/alumina catalyst was found to be the best choice in terms of bio– oil yield, then it was decided to further optimize the catalyst weight percentage for the 25% PP added to PJ during HTL reaction by investigating the bio-oil yield obtained when a catalyst loading of 1, 2, 3, 4, and 5 wt.% was used. At the low catalyst loading of 1 wt.%, 24% solids, 42% bio-oil, 17.2% aqueous phase and 16.8% gases were produced. The low bio-oil liquid yield obtained can be attributed to the limited catalyst amount used to drive the conversion to oil, therefore, the oil production is rivalled by gas and solid product formation. The increase in catalyst load from 1 wt.% to 2 wt.%, increased the bio-oil yield to 59.4%, while decreasing solid residue formation by 50% (24%–1 wt.% to 12%–2 wt.%) suggesting the effective conversion of organic compounds into HTL products. Further increasing catalyst loading results in an increase in gas phase formation due to further decomposition of low molecular weight hydrocarbons from bio-oil due to strong catalyst acidity. Gradual increase in solid residue formation was also observed at increased catalyst load which would have been due to the re–polymerization of oil intermediates. Hence, 2 wt.% of Nb/alumina catalyst was the optimum loading for the HTL reactions. Table 3 shows the physicochemical properties of the bio-oils obtained from the non-catalytic and catalytic HTL tests conducted at 420 °C and with 25% PP addition to PJ biomass. The 75% PJ–25% PP blend feedstock contains 55.7% C and 6.4% H with a net hydrogen to carbon ratio (H/C)eff of only 0.38. C and H in the bio-oil obtained from the non-catalytic HTL was 59.5% and 6.7%, respectively, with an increased (H/C)eff of 0.52. Due to the low bio-oil yield and carbon loss through solid and gas products, only 40.1% carbon was recovered into the oil phase. Evidently, the catalytic runs produced much improved bio-oil in terms of (H/C)eff and carbon recovery as can be seen in Table 3. With Nb catalyst, about 79% carbon was recovered to the oil phase with (H/C)eff as 1.13.In terms of oxygen, the non-catalytic bio-oil contained 32.3% oxygen corresponding to 16.5% bio-oil deoxygenation when compared to the feed. During the catalytic runs, up to 65.1% deoxygenation was achieved with Nb/alumina catalyst as occurrence of several reactions during the HTL, which is discussed in the following sections. It must be also noted that 1.7% sulphur was present in the feed and brought down to 0.1% when Nb/alumina catalyst was used for the HTL reaction, supporting the evidence that Nb is an efficient catalyst for desulphurization reactions as well [39]. Nb catalysts has been reported prominent for dehydration due to its Lewis and Bronsted acid sites [40]. Higher heating value (HHV) is the heat produced upon complete combustion is one of the vital properties of bio-oil which is influenced by factors such as % of heteroatom, H/C ratio, etc. The HHV of the feedstock was 23.54 MJ/Kg which was also improved while employing a catalyst and a maximum of 35.08 MJ/Kg was observed with Nb catalyst.Overall, by comparing the bio-oil properties, Nb/alumina catalyst performed exceptional in terms of improving bio-oil properties such as HHV and in terms of % deoxygenation and % carbon recovery.With respect to product distribution, the bio-oil obtained from the catalyzed HTL of PP alone at 420 °C was analyzed by GC-MS and n–paraffin, i–paraffin, olefin, naphthene and aromatics hydrocarbons in the range of C7 to C18 were observed. The main products identified from GC-MS were methyl cyclohexane (C7H14), methylhexane (C7H16), 2,4–dimethyl–1–heptene (C9H18), 2– decene–2,4–dimethyl (C12H24), hexyl cyclohexane (C12H24), 3–ethyl–5–methyl–1–propyl cyclohexane (C12H24), 1,4–dicyclohexylbutane (C16H30), undecylcyclohexane (C17H34), pentadecene (C17H34), and dodecylcyclohexane (C18H36). The aromatic product includes toluene (C7H8), trimethylbenzene (C9H12) and xylene (C8H10). Escola et al. has reported the formation of C1-C5 range products with highly acid catalyst formed by the end-chain scission reaction [41]. The other two conversion pathways of PP are: (i) oligomerization of the produced olefinic gas products, and (ii) the cracking reactions occurring at random position of the polymer chain [41].The composition of the obtained bio-oils from PJ and the co-liquefaction studies with different PP concentrations were quantified also by GC-MS analysis and categorized into seven major classes. (i) guaiacolics, (ii) aromatic hydrocarbons, (iii) acids, aldehydes, and ketones (iv) alkyl phenolics, (v) catechols, (vi) naphthalene oligomers, and (vii) alkanes. The detected naphthalene compounds and undetected large naphthalene molecules were considered as naphthalene oligomers. Biomass undergoes decomposition and de–polymerization during the initial HTL process temperature which further produce smaller molecules through addition, cracking, hydrogenation, oxidation and nucleophilic reactions [42]. Fig. 4 shows the selectivity towards bio-oil components for the non-catalytic and catalytic HTL runs with 25% PP in PJ at 420 °C for 60 min and 2 wt.% catalyst. The non– catalytic HTL of 25% PP blend produced guaiacolics (42%), followed by acids, aldehydes, and ketones (26%). 12% of completely oxygen-free compounds: aromatic hydrocarbons were produced as a result of the deoxygenation reaction taking place under non-catalytic hydrothermal condition. It was noticed that 6% alkyl phenolics, 6% catechols, and 8% naphthalene oligomers were the other product classes identified (Fig. 4). During the HTL reaction, repolymerization and condensation reactions occur that produce oligomers. There were no alkanes detected from GC– MS. As it is well known, biomass constitutes cellulose, hemicellulose, and lignin components. The presence of major derivatives compounds from lignin in the bio– oil can be attributed to the more solubility of cellulose derived compounds in water.During the catalytic HTL reaction of 25% PP added to PJ, interesting results were observed. The guaiacolics selectivity decreased from 42% in case of non-catalytic to 32%, 34%, 33% and 29%, when the Mo, Ni, W and Nb catalysts are used, respectively. There was a simultaneous increase in the yield of aromatic hydrocarbons which can derive to the inference that the deoxygenation of guaiacolics takes place in the presence of metal oxide catalysts to produce aromatic hydrocarbons. Moreover, the bio-oil average molecular weight was calculated by GPC and results shown in Table 3. The non-catalytic HTL of 25% PP blend produced bio-oil with an average molecular weight of 692 g/mol. Whereas, during the catalytic HTL, the average molecular weight decreased to 526, 582, 424 and 368 g/mol for alumina supported Mo, Ni, W and Nb catalysts, respectively. This indicates the effective cleavage of C–C bonds in biomass in the presence of catalyst. It has been reported that Nb2O5 possess exceptional hydrogenolysis activity by selectively cleaving Caromatic – C lignin bonds, while suppressing hydrogenation reaction, when compared to other supports such as ZrO2, Al2O3, TiO2 [43]. The exceptional dehydration capacity of Nb catalyst can be ascribed to the oxophilic nature of Nb2O5 (XPS spectra in Fig. S5 in ESI) that possess an unique dehydration potential due to the strong interaction between the Nb5+/Nb4+ and the oxygen atom of the guaiacol molecule [44]. Xia. et al. reported that the C– O bond cleavage in tetrahydrofuran ring performed by Nb– O– Nb is the result of an increase of acidity of NbOx that favors an increase in the rate of dehydration reaction [45]. The direct conversion of biomass derived carbohydrates and glucose involves the dehydration to produce hydroxymethylfurfural and Nb based catalysts has been reported to be promising for this direct conversion [46,47]. Additionally, the selectivity to acids, aldehydes, and ketones, were lesser during catalytic HTL when compared to the non– catalytic conversion (from 26%– non catalytic to 19%, 12%, 13%, and 17% with Mo, Ni, W and Nb, respectively).The decrease in selectivity to this group of compounds along with the associated increase in CO2 and CO gases infer that decarboxylation and decarbonylation reactions of the acid, aldehyde and ketone groups were taking place [48]. The decarboxylation and decarbonylation reactions are accompanied by the formation of CO2 and CO, respectively which can be observed in the gas products (Fig. S6 in ESI).In contrast, catalytic HTL increased the selectivity to alkyl phenolics when compared to non–catalytic HTL (from 6%– non catalytic to 9%, 11%, 12%, and 12% with Mo, Ni, W and Nb, respectively). Alkylation of guaiacol aromatic ring is a common reaction that occurs in the presence of an acidic catalyst and an alkyl source under hydrothermal conditions [49]. Demethylation and demethoxylation of guaiacol produces CH4 and CH3OH, where this methyl group could alkylate the aromatic ring producing alkylated products as a result of the catalyst acidity (Table 1) [5]. The next class of product compound, catechols and naphthalene oligomers showed a yield decrease due to the presence of catalyst. The catechols and naphthalenes have been reported to produce coke through condensation reactions, [50] therefore, the decrease in selectivity to naphthalene oligomers and catechols implies the decomposition of these coke precursors in the presence of catalyst. The presence of acidic catalyst also promotes the formation of gases from PP by the end-chain cleavage mechanism. A complete deoxygenation of guaiacolics and alkylphenolics results in the formation of aromatic hydrocarbons and alkanes as can be seen in Fig. 4.The efficiency of the Nb2O5/alumina catalyst was further investigated for its catalytic reusability. After reaction, the catalyst was separated from the reaction mixture and regenerated by burning off the deposited coke at 400 °C in a muffle furnace. The reusability tests were then conducted using the same reaction conditions as the fresh one. There is a catalyst loss of about ∼3.4% every time which was compensated from a fresh batch. The yield of bio-oil, aqueous phase, biochar, and gases from each reaction were quantified and the corresponding % deoxygenation and carbon recovery to bio-oil phase were calculated, and the data presented in Fig. 5 . The catalyst performed remarkably up to 10 reaction cycles maintaining a high bio-oil yield with a marginal decrease (59.4% yield–1st cycle to 55.2%–10th cycle). The decrease in bio-oil yield was also followed by a decrease in the % carbon recovery to bio-oil from 78.9% in 1st cycle to 75.8% in 10th cycle which is not a significant loss. An increase in the biochar yield was observed from 12% to 15% after 10th cycle. These results demonstrate that the catalyst is promising for reusability.For a comparison, the catalyst retrieved after first cycle was tested for reusability without regeneration (burning coke at 400 °C), and as expected the bio-oil yield decreased from 59.4% to 55.6% whereas gas and biochar yield sharply increased. This is due to the coke deposited on the catalyst surface covering Nb2O5 active site, therefore, hindering the contact of reactant with the catalyst active acid species. Therefore, the regeneration of the catalyst was essential after every reaction cycle. The catalyst deactivation during HTL reactions have been reported to occur due to multiple factors such as leaching of active metals to the liquid medium,[51] catalyst coking that blocks the pores and masks the active sites,[51] the presence of high concentration of hetero atoms in the feed,[52] etc. On the other hand, ɣ-alumina tend to deactivate in hot water and change phase to aluminium oxide hydroxide (boehmite) that also contains Lewis acid sites [53]. However, this phase change could lead to catalyst deactivation [54].In this study, PJ was converted to bio-oil by hydrothermal liquefaction process that can be further upgraded to biofuels or platform chemicals. To improve the bio-oil yield and quality, a hydrogen rich co-reactant PP and solid acid catalysts were employed. The synergistic interaction between the PJ and PP (25% PP substitution to PJ) at HTL reaction temperature of 420 °C increased the oil yield to 46.5% from 42.5% obtained when using PJ alone. Among the catalysts, Nb based catalysts showed high selectivity and efficiency for deoxygenation of liquid biomass compounds resulting in high hydrocarbons with reduced oxygen content, making them very suitable for conversion into transportation platform fuels. Nb/Al2O3 was reasonable stable up to 10 reaction cycles. This strategy could be useful both for efficient valorisation of PJ and recycling of PP waste into high value fuels and chemicals which could potentially benefit to Indian farmers, rural industries-based bioeconomy, and municipalities strategies for plastic waste management.This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Swathi Mukundan: Investigation, Writing – original draft, Supervision, Funding acquisition. Jonathan L. Wagner: Writing – review & editing, Conceptualization. Pratheep K. Annamalai: Writing – review & editing, Conceptualization. Devika Sudha Ravindran: Formal analysis. Girish Kumar Krishnapillai: Writing – review & editing, Supervision. Jorge Beltramini: Writing – review & editing, Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Dr. SM thankfully acknowledge the University Grants Commission- Dr. D. S. Kothari Postdoctoral Fellowship Scheme for sponsoring the research. The first author sincerely appreciates the facilities provided by SAIF STIC, Cochin University of Science and Technology, Kochi. Supplementary material Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuproc.2022.107523.

This study reports an efficient conversion route for prosopis juliflora (PJ) biomass into high-quality bio-oil through catalytic hydrothermal liquefaction (HTL) process with systematically substituted hydrogen-rich plastic waste ‘polypropylene (PP)’, and using alumina supported metal oxide (Mo, Ni, W, and Nb) catalysts. The HTL treatments of PJ with PP (0-75 wt.%) were investigated in both sub and supercritical water conditions. An excellent synergy between PP and PJ was observed even in subcritical conditions (97.6% synergy at 340 °C at 25% PP to PJ), while efficient liquefaction of PP alone was observed only in the supercritical conditions. The optimum temperature, and PP substitution were found to be 420 °C and 25% respectively, with 46.5% bio-oil yield, high deoxygenation (65.1%), and carbon recovery (78.9%) when using Nb/Al2O3 as the catalyst. An in-depth analysis of physicochemical properties and the bio-oil product distribution with respect to each catalyst and PP/PJ substitution ratio are discussed in detail. Among all, the Nb/Al2O3 catalyst performed well with remarkable recyclability up to 10 cycles. The produced bio-oil mixture due to its low oxygen content is very promising to be upgraded to precursors for chemicals and transportation biofuels.

• This study did not generate a new code. • The supplemental information includes all datasets generated and analyzed during this study. • Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. This study did not generate a new code.The supplemental information includes all datasets generated and analyzed during this study.Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.Syngas, a mixture of CO and H2, is widely used in modern chemical industries such as hydroformylation, Fischer-Tropsch synthesis, and the production of alkanes, olefins, and oxygenates. 1 , 2 , 3 Given its industrial importance, a variety of syngas production methods have been developed, mainly using natural gas, petroleum, or coal as raw materials. Among all the synthesis pathways, dry reforming of methane (DRM) attracted considerable attention recently, because this process would convert two greenhouse gases, methane and carbon dioxide, into useful syngas. 4 , 5 , 6 In addition, compared with an alternative steam reforming of methane, 7 , 8 DRM has the advantage of producing syngas with a low H2/CO ratio.Despite the great potential of DRM as one can imagine, there still exist some technical problems to industrialize this process. The most important one is the absence of an efficient catalyst. Many noble metal catalysts show excellent catalytic performance but are rather expensive, preventing the large-scale usage of these catalysts in the industry. 9 , 10 , 11 It was found that the inexpensive earth-abundant transition metal nickel gives rise to good catalytic activity for the DRM reaction as well. 9 , 12 , 13 , 14 , 15 Nevertheless, there exists another inevitable drawback of nickel, which is deactivation due to the deposition of coke. It was found that the deactivation mainly results from the formation of inert carbon structures on the catalyst surface during the DRM process. In the reaction network, the relevant carbon production reactions are the dissociation of methane and the disproportionation of carbon monoxide. In addition, the accumulation of coke on nickel is a macroscopic scale phenomenon contributed by different surfaces. According to our previous work, 16 , 17 (211) surface of nickel is a typical surface where coke can form easily, while (111) is not a suitable surface for carbon deposition.As carbon atoms deposit on the nickel surface, two types of Ni-carbon structures would be possible. One structure is formed through atomic carbon diffusing into the bulk phase of Ni and forming a Ni carbide structure, namely a Ni-C bulk structure. 18 , 19 , 20 The other is that carbon atoms adsorb on the surface of Ni and poison the active site, forming a Ni-C surface structure. While most previous studies focused on the Ni-C bulk structure, the characteristics of the Ni-C surface structure in tuning DRM activity were rarely reported. It has been reported previously that Ni(100) is a surface suitable for the deposition of carbon atoms on its surface sites. 21 , 22 , 23 Therefore, Ni(100) can be a typical surface for the formation of Ni-C surface structure. Besides, according to previous Wulff construction study results, Ni(100) is the secondly abundant surface; the proportion is even higher than Ni(211), over Ni nanoparticles with a size around 5–10 nm, the typical size of Ni particles for the DRM reaction. 24 , 25 , 26 In order to obtain deeper understandings on the formation and dynamic behavior of this type of surface structure on Ni(100) surface, a combined density functional theory (DFT) calculation and microkinetic modeling (MKM) study was performed here. The kinetics of the DRM reaction and deposition of carbon atoms on the pristine Ni surface were studied first. Based on the results obtained, DFT calculations of the surface structures at different carbon coverages were exerted. It is worth mentioning that surface reconstruction has also been observed in the current work here, which was reported experimentally before. 27 , 28 , 29 , 30 , 31 Finally, we analyzed the kinetics of the DRM reaction features over carbon-covered Ni(100) and discovered a unique carbon-based Mars-van-Krevelen (MvK) mechanism. 32 The reaction network considered here is the same as the one reported in our previous study. 16 , 17 , 33 All the surface adsorbates are formed from the product of methane and carbon dioxide dissociation. According to early systematic studies, among all the CHx (x = 0–3) species, only C∗ and CH∗ showed noteworthy reaction activity. 18 , 34 Therefore, we only introduce here five main routes of CHx oxidation in the reaction network, namely C + O, C + OH, CH + O, CH + OH (COH), and CH + OH (CHO). The intermediates produced by these oxidation reactions finally transform into carbon monoxide, the product of the DRM reaction.Structures of all the adsorbates and transition states on pristine Ni(100) are optimized and presented in Figure S1. One can find that most adsorbates, except CH3∗ and CHOH∗, prefer the 4-fold hollow site. Based on the energies calculated for these structures, the activation barriers and reaction energies of all elementary steps involved in the network can be obtained, which are shown in Figure 1 . Compared with all other elementary steps, the oxidation reactions of C∗/CH∗ possess much higher activation energies. Among these oxidation reactions, only C∗+OH∗ has a slightly lower energy barrier of 1.68 eV, while the other three oxidation reactions all have higher energy barriers of around 2 eV.Based on these energies calculated by DFT on pristine Ni(100), MKM can be further performed to obtain important kinetic information. We plot in Figure 2 the steady-state coverage of surface dominant species, the logarithm of turnover frequency (TOF), and degree of rate control (DRC) obtained from MKM studies against reaction temperatures. One can clearly find that the steady-state coverage of C∗ is almost 1 monolayer (ML), suggesting that C∗ would occupy almost all the 4-fold hollow sites. This result is consistent with the tendency of carbon accumulation on Ni(100) surface reported before. 35 The TOF of the overall reaction varies to a small extent as temperature increases. The calculated TOF related to the key elementary steps that determine the preferred reaction pathway is shown in Figure S4, and the dominant route over pristine Ni(100) is found to be C + O/OH.One can find from Figure 2C that the DRC results of reaction intermediates are quite simple since C∗ is the only adsorbate with high coverage (∼1 ML) on the surface. Therefore, C∗ is the rate-controlling intermediate with a constant DRC value of −2. The transition states of the oxidation of surface carbon, i.e. C-O and C-OH, have higher DRC values at a relatively low temperature. As temperature increases, DRC of the transition state of CO2 dissociation (CO-O) increases rapidly and is finally close to 1. More importantly, it is obvious that all the transition states with high DRC are related to carbon elimination. According to detailed MKM analyses performed by our group, when the DRC value of carbon elimination process is high, carbon atoms would be accumulated on the surface as the C∗ formation process is quasi-equilibrated and fast. 8 , 16 , 17 , 33 , 36 , 37 It should be mentioned that we also added the side reaction of water formation into the DRM reaction network and found that the influence of such reverse water-gas shift reaction on the kinetics is almost negligible (see Figure S4), which is consistent with previous studies by our group. 17 , 33 Recent studies from our group reported that the conventional MKM approach presents some major failures. 37 , 38 , 39 , 40 , 41 As one can find in the above section, the main discrepancy here is that the C∗ coverage obtained from MKM at the steady state, i.e. 1 ML, is not consistent with the initial coverage used for DFT energy calculations, which is 0 on pristine Ni(100). Meanwhile, it was observed by DFT calculations that the adsorption of carbon atoms at the subsurface sites of Ni(100) is less stable than that on the surface. 23 Furthermore, according to the literature, the appropriate temperature window for DRM reaction is 650°C–850°C, 42 also indicating the results in Section 3.1 are not consistent with the experimental results reported before. Therefore, the most useful information one can obtain from the results in Section 3.1 is that surface carbonization is very likely to occur on Ni(100). Based on this observation, we further calculated the adsorption free energy of carbon as a function of coverage over Ni(100).Since a p(4 × 4) supercell of Ni(100) is used here, we only considered the surface structures covered by carbon atoms with an increment of 1/16 ML. In addition, we find that the surface will get reconstructed at high carbon coverage after structural optimization. According to a previous study, this phenomenon is named as a clock-type surface reconstruction, which is common on Rh(100), Pd(100), and Ni(100). 27 , 28 , 29 , 30 , 31 Therefore, all the possible structures of both original and reconstructed surfaces under different carbon coverages were considered in the DFT calculations. We find from the results that the preferred adsorption structures of multiple carbon atoms would give rise to minimized interactions between atoms.Differential adsorption free energies of carbon atom on the most stable surface structures are plotted in Figure 3 against carbon coverage from 1/16 ML to 10/16 ML. Since our slab model for Ni(100) is a four-layer p(4 × 4) supercell, each addition of carbon atoms on the surface corresponds to a carbon coverage increment of 1/16 ML. The red and green curves show the differential adsorption free energy of C∗ on the original and reconstructed Ni(100) surface, respectively. The differential adsorption free energy of carbon atom at different coverages can be calculated with the following equation: (Equation 1) G a d s ( N 16 ) = G s l a b + N × C − G s l a b + ( N − 1 ) × C − ( μ C H 4 − 2 μ H 2 ) where G a d s ( N 16 ) , G s l a b + N × C , and G s l a b + ( N − 1 ) × C are differential adsorption free energy of the Nth C∗, total free energy of the surface structure with N C∗, and (N-1) C∗, respectively. μ C H 4 and μ H 2 are the chemical potential of methane (0.5 bar) and hydrogen (0.05 bar), respectively, at 873 K.We find from Figure 3 that, when carbon coverage is lower than 6/16 ML, the reconstructed surface is unstable and will change back to the original structure after optimization, giving rise to the same adsorption energies. With surface carbon coverages between 6/16 and 8/16 ML, carbon adsorption on the reconstructed surface is more stable than that on the original one. More importantly, the differential adsorption free energies of carbon over these structures are all close to the one over the pristine surface (−0.74 eV), e.g. −0.63 eV at 6/16 ML, −0.67 eV at 7/16 ML, and −0.50 at 8/16 ML. In addition, it is worth noting that the differential adsorption free energy will not be above zero until coverage reaches 9/16 ML. From the structures shown in Figure 3, the ninth carbon atom has to locate at the unstable 3-fold hollow site on the reconstructed surface. Meanwhile, upon the adsorption of the 10th carbon atom, the surface structure is significantly deformed, further suggesting that the adsorption of more carbon atoms will be strongly endergonic.It is also worth noting that, from previous experiments, the reconstruction will not occur until the coverage reaches 0.33 ML, 30 which is consistent with our DFT results (between 5/16 and 6/16 ML). From both DFT results and experimental evidences, it is reasonable to assume that pristine Ni(100) surface will accumulate carbon atoms and finally form a stable Ni-C surface at 8/16 ML carbon coverage.Over the reconstructed Ni(100) surface with 8/16 ML of carbon coverage, the structure of which is shown in Figure 3C, there would be two possible mechanisms of the DRM reaction, namely the surface reaction mechanism and the carbon-based MvK mechanism.Regarding the surface reaction mechanism, the reaction pathways are identical to those studied over the pristine Ni(100) surface, and five routes introduced in Section 3.1 are considered here, i.e. the C + O, C + OH, CH + O, CH + OH (COH), and CH + OH (CHO) routes. It should be mentioned that, over the carbon-covered reconstructed Ni(100), we calculated the adsorption of surface species over all possible sites, including the 3-fold metallic sites and the 4-fold sites with one carbon atom at the center, and the structures presented in Figure S2 are the most stable ones. From these stable structures, we find all the reactions occur at 3-fold metallic sites and therefore denote these active sites as ∗.The energy profile of the DRM reaction following this mechanism is shown in Figure 4 A. Comparing this energy profile with the one shown in Figure 1 over the pristine Ni(100), one can find that the desorption of products is much more difficult over the carbon-covered reconstructed Ni(100). In addition, the dissociation reactions of reactants have to overcome high activation energies and reaction energies, e.g. CO2 dissociation gives an activation energy of 3.01 eV and a reaction energy of 1.98 eV, suggesting that both reactants may be difficult to dissociate on this surface. Meanwhile, oxidation of C∗ or CH∗ is much easier, e.g. CH + OH shows an activation energy of 0.95 eV. These results indicate that carbon atoms are difficult to form and easy to be eliminated on this surface. We will show later that this is important to understand the different steady-state carbon coverage obtained from MKM simulations.The other mechanism considered is the carbon-based MvK mechanism (see Scheme 1 ). In this catalytic cycle, the DRM reaction would be initiated by the elimination of one surface carbon atom, through the reaction between CO2 and this carbon atom, to produce two CO molecules and to leave one carbon-vacancy site on the surface. This reaction is the reverse reaction of the classical Boudouard reaction, following which CO2 is dissociated in a concerted way. The transition state structure of this reaction is shown in Figure 4B. This reaction is “spin forbidden” in gas phase and thus needs to overcome a high energy barrier. 43 However, according to experiments, nickel is a suitable catalyst for this reaction, 44 which is consistent with our computation result. Subsequently, CH4 will get dissociated at this carbon-vacancy site to close the catalytic cycle. This mechanism is quite similar to the traditional oxygen-based MvK mechanism, which is an important reaction mechanism over metal oxide catalysts, 32 and normally the key species participating in the catalytic cycle is lattice oxygen and oxygen vacancy.The energy profile of this carbon-based MvK mechanism is presented in Figure 4B. Interestingly, we find all the reactions occur at 4-fold metallic sites and denote these active sites as #. It is obvious that dissociation of CH4 is much easier with a barrier of 1.19 eV compared with that in the surface reaction mechanism (the barrier is 1.66 eV). All the elementary steps of methane dissociation share similar energy trends compared with those on a pristine nickel surface. In addition, the barrier for carbon-assisted CO2 dissociation is 1.43 eV, which is slightly more difficult than at the 4-fold hollow sites on pristine surface but much easier than at the 3-fold sites on the reconstructed surface.We plot the steady-state coverage of surface-dominant species, logarithm of TOF and DRC obtained from MKM studies concerning the surface reaction mechanism, and the carbon-based MvK mechanism in Figure 5 against reaction temperatures. The influence of side reverse water-gas shift reaction is also found negligible over this surface (see Figure S4).Distinct steady-state coverage of surface carbon can be found from Figures 5A and 5D, when the surface reaction mechanism or the carbon-based MvK mechanism is considered. Almost no adsorbate has distinctive coverage when the surface reaction mechanism is considered, and therefore only a negligible amount of carbon (<10−10 ML) would be observed at the steady state. In comparison, the carbon-based MvK mechanism would suggest that the carbon atoms should occupy almost all the active sites. It should be noted that the 4-fold hollow sites are taken as the active sites in the kinetic model for the carbon-based MvK mechanism; therefore, the carbon vacancy should be readily replenished and the catalytic cycle can be closed.TOF of the carbon-based MvK pathway is much higher than that of surface reaction pathway, suggesting that these two kinetic models would give significantly different reactivity of the DRM reaction. In addition, compared with the TOF over pristine surface, TOF of carbon-based MvK mechanism is even higher. These results indicate that the mechanism driven by CO2 concerted dissociation reaction has higher reactivity than conventional DRM mechanism at high carbon coverage. This provides a detailed understanding on the observations reported in a recent study at the molecular level. 45 For surface reaction mechanism, the calculated TOF related to the key elementary steps that determine the preferred reaction pathway is shown in Figure S4, and the dominant route is found to be CH + OH (CHO).DRC results obtained for the surface reaction mechanism are presented in Figure 5C. The transition states with non-zero DRC values are CO2 dissociation and oxidation of surface CHx species, mainly CH-OH and C-OH. In contrast, the only rate-controlling transition state determined from the carbon-based MvK mechanism is C-CO2, the transition state of carbon-assisted CO2 dissociation, as shown in Figure 5F. Since almost all the 4-fold hollow active sites are filled by C∗ at steady state, only C∗ has a non-zero DRC value.To analyze the relevance of these two mechanisms, we also constructed a microkinetic model combining all these reactions at all possible sites, and the migration of all adsorbates between different sites was also included. The microkinetic modeling results are presented in Figure S5. One can find that all the active # sites are occupied by carbon atoms, while there is almost no adsorbate adsorbed in the active ∗ sites. TOFs related to CO production obtained from the combined model are found to be identical to those of the preferred carbon-based MvK mechanism. According to the DRC results, the important transition state and intermediate are both within the carbon-based MvK mechanism. These results further suggest that the carbon-based MvK mechanism should be preferred.According to the results presented above, the C∗ formation process can be considered quasi-equilibrated in both mechanisms. This means that once the vacancy at the 4-fold hollow site is created, it always tends to be replenished by C∗. Meanwhile, the C∗ is difficult to form at the 3-fold hollow sites. In our simulation model, this reaction process can be described as an oscillation between 7/16 ML carbon-covered and 8/16 ML carbon-covered surfaces. Once an 8/16 ML carbon-covered surface structure is formed, the further accumulation of surface carbon will be prevented. In the meantime, a new reaction pathway, i.e. the carbon-based MvK pathway, becomes dominant. This reaction pathway firstly creates a vacancy at the 4-fold hollow sites and a 7/16 ML carbon-covered surface structure is formed, then changes to the 8/16 ML carbon-covered surface structure upon dissociation of CH4 at the vacancy sites.The above result indicates that carbon-based MvK mechanism is favored over Ni(100) surface at high carbon coverage. It is intriguing to extend this result to other typical surfaces of nickel nanoparticles. In our previous studies, the activity of Ni(111) and Ni(211) has been thoroughly studied. 17 , 36 Combining all the results reported, we found that the TOF is increasing from Ni(100) and Ni(111) to Ni(211). Therefore, one can find that Ni(211) possesses the highest activity for the DRM reaction, but the high coverage of carbon might result in deactivation and coke formation, which would be an interesting topic for future studies. It should be mentioned that a similar analysis approach introduced in the current work would be helpful for future studies on Ni(211).In addition, several key information that can be used to help design more stable Ni catalysts for the DRM reaction can be obtained from our results. Firstly, not all the surface sites would contribute to the formation of coke during the DRM, which is consistent with recent work that only part of the surface sites on nickel nanoparticles lose their activity during DRM reaction. 45 More importantly, different from the catalyst designing idea widely reported previously to find the surface where no carbon tends to accumulate, our research presents a new idea that the adsorbed carbon may prevent the formation of poisonous coke. Combining both strategies of preventing carbon deposition, we suggest that a low-index flat surfaces, i.e. Ni(100) and Ni(111), tends to be coke resistant, although the mechanisms might be different over these surfaces, while surfaces with defects may deactivate quickly, e.g. Ni(211). Therefore, in order to prevent coke-induced deactivation of Ni catalyst during the DRM reaction, the catalyst particles should possess as few defects as possible.In the current work, thorough understandings on the DRM reaction over Ni(100) are obtained. We find that carbon deposition and accumulation will be spontaneous over the pristine Ni(100) surface, because the steady-state coverage of carbon obtained from microkinetic modeling is as high as 1 ML. In the meantime, as the coverage of carbon on the surface increases, surface reconstruction would happen and form a stable Ni-C surface structure, which brings more interesting properties. The optimal coverage of carbon over Ni(100) surface is found to be 0.5 ML and further adsorption of carbon is endergonic. Through comparing the surface reaction and the carbon-based MvK mechanisms for the DRM reaction over carbon-covered reconstructed Ni(100) surface, we find that the latter dominates under reaction conditions and shows even higher activity and promoted coke resistance. The unique carbon-based MvK mechanism was rarely reported in previous studies on the DRM reaction, and the current work not only provides evidence of the existence of this mechanism but also quantitatively determines the detailed kinetic information such as reaction rate and the rate-controlling step over the carbon-covered reconstructed Ni(100) surface. The possible strategies for the promotion of Ni catalyst stabilities during the DRM reaction are proposed.Our work is a theoretical research of DRM reaction behavior on Ni(100) by combining DFT calculations and microkinetic modeling. The catalyst design strategy is proposed that augmenting proportion of Ni(100) and Ni(111) is beneficial to coke resistance, while the experimental investigation over this topic needs to be further conducted. REAGENT or RESOURCE SOURCE IDENTIFIER Software and algorithms VASP 5.4.1 VASP Software GmbH https://www.vasp.at CATMAP SUNCAT, Stanford University https://github.com/SUNCAT-Center/catmap Further information and requests should be directed to and will be fulfilled by the lead contact, Bo Yang (yangbo1@shanghaitech.edu.cn).This study did not generate new unique material.All the DFT calculations were performed with the Vienna Ab-initio Simulation Package (VASP). The electron-ion interaction was described by the projector-augmented wave (PAW) formalism, with an energy cutoff of 500 eV. 46 Bayesian error estimation functional with van der Waals correlations (BEEF-vdW) was used to describe electron exchange and correlation. 47 Spin polarization was considered for all the calculations. The Methfessel-Paxton smearing method was used here with a broadening of 0.1 eV. 48 All adsorption configurations were optimized by a force-based conjugate gradient algorithm and the transition states were determined by using a constrained minimization method. 39 , 49 , 50 All the identified transition states were further confirmed with vibrational analysis to ensure that only one imaginary frequency, corresponding to the bond breaking/formation, was obtained. The force convergence criterion was set to 0.05 eV/Å while the total energy convergence criterion was 10−4 eV. From previous studies, theoretical calculations with such parameter settings are sufficient to simulate the experiment results. 39 , 51 , 52 , 53 We built a 4-layer p(4 × 4) supercell for the Ni(100) surface and the bottom two layers were frozen during structural optimization. A vacuum layer with a height of 15 Å was used here to avoid interactions between periodic structures. A k-point mesh of 3 × 3×1 was used here for all the adsorption structure optimization and transition state searching.Adsorption energy was calculated from (Equation 2) E a d = E s l a b + a d s − E s l a b − E a d s where E a d , E s l a b + a d s , E s l a b , and E a d s are adsorption energy, energy of the slab after adsorption, energy of the slab and energy of the adsorbate in gas phase. Although DFT-calculated gaseous energies of CO2 and CO may be inaccurate, and this error might result in flawed microkinetic modeling results, 51 , 52 , 53 we found that no other correction should be made for the energies of CO2 and CO while studying the DRM reaction using the above DFT parameter settings. 17 We utilized the CatMAP package in the microkinetic simulations. 54 Steady state approximation and transition state theory were used for microkinetic modeling in this package. The temperature considered in our MKM study is within the range of 873–1073 K. To be consistent with the experimental conditions reported, the total pressure of reactants here was selected as 1 bar. The ratio between two reactants, CH4 and CO2, was fixed at a proportion of 1:1, and the conversion considered was 5%. Thermodynamic corrections were included here for both gas-phase molecules and surface adsorbates with Shomate equations and harmonic approximation, respectively. Here, we consider a model with two types of active sites, one for hydrogen atoms and the other for the remaining adsorbates. In this approach, hydrogen is adsorbed at a special “hydrogen reservoir” site and does not compete with other adsorbates, because hydrogen has almost zero interaction with all adsorbates, including itself. This approach has been widely used in MKM studies by several groups. 51 , 55 , 56 More details regarding the parameter settings used in MKM studies can be found in the supplemental information.The degree of rate control (DRC) analysis method, developed by Campbell and co-workers, was applied in our study to obtain deeper understandings of the MKM results. 57 , 58 , 59 , 60 , 61 The values of DRC can be calculated with (Equation 3) X i = ( − ∂ ln r ∂ ( G i 0 / k B T ) ) G j ≠ i 0 where X i represents the DRC of a transition state or an intermediate i, r is the rate of the overall reaction and G i 0 is the Gibbs free energy of i. The relative magnitude of DRC is in line with the response of reaction rates to the change in the free energy of a given intermediate or transition state. A positive DRC value means the reaction rate increases when lowering the Gibbs energy of i, while a negative value implies the opposite influence. One can find that a transition state should normally possess a positive value. In addition, the higher the DRC value is, the more important this transition state would be to the overall reaction. The maximum for a DRC value of a transition state is 1, which indicates this transition state controls the overall reaction rate completely. Meanwhile, DRC value of an intermediate is negative and equal to the coverage of the intermediate multiplied by the number of active sites required for the elementary reaction.This work is financially supported by the National Natural Science Foundation of China (22072091, 91745102, 92045301), Shanghai Rising-Star Program (20QA1406800), and ShanghaiTech University. We thank the HPC Platform of ShanghaiTech University for computing time.Z.G. performed the DFT calculations and microkinetic simulations related to carbon-covered surfaces. S.C. performed DFT calculations related to clean surface. B.Y. conceived the problem. All the authors contributed to writing the paper.The authors declare no competing interests.Supplemental information can be found online at https://doi.org/10.1016/j.isci.2023.106237. Document S1. Figures S1–S5 and Table S1

Dry reforming of methane (DRM) is an efficient process to transform methane and carbon dioxide to syngas. Nickel could show good catalytic activity for DRM, whereas the deactivation of nickel surfaces by the formation of inert carbon structures is inevitable. In this study, we carry out a detailed investigation of the evolution and catalytic performance of the carbon-covered surface structure on Ni(100) with a combined density functional theory and microkinetic modeling approach. The results suggest that the pristine Ni(100) surface is prone to carbon deposition and accumulation under reaction conditions. Further studies show that over this carbon-covered reconstructed Ni(100) surface, a carbon-based Mars-van-Krevelen mechanism would be favored, and the activity and coke resistance is promoted. This surface state and reaction mechanism were rarely reported before and would provide more insights into the DRM process under real reaction conditions and would help design more stable Ni catalysts.

Methane is the primary component of natural gas, shale gas, biogas and combustible ice. Methane conversion has been a hot topic for several decades. With the fast exploration of shale gas, methane conversion has even attracted increasing attentions (Bian et al., 2017; Yi et al., 2015; Pan et al., 2010). Among various options for methane conversion, CO2 reforming of methane or dry reforming of methane (DRM) is promising, because this reaction can convert methane and carbon dioxide into valuable chemicals in a large scale with reasonable H2/CO ratio for further syntheses. Compared with the noble metal-based catalysts, the nickel-based catalyst has high activity but relatively low price. It has been therefore widely investigated for CO2 reforming (Gonzalez et al., 2016; Vasiliades et al., 2018; Pudukudy et al., 2017). During the DRM reaction, methane decomposition to carbon and hydrogen was reported to be the first and rate-determining step. The formed carbon intermediate can be then oxidized by CO2 to CO. The reactivity of such formed carbon is very important for the stability of the catalyst. If the reactivity of the formed carbon is not good, the carbon will tend to aggregate into more stable carbon species with higher graphitization degree (e.g., graphite and carbon filaments). This would cause a serious carbon deposition on the catalyst surface (Aramouni et al., 2018). The active sites of the catalyst would be covered, which further suppresses the following reaction steps and finally deactivates the catalyst (Bayat et al., 2016). Therefore, the carbon intermediate for DRM must be active and easy to be removed by CO2. Rationally controlling the methane decomposition and carbon deposition is of great importance to sustain the DRM activity.The morphology and structure of the nickel-based catalyst play key roles in the structure of the carbon species produced from methane decomposition. It was previously reported (Beltrán et al., 2017; Goula et al., 1996) that the methane dissociation is a structure sensitive reaction on nickel surface with different activation energies on various nickel facets. For example, the activation energy of the dissociation of CH* on Ni (111), Ni (211), Ni3C (111) and Ni3C (001) facet was 1.35 eV, 0.52 eV, 1.14 eV and 0.86 eV, respectively (Liu et al., 2019; Wang et al., 2014). Ni (111) shows the highest activation energy, which means that the rate of carbon deposition on Ni (111) facet is the slowest. The Ni nanoparticle size also significantly affects the diameter of carbon nanotube (Seshan et al., 1998) and the growth rate of carbon filament (Ermakova and Ermakov, 2002; Lou et al., 2017).We previously reported that the decomposition of nickel precursor by dielectric barrier discharge (DBD) plasma, followed by the thermal treatment in the absence of the plasma, causes the Ni catalyst with enhanced coke resistance for DRM (Hu et al., 2019). In this work, the influence of reaction conditions and plasma catalyst decomposition on the structural properties (e.g., species, morphology and size) and the reactivity of the carbon intermediate formed from methane decomposition was further investigated. Ni/ZrO2 catalysts were prepared via two methods: one was decomposed by the DBD plasma, and the other one was prepared by thermal calcination. Compared with the calcined Ni/ZrO2 catalyst, the DBD plasma decomposed Ni/ZrO2 catalyst showed a significantly different Ni structure, which leads to the formation of carbon with improved reactivity towards CO2, which explains the enhanced coke resistance for DRM.The ZrO2 support was prepared by calcining Zr(NO3)4·5H2O (Tianjin Kemiou Chemical Reagent) at 500 °C for 3 h. A certain amount of Ni(NO3)2·6H2O (Tianjin Kemiou Chemical Reagent) was dissolved in distilled water. The ZrO2 powder was then incipiently impregnated with the prepared aqueous solution and aged at ambient temperature for 12 h. After drying at 110 °C for 12 h, one part of the dried sample was calcined at 700 °C for 2 h. The obtained catalyst was denoted as Ni/ZrO2-C. The other part was decomposed by DBD plasma, operated ca. 150 °C, under argon atmosphere for 1 h. In order to remove the undecomposed nickel nitrate, the DBD decomposed sample was washed by ionized water and alcohol, and then dried at 75 °C for 8 h. The obtained catalyst was denoted as Ni/ZrO2-P. Both of the two catalysts were reduced by hydrogen in the absence of the plasma before the activity tests for methane decomposition.The DBD plasma is a typical kind of cold plasmas with plentiful of various energetic species, like electrons, ions, radicals and excited species. The DBD plasma decomposes the nickel precursor in a rapid way. It causes a rapid nucleation but relatively slower crystal growth. This makes the DBD plasma decomposition different from the conventional thermal decomposition. The DBD plasma setup has been described in our previous works (Yan et al., 2015; Tan et al., 2018). A sinusoidal voltage with a frequency of ca. 22 kHz was generated via a high voltage generator (CTP-2000 K; Corona Laboratory, Nanjing, China). The sample powders were laid on a quartz reactor. The diameter of the quartz plate is 90 mm with a thickness of 8 nm. The DBD plasma is generated in the quartz reactor via applying an average voltage of 14 kV to a steel plate electrode.Thermal gravimetric analyses (TGA) with differential scanning calorimetry (DSC) were applied to obtain the weight loss and heat change of the catalysts. It was carried out on a Netzsch STA 449 F3 system with a heating rate of 10 °C/min (from 35 to 800 °C) under a flowing air of 100 mL/min (O2 20 mL/min, N2 80 mL/min).X-ray powder diffraction (XRD) patterns were recorded by a Rigaku D/max-2500 diffractometer with Ni-filtered Cu target and a Kα radiation source (λ = 1.54056 Å). The 2θ range was from 10° to 80° with a scanning speed of 4°/min. The acquired XRD patterns were compared using Joint Committee on Powder Diffraction Standards (JCPDSs) to identify the crystal phase of different samples. The metal particle size was calculated using Scherrer equation based on the characteristic diffraction peak.Transmission electron microscopy (TEM) was applied to study the morphology and the size of different catalysts. TEM images were obtained from a Philips Tecnai G2 F20 system equipped with an energy-dispersive X-ray spectrometer (EDX) operated at 200 kV. The catalyst powder was first suspended into ethanol and then ultrasonically dispersed for 40 min. One drop of the suspension was dripped onto a copper grid for TEM observation.Raman spectroscopy is a common method to quantitatively analyze the graphitization degree of carbon species. Raman spectra were collected on a Renishaw inVia reflex spectrometer, using a laser with an excitation wavelength of 532 nm.Temperature programmed reaction with CO2 (CO2-TPR) was performed on AutoChem II 2920 adsorption apparatus combined with a mass spectrometry. A certain amount of sample was blown with 30 mL/min He at 50 °C for 1 h to remove surface physically absorbed gas. Then, the sample was heated from 50 °C to 800 °C, at a heating rate of 10 °C/min under a gaseous mixture of CO2:He = 1:2 (30 mL/min). The gas product was analyzed by mass spectrometers.Methane decomposition was carried out in a quartz tubular fixed-bed reactor under atmospheric pressure. 80 mg catalyst was loaded in the quartz reactor and linearly heated to 700 °C under argon atmosphere. The reactor was then fed with hydrogen at 40 mL/min to reduce the catalysts for 1 h at 700 °C. After that, it was cooled down to reaction temperature under argon atmosphere. A mixed gas flow (CH4:Ar = 1:3 or CH4:Ar = 2:3) was fed into the reactor at a certain temperature. Carbon was then formed on the catalysts. The range of reaction temperature is from 450 °C to 600 °C.Generally, carbon species deposited on Ni catalysts can be divided into three types: atomic carbon (Cα), amorphous carbon (Cβ) and graphite (Cγ) (Zhang et al., 2015). As a highly reactive intermediate in DRM, Cα can be easily oxidized by oxygen under 100 °C (Noh et al., 2017). Cβ, which is converted from Cα, can be removed at 300 °C or turned into more stable form (Cγ). Onion-like carbon, carbon fibers and carbon nanotubes (CNTs), as the Cγ species, are the most inactive carbon species and need higher temperature to be oxidized. This is the main reason for the deactivation of the catalyst during DRM (Bartholomew, 2001; Al-Fatesh et al., 2017).In this work, methane decomposition under different temperature and reaction time were conducted. After methane decomposition, the TG/DSC analyses under air atmosphere were conducted to measure the amount and reactivity of carbon deposition through the mass losses and exothermic peaks of the used catalysts. The effect of reaction temperature on carbon deposition is shown in Fig. 1 . The mass losses of carbon on the plasma-treated and calcined catalysts are almost equal, indicating the same amount of carbon deposition on two catalysts. After methane decomposition at 450 °C, 500 °C and 550 °C for 30 min, the amount of deposited carbon is 15 wt%, 29 wt% and 38 wt%, respectively. Therefore, the carbon deposition increases with increasing temperature. Carbon oxidization is an exothermic process, as reflected by DSC curves. The peaks were fitted by a Gaussian-type function to figure out the different types of carbon (Fig S1 and Fig S2). For every sample, the peak at the lowest temperature should belong to the oxidization of metallic Ni. The exothermic peaks located at temperature higher than 400 °C for every sample should be assigned to the oxidation of onion-like carbon (<500 °C) and CNTs (>500 °C) (Guo et al., 2004, Zhang et al., 2015). Ni/ZrO2-P shows lower ratios of onion-like carbon to CNTs than Ni/ZrO2-C (Table 1 ). The onion-like carbon would encapsulate Ni particles, leading to the deactivation of active sites of the Ni catalyst soon. On the contrary, CNTs with Ni particles at the top would not completely cover the Ni active sites, as discussed below. Obviously, the degree of graphitization of the onion-like carbon should be lower than that of CNTs. Therefore, the onion-like carbon requires lower oxidation temperature. Yan et al. (2015) have reported that the plasma treated Ni catalysts with fewer Ni defect sites facilitate the formation of CNTs along with less onion-like carbon. Hence, Ni/ZrO2-P would present a higher resistance to deactivation than Ni/ZrO2-C. With the increasing reaction temperature, the peaks shift towards high temperatures with increasing CNTs. It can also be found that the exothermic peaks of Ni/ZrO2-P always center at lower temperatures than those on Ni/ZrO2-C. This means that the carbon deposited on Ni/ZrO2-P exhibits lower degree of graphitization. Fig. 2 presents the TG curves of the used Ni/ZrO2-P and Ni/ZrO2-C after methane decomposition for different time. The mass losses of both catalysts are less than 10 wt% after methane decomposition for 10 min. When the reaction lasts for 60 min, the mass loss (48.02 wt%) of the calcined catalyst is slightly higher than that (45.01 wt%) of the DBD catalyst. The DSC curves of the used catalysts are displayed in Fig. 3 , and the ratios of CNTs to onion-like carbon are listed in Table 1. With the time increasing, the peaks gradually shift to higher temperatures, so the deposited carbon is more difficult to be oxidized. As compared with Ni/ZrO2-C, Ni/ZrO2-P always produces less onion-like carbon. And, CNTs on Ni/ZrO2-P need lower oxidization temperature. Therefore, the DBD plasma decomposed catalyst shows a stronger carbon resistance than the calcined one. This suggests that the carbon formed on Ni/ZrO2-P has higher reactivity, leading to an improved stability for DRM (Hu et al., 2019).XRD was used to estimate the graphitization degree of deposited carbon and nickel particle size for both catalysts. The XRD patterns of the catalysts after reactions at 450 °C, 500 °C and 550 °C for 30 min are shown in Fig. 4 . The typical peak of graphite (002) facet at 2θ = 26° can be detected for Ni/ZrO2-C and its intensity increases with increasing temperature. Ni/ZrO2-P exhibits weaker graphite peaks than Ni/ZrO2-C, indicating lower degree of graphitization of the carbon formed. The average diameters of nickel particles were calculated using Scherrer equation, based on the typical peak of Ni (111) facet at 2θ = 44.5°. The results are listed in Table 2 . After methane decomposition at 450 °C for 30 min, the average diameters of Ni particles of Ni/ZrO2-P and Ni/ZrO2-C are 12.90 nm and 15.19 nm, which are similar to those of fresh catalysts (12.90 nm and 15.59 nm). No obvious sintering was observed, meaning that the catalysts kept stable structures. The particle size of Ni significantly increases when the reaction temperature reaches 550 °C, indicating that the Ni catalysts are likely to move and aggregate into larger particles. However, the size of Ni particles on Ni/ZrO2-P is still smaller than that on Ni/ZrO2-C. It has been reported that Ni/ZrO2-P mainly possesses Ni (111) as the principally exposed facet with fewer Ni defect sites, while the smaller Ni particles can provide more Ni (111) active sites (Jia et al., 2019). Therefore, Ni/ZrO2-P improves the formation of CNTs with lower graphitization degree, supporting the TG/DSC and XRD results.XRD patterns of the catalysts after reaction at 500 °C with different time are shown in Fig. 5 . With the increasing reaction time, the graphite (002) peak gradually gets stronger for both catalysts. Methane decomposition for longer time would produce more graphite-like carbon with higher crystallinity. In addition, the size of Ni particles is also influenced by the decomposition time. According to Table 3 , the Ni particles exhibit obvious aggregation with time increasing. The diameters of Ni on Ni/ZrO2-P are always smaller than those on Ni/ZrO2-C.TEM was applied to directly investigate the morphology and distribution of the CNTs on the used catalysts (Fig. 6 ). The CNTs, attached to Ni nanoparticles at the top, and the onion-like carbon, covering Ni nanoparticles, can be observed on both catalysts. The Ni particles are not completely covered by CNTs. They should be still active for methane decomposition (Li et al., 2011; Monthioux et al., 2007; Li et al., 2009). With the increasing temperature, the amount and the length of the CNTs increase rapidly. The size distribution of CNTs after methane decomposition at 550 °C for 30 min is presented in the right of Fig. 6. The average diameter of CNTs on Ni/ZrO2-P and Ni/ZrO2-C is 21 nm and 40 nm, respectively. It has been proved that the rate of carbon deposition is positive correlation with the size of Ni particles (Seshan et al., 1998). The above results support that smaller Ni particle size of Ni/ZrO2-P contributes to the lower rate of carbon deposition and more uniform distribution of CNTs with smaller diameters.The structure of the carbon deposited on the catalysts was then analyzed by Raman spectroscopy. Fig. 7 shows the Raman spectra of used catalysts after methane decomposition at 450 °C, 500 °C and 550 °C for 30 min. The G band located at 1580 cm−1 is attributed to the in-plane C-C stretching vibration of graphite. The peak at 1350 cm−1 is named as D band, derived from amorphous carbon or imperfect graphite. The relative intensity ratio in the form of ID/IG is used to quantitatively estimate the graphitization degree of deposited carbon (Yu et al., 2017). Namely, when the degree of graphitization increases, the ratio of ID/IG decreases. The ID/IG value of the deposited carbon on Ni/ZrO2-P is 1.52 and 1.50 at 450 °C and 550 °C, respectively. The ID/IG value of the deposited carbon on Ni/ZrO2-C is 1.42 and 1.36 at 450 °C and 550 °C, respectively. Larger ID/IG values of Ni/ZrO2-P indicate lower graphitization degree of deposited carbon, consistent with the results of TEM and XRD.CO2-TPR was used to evaluate the reactivity of the deposited carbon on catalysts. The produced CO was detected by a mass spectrometer. The temperature of the CO signal reflects the reactivity of the carbon with CO2. As shown in Fig. 8 , the peaks of Ni/ZrO2-P are found at 480 °C, 530 °C and 562 °C, while those of Ni/ZrO2-C appear at 491 °C, 554 °C and 578 °C. Therefore, the peaks of Ni/ZrO2-P shift to lower temperatures, suggesting lower graphitization degree of the carbon deposited. According to the DSC results, the peaks at 562 °C and 578 °C for Ni/ZrO2-P and Ni/ZrO2-C should belong to CNTs, and the peaks at lower temperatures should belong to onion-like carbon. The ratio of onion-like carbon to CNTs of Ni/ZrO2-P is 2.8, lower than that of Ni/ZrO2-C (3.9), indicating less onion-like carbon encapsulating Ni active sites formed on Ni/ZrO2-P. This result is consistent with the TG/DSC results, supporting that Ni/ZrO2-P has a higher resistance to deactivation due to onion-like carbon deposition.The present work demonstrates the significant effect of catalyst preparation methodology on the structure and carbon deposition of the Ni/ZrO2 catalyst for methane decomposition. The decomposition of nickel precursor by DBD plasma, followed by the thermal treatment in the absence of the plasma, causes the Ni catalyst of smaller Ni particle size with Ni (111) as the principally exposed facet. This unique catalyst structure favors the formation of the carbon with enhanced reactivity towards oxygen and carbon dioxide. This means the carbon formed on the plasma prepared Ni/ZrO2 catalyst is more easily to be removed, leading to higher reactivity and stability for DRM, as confirmed by our previous study (Hu et al., 2019). This work will be helpful for the future catalyst design beyond DRM. Xue Hu: Methodology, Formal analysis, Investigation, Writing - original draft, Visualization. Xinyu Jia: Visualization, Investigation, Writing - review & editing. Chang-jun Liu: Resources, Investigation, Supervision, Writing - review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the National Natural Science Foundation of China (21536008 and 21621004).Supplementary data to this article can be found online at https://doi.org/10.1016/j.cesx.2021.100104.The following are the Supplementary data to this article: Supplementary data 1

In this work, the structural effect of the plasma decomposed Ni/ZrO2 catalyst and the thermally calcined Ni/ZrO2 catalyst on the formation and the reactivity of the carbon formed from methane decomposition was investigated. The plasma prepared Ni/ZrO2 catalyst possesses smaller Ni particle size with Ni (111) as the principally exposed facet, which favors the formation of the carbon with enhanced reactivity towards carbon dioxide. The Ni (111) facet benefits the formation of carbon nanotubes (CNTs) attached to Ni nanoparticles at the top and suppresses the formation of onion-like carbon which encapsulates Ni active sites. Besides, CNTs on the plasma-treated catalyst show smaller diameter sizes with lower graphitization degree, which are easier to be converted. Therefore, the plasma prepared catalyst presents a higher carbon resistance. The present work well explains the improved activity and stability of the plasma prepared Ni/ZrO2 catalyst for CO2 reforming of methane.

No data was used for the research described in the article.As societal understanding of the adverse implications of global warming is increasing, so is the dependence on renewable energy sources such as solar and wind [1,2]. Water electrolysis provides a means to store this intermittently produced energy in the form of hydrogen, thus facilitating the global transition away from fossil fuels [3,4]. While the hydrogen evolution reaction (HER) is a relatively fast and efficient process, its anodic counterpart, the oxygen evolution reaction (OER), suffers from slow kinetics due to the complex, four-electron transfer pathway [5,6]. Commercial OER catalysts for polymer electrolyte water electrolyzers (PEWEs) are based on non-abundant, thus expensive, Ir and Ru; consequently, research efforts target the development of efficient and stable non-noble metal catalysts for alkaline water electrolyzers (AWEs) [7,8]. Transition metal oxides containing Ni, Co and Fe are promising candidates for alkaline OER catalysis due to their relative abundance, tunable 3d electron configuration, and the versatility of available crystal structures [2,9,10]. In particular, Ni-based materials possess high activity and stability and, unlike their Co-based counterparts, have a comparatively cleaner supply chain with reduced geopolitical risk [11].The rational design of new Ni-based catalysts requires the development of structure–activity relationships in order to correlate the electronic properties, local and long-range structure, and morphology of materials with their catalytic performance. However, it is well established that Ni and Co-based catalysts undergo such significant transformations of surface and sub-surface atoms under OER conditions that structure–activity relationships are difficult to qualify. Indeed, the as-synthesized structure is considered merely a “pre-catalyst,” which undergoes dynamic reconstruction under oxidative conditions to form an amorphous, active surface known as the oxyhydroxide layer [12,13]. Simultaneously, many other reconstruction processes occur under OER conditions. Reversible processes include other potential-dependent phase transformations, the electrochemically-driven dissolution and re-deposition of surface atoms, and the adsorption and desorption of OER intermediates during the catalytic process (accompanied by the associated redox transformations of catalytic centers, and vacancy generation and refilling). Differently, irreversible transformations can include phase transformations and morphological or structural changes [14]. This review will examine the following operando transformations of Ni-based OER catalysts: phase changes involving the formation of a surface layer with a new crystalline structure; oxidation state changes of interfacial cations including Ni and Fe; the extent of lattice oxygen participation in the OER; and the uptake of Fe from an impure electrolyte into the crystal lattice.Ni oxides undergo phase transformations as a function of applied potential, with the reversible formation of an OER active surface layer under oxidative conditions being a key prerequisite for the high activity of these materials [15]. The electrochemical stabilities of Ni metal and its oxide, hydroxide and oxyhydroxide derivatives have been calculated by Huang et al. using standard Gibbs free energies of formation (ΔfG) obtained both experimentally and using DFT, across a range of pH values [16]. The resulting Pourbaix diagrams (Figure 1 a, b), experimentally verified by electrochemical impedance spectroscopy (EIS) and surface-enhanced Raman spectroscopy (SERS), illustrate the formation of new phases at the surface of NiO with applied anodic potential Ni(2+)O → Ni(2+)(OH)2 → Ni(3+)OOH → Ni(4+)O2, and the instability of these phases at low pH. The limitations of such DFT calculations are exposed by the discrepancy in experimentally determined and DFT-obtained phase stability windows and Ni(OH)2 oxidation potentials; though the latter would likely arise from an irreversible phase transformation of the disordered, hydrous α-Ni(OH)2 to the crystalline, anhydrous β-Ni(OH)2 polymorph with electrochemical cycling. In addition, the study proposed probability profiles (Figure 1c) that indicate multiple phases are present at the NiO-NiOOH boundary including Ni3O4, Ni2O3, NiO2, NiO, and Ni(OH)2, and this may contribute to discrepancies in reported experimental oxidation potentials of NiO or Ni(OH)2 [16].These electrochemically-driven phase transformations are associated with significant structural changes. The Ni(2+)(OH)2 “precatalyst,” with the brucite structure (P3m1), exists as two polymorps: β-Ni(OH)2, which consists of Ni2+ and OH- ions in a hexagonal close packed arrangement, and α-Ni(OH)2, which comprises planes of β-Ni(OH)2 with intercalated H2O and electrolyte ions. Under OER conditions, α-Ni(OH)2 and β-Ni(OH)2 experience reversible phase transformations to form their corresponding oxyhydroxides, γ-NiOOH (with a Ni oxidation state of 3.3–3.67+), and β-NiOOH (with a Ni oxidation state of 2.7–3.0+), respectively [17–20]. While the high γ-NiOOH oxidation state range can be attributed to the presence of Ni4+, that of β-NiOOH is less easily understood [20]. Additionally, in concentrated alkaline solutions, the γ-NiOOH phase can form from the irreversible overcharging of β-NiOOH. Attempts to track the formation of these –OH and –OOH species with Raman spectroscopy have produced differing results, suggesting that the exact phase transformations experienced by a material depend on the precursor structure and specific heteroatom doping. For instance, Dürr et al. used operando Raman to identify the direct and irreversible formation of γ-NiOOH from their initial NiMoO4 nano-flower catalyst, without an intermediate hydroxide step [21]. Conversely, with comparable reaction conditions and spectra acquisition time, Saguì et al. used operando Raman to identify an initial, irreversible transformation of their Ta-doped NiO films to α-Ni(OH)2 upon immersion in the alkaline electrolyte, followed by a reversible transformation to γ-NiOOH with applied anodic potential [22]. Likewise for Fe/Co-based materials (hydr)oxide catalysts undergo a similar potential-dependent surface transformation to form the OER active oxyhydroxide phase [23,24]. For example, D. Grumelli et al. used operando X-ray diffraction to observe the surface reconstruction of Fe3O4 in OER conditions [24]. For the highly active spinel Co3O4, this process is also dependent on the Co-ion geometry: only the tetrahedral Co2+ is capable of releasing electrons under applied potential to form the surface CoOOH layer [25].Similar to the phase transformations of Ni(OH)2 polymorphs to the corresponding NiOOH structures, other Ni-based structures such as perovskite or spinel oxides experience an electrochemically-driven surface reconstruction associated with the dynamic formation of an oxyhydroxide surface layer [26–28]. Baeumer et al. used operando UV-vis spectroelectrochemistry to investigate the behavior of Ni-terminated LaNiO3 (LNO) films, and discovered the formation of a 4 Å-thick NiOOH surface layer at the Ni2+/Ni3+ redox peak potential during the anodic CV sweep [28]. Post-mortem low-energy electron diffraction (LEED) revealed that this surface reconstruction is associated with the disappearance of the perovskite diffraction pattern. While the authors hypothesized an irreversible loss of long-range order in the material, it is important to note that LEED probes approximately a < 1 nm depth. Conversely, Liu et al. used scanning transmission electron microscopy (STEM), probing the entire sample, to identify the localized amorphization of only two LNO surface layers after OER catalysis [29]. While the structural amorphization of LNO is a fully irreversible process, the formation of the –OOH layer is theoretically reversible with applied potential, according to the thermodynamic stability windows outlined in Pourbaix diagrams calculated by Huang and Zhou [16,30]. However, Fabbri et al. used operando X-ray absorption spectroscopy (XAS) to correlate the oxyhydroxide layer formation on the perovskite Ba0.5Sr0.5Co0.8Fe0.2O3-δ (BSCF) with an irreversible oxidation of Co atoms [12]. This implies that surface –OOH adsorption is in fact not fully reversible, though an analogous claim has not yet been proven for Ni perovskites.Under OER conditions, interfacial cations including Ni and Fe (where present) will undergo reversible changes in oxidation state. For nickel oxide, hydroxide and perovskite catalysts, recent evidence has pointed to the reversible oxidation of Ni centers Ni(2+) → Ni(3+) → Ni(4+) during OER catalysis [12,28,31–33]. The formation of Ni(4+)OO under OER conditions was first identified by Diaz-Morales et al. using in situ SERS combined with an 18O-labeled electrolyte [34]. Due to its high oxidation state, the 3d orbitals in Ni(IV) are lowered in energy to allow optimal overlap with the O 2p orbitals. This, in turn, energetically favors OH- adsorption onto the material surface [27,35]. Similar transformations occur in Co-based oxides, as evidenced by Hu et al., who used in situ XAS to identify the formation of surface Co4+ sites on CoOOH as a function of applied potential [36]. Their findings were echoed by operando XAS studies conducted by Zhang et al., who identified the transformation of CoOOH to Co3+/4+OOH1-x under oxidative conditions via a potential dependent deprotonation reaction [37]. Ni-based materials are often doped with Fe to enhance the OER activity, though the reversible electronic transformations of Fe under OER conditions are still widely debated [38]. For instance, XAS studies conducted by Friebel et al. and Görlin et al. determined Fe3+ to be the maximum oxidation state of Fe under OER conditions, whereas Hunter et al. used Mössbauer and UV-vis spectroscopy to detect Fe4+ states stabilized as ∗Fe4+=O, a finding supported by DFT calculations performed by Martirez et al. [39–43]. Despite the discrepancies in reported electronic structure, it is evident that synergistic electronic interactions between Ni and Fe result in reduced overpotentials for the OER on Fe-doped materials. The bridging oxygen (μ–O) in Ni-O-Fe bonds can facilitate partial electron transfer as π-donation between the Ni/Fe d-orbitals, and the Ni-Fe synergy might enable a bimetallic mechanism to proceed through bridging O2 intermediates Ni-∗O-O∗-Fe [44]. Aside from electronic effects, Abbott et al. used operando XAS to observe that the Fe-doping of nickel oxides increases the structural stability of the β-Ni(OH)2 phase [45]. The transformation of β-Ni(OH)2 to a layered β-NiOOH structure (and disordered α/γ-phases) facilitates ionic diffusion to previously inaccessible metal centers, increasing the electrochemically active surface area. However, since Fe-doping enhances the intrinsic activity of the available active sites in β-Ni(OH)2, this transformation is no longer a prerequisite of high activity.Wang et al. studied the effect of Fe doping of α-Ni(OH)2 in an anion exchange membrane electrolyzer, and reported a current of 2 A cm−2 at 2.046 V and 50 °C, a performance on par with proton exchange membrane alternatives [46]. Initial rotating disk electrode (RDE) studies revealed a positive shift in the Ni2+/Ni3+ redox peak potential attributed to the Ni to Fe charge transfer, and the trigonal distortion of the octahedral symmetry that arises from Ni-O bond contraction as a consequence of Fe doping [47,48]. A similar distortion was also observed for Fe-doped LaNiO3 films by Bak et al., who demonstrated the oxygen-octahedron distortion results in a significant increase of the DOS of both the O 2p and Ni/Fe 3d orbitals near the Fermi level, facilitating the charge transfer from transition metals to adsorbates via oxygen (Ni3+-O(OH∗) → Ni4+-OO∗) [49]. Wang et al. used DFT + U calculations to model the OER mechanism on the (001) facet of γ-Fe0.25Ni0.75OOH (Figure 2 a), and created a free energy diagram from the reaction energies of each elementary step, as shown in Figure 2b. The most favorable surface configuration was achieved with a local arrangement of one Fe3+ and two Ni4+ atoms, resulting in a low overpotential of 0.57 V (Figure 2c). However, they provided no experimental evidence that such a high local concentration of strongly oxidized Ni4+ could be achieved, and it appears that they did not consider the possibility of Fe4+. Besides, it is vital to consider that irreversible, OER-driven processes such as cation dissolution or surface amorphization will likely evolve the local surface structure when evaluating DFT facet calculations.There are two main classes of OER mechanism: the conventional adsorbate evolution mechanism (AEM), in which all oxygen-containing intermediates originate from the electrolyte, and the lattice-oxygen mediated mechanism (LOM), in which the lattice oxygen participates in the reaction [50]. The latter is associated with an increased OER reactivity for Ni-based perovskites, oxides and hydroxides; thus, developing surface-sensitive spectroscopic techniques that can directly detect oxygen vacancies in situ is paramount for optimizing future catalyst design [51–54]. Ex situ neutron diffraction has provided indirect evidence for the oxygen vacancy content of OER catalysts, though it has crucial limitations: highly crystalline samples are required, surface changes in vacancy concentration are difficult to detect with bulk sensitivity, and a large amount of material is necessary, which precludes the possibility of operando studies [55,56]. Soft X-ray absorption spectroscopy (sXAS) at the oxygen K-edge likewise indirectly monitors oxygen vacancy dynamics; in combination with total electron yield (TEY) acquisition, a high surface sensitivity (up to 5 nm) is possible [57,58]. Thus far, the design of an operando cell with negligible electrolyte interference in the oxygen K-edge absorption spectra limits definitive sXAS data interpretation. Recently, Mom et al. used sXAS to study IrOx thin films of 100 nm thickness while eliminating electrolyte oxygen contribution, through a back-contacted electrolyte/IrOx interface [59]. Nevertheless, the use of TEY detection at the ‘front’ negates the concept of surface sensitivity, and illustrates the multifaceted problem of optimal cell design.As a result, conclusions about the extent of oxygen vacancy participation in Ni-based materials must be examined with caution based on the experimental methods employed. For instance, Zhang et al. used sXAS of the oxygen K-edge to determine that oxygen vacancy participation in FeCoCrNi thin films can be promoted by the reversible formation of Ni4+ [57]. The generation of Ni4+ under OER conditions results in downshifted Ni 3d orbitals (as predicted by partial density of states (PDOS) calculations); this induces the formation of oxygen ligands with localized holes in their p-orbitals (i.e. O(2−δ)-) [57]. As outlined by Nong et al., the enhanced electrophilic character of these oxygen ligands increases their reactivity towards nucleophilic acid–base-type O-O bond formation (i.e. nucleophilic attack of electron-deficient O(2−δ)- ligands), facilitating the LOM and improving OER activity [60]. However, obtaining these ex situ sXAS measurements involved freeze-quenching the post-mortem samples in liquid N2, before transferring them in air to the vacuum chamber for measurement; the impact of this preparation process on the surface oxygen is unknown. Moreover, systematic studies performed by Cheng et al. concluded that varying the oxygen vacancy content in Ni and Co-based perovskites was accompanied by changes in other physiochemical properties including conductivity, degree of structural (dis)order and cation oxidation state; the predominant mechanism is determined by the combination of these and other factors [55,61,62].Ni-based electrocatalysts uptake trace Fe impurities from unpurified KOH electrolyte, resulting in significantly enhanced OER activity and cycling stability [63,64]. Therefore, it is vital to understand the mechanism of this Fe incorporation and its role in improving OER kinetics in order to decouple and optimize the intrinsic activity of Ni catalysts. Kuai et al. used operando XAS to determine that electrolyte Fe incorporation into 2D Ni(OH)2 nanosheets is an electrochemically driven process, occurring at the OER reactive potential during the anodic CV sweep [65]. Furthermore, by using X-ray fluorescence microscopy (XFM) to generate elemental distribution maps, they determined that Fe incorporation occurs predominantly at edge sites, which feature a higher concentration of oxygen vacancies and show higher OER reactivity [65]. The group identified a non-linear increase in OER current with Fe atomic ratio, and attributed this to the irreversible formation of a separate, insulating FeOOH phase at high overpotentials, although other disruptive electronic or structural effects may also influence OER activity. Furthermore, operando soft XAS suggested that Fe surface uptake enhances the reducibility of Ni, increasing the concentration of oxygen vacancies and improving the OER activity [65].The surface Fe sites exist in a dynamic equilibrium, undergoing reversible dissolution and re-incorporation in aqueous KOH [63]. Chung et al. investigated the effect of electrolyte Fe concentration on the OER activity of MOxHy (M = Ni, Fe, Co), and determined that Fe adsorption saturates at electrolyte concentrations as low as 0.1 ppm (Figure 3 a, b) and the OER activity increases linearly with Fe surface coverage (Figure 3c). Importantly, maintaining the high OER activity was conditional on achieving “dynamically stable” surface Fe with continued re-deposition after dissolution (Figure 3e) [63]. Farhat et al. reported that Ni(Fe)OxHy subsequently cycled in purified (Fe-free) KOH will experience a loss of OER activity as the active, surface Fe atoms move into inactive bulk sites, though they provide only electrochemical evidence to support their claims [66]. The authors hypothesize that Ni atoms at the surface are now able to dissolve and redeposit in the structure preferentially with respect to Fe in so-called “Fe-free” KOH, as the purification process involves dissolving nickel nitrate in the electrolyte [66,67]. However, Chung et al. used ICP-MS combined with stationary probe rotating disk electrode studies (SPRDE) to prove that Ni sites have a comparatively high stability with respect to Fe in (Ni/Fe)OxHy, and Ni dissolution is negligible [63].In summary, Ni-based OER catalysts experience significant operando transformations, often leading to a highly active final state. Reversible changes include phase transformations, oxygen vacancy dynamics, intermediate adsorption/desorption, and surface atom dissolution/redeposition. Irreversible changes comprise structural and morphological transformations. This begets the obvious question for further research: is the initial ‘precatalyst’ structure a prerequisite of the high activity final state, or should the future development of Ni-based materials focus on direct synthesis of this final state? Accordingly, gaining a greater fundamental understanding of the OER mechanism is a broad but pressing concern. In particular, this should involve developing operando methods and innovative cell designs to enable direct observation of oxygen vacancy dynamics (with bulk and surface sensitivity, a high time resolution, and minimal interference from atmospheric, electrolytic, or binder oxygen). Alongside the expansion of fundamental mechanistic studies, further exploration of new classes of materials with less well-known structure-activity correlations would advance the field in new directions. In particular, Ni-based metal-organic framework (MOF) catalysts, noble metal-free high entropy alloys, and amorphous Ni-based materials provide underexplored yet promising avenues for research [68–71]. In addition, while it is worthwhile to decouple the intrinsic activity of Ni-based catalysts from changes induced by operando uptake of Fe from unpurified electrolyte, shifting the focus of future research to optimize Ni-based catalysts for operation in Fe-doped electrolyte may be a valuable, application-oriented approach. Finally, approaching DFT calculations with careful consideration of the reversible and irreversible changes of Ni-based catalysts during operation will enable this valuable computational method to direct and enhance experimental work, rather than acting as an addendum.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors gratefully acknowledge the Swiss National Science Foundation through its SNSF PRIMA grant (grant no. PR00P2_193111).

Nickel-based catalysts for the alkaline oxygen evolution reaction (OER) demonstrate excellent catalytic performance and stability. However, a lack of fundamental understanding of the dynamic electronic and structural changes that occur under OER conditions inhibits the rational design of new materials. Recent advances in operando spectroscopy and computational modeling techniques have helped to elucidate the electrochemically-driven transformations of Ni-based materials. For reversible transformations, this encompasses an increased understanding of the redox transformations of Ni/Fe centers, the adsorption and desorption of reaction intermediates, oxygen vacancy dynamics, phase transformations, and the mechanism of dissolution and redeposition of surface atoms. Likewise, there have been great advances in scientific understanding of irreversible transformations including phase transformations related to ageing, as well as operando surface reconstruction which involves the growth of new OER active phases.

Data will be made available on request.The global climate is facing two main challenges: first, the increased energy demand due to the growing world population, and second, growing anthropogenic CO2 emissions due to the use of non-renewable fossil fuels in major economic sectors. Ambitious targets like limiting the rise of global temperature to stay below 2 °C in the Paris Agreement and achieving net-zero CO2 emissions by 2050 in the European Green Deal will drive the current energy transition towards a low or neutral renewable carbon energy system. In line with this, the United Nations (UN) Sustainability Development Goals (SDGs) adopted by all UN members also provided a blueprint to tackle detrimental climate change and achieve a better sustainable society for all. Thus, the use of renewable energy in tackling climate change is inevitable. Renewable liquid biofuels fall under this category and provide an immediate solution for sectors like transportation.Biomass feedstock can be divided into three main categories: (1) sugar-based feedstocks such as sugar beet, sugar cane, and corn, (2) triglyceride feedstocks like animal fats, vegetable oil, and waste cooking oil, and (3) lignocellulosic feedstocks like wood and forestry residues, bagasse, grass, and leaves. The types of biofuels depend largely on the biomass source. Renewable liquid fuels, also known as advanced biofuels, produced from the non-crop and waste-based bio-feedstocks, represent an excellent option as an alternative fuel and also serve the role of bridging the transition period for existing conventional combustion engine-based fleets. For instance, hydroprocessed esters and fatty acids (HEFA) as hydrotreated vegetable oils (HVO), are the only drop-in biofuels that are commercially produced in refineries. Several examples of such commercial technologies are NEXBTL™, Ecofining™, Vegan™, and Hydroflex which produce these advanced biofuels. Apart from these commercially available examples, the next-generation biofuels such as those derived from pyrolysis oil also possess advantages in reducing greenhouse gas (GHG) emissions and fossil fuel dependency. Pyrolysis oil can be produced using different processes, one of which is fast pyrolysis or thermal liquefaction of biomass feedstocks [1,2]. The conversion of solid biomass via a thermochemical process like fast pyrolysis results in bio-oils that can be subsequently upgraded via catalytic hydrotreating into biofuels and high-value platform chemicals. Another potential advanced feedstock like lignin can also be used to substitute fossil-based feeds. Lignin is a biopolymer consisting of phenylpropane units (coniferyl, sinapyl, and p-coumaryl alcohol) [3]. It is an important renewable carbon source and accounts for 20–30% of the major mass of lignocellulosic biomass. Due to the large utilization of cellulosic and hemicellulosic materials in the existing biorefineries, the remaining lignin fraction is considered a byproduct and is often burnt to produce heat and power for the mill. Thus, lignin can serve as a sustainable feed for liquid fuel production or value-added fine and platform chemicals.However, there are a few common undesired properties of the bio-feedstocks such as high oxygen content depending on the biomass ( Table 1) and acidic nature caused by the presence of carboxylic acids. The high oxygen content contributes to detrimental properties of bio-oils like high viscosity and low heating value as compared to fossil-derived fuels [4]. Owing to the various negative characteristics of the bio-oils from these feedstocks, it is difficult to use these bio-liquids directly as engine fuel. Therefore, a refining process is required to improve the quality of the products so that the produced liquid fuel is compatible with the existing fuel grades. This process involves conventional hydrotreating technology such as hydrodesulfurization (HDS), hydrodenitrogenation (HDN), hydrodeoxygenation (HDO), and hydrodemetallization (HDM) processes. These processes serve to remove or reduce the sulfurous, nitrogenous compounds, oxygenates, and metals from fossil feeds. Catalytic hydrodeoxygenation (HDO) has been implemented in the refineries to remove excess oxygen in the form of H2O, CO, and CO2 at various temperatures and pressure with hydrogen as a co-reactant. Moreover, the reaction is catalyzed by a selective hydrotreating catalyst. The key elements in such a process are the choice of catalyst material, reaction conditions, type of reactors, and feedstocks that are upgraded. Over the past few decades, transition metal sulfides/noble metals and non-sulfided catalysts have been studied extensively for valorizing bio-based feedstocks (Triglycerides/Fatty acids/pyrolysis oil/ lignin-derived bio-oil, etc.). In this review, we focus on the application of metal sulfides as catalysts for advanced biofuel production.Krauch and Pier discovered the transition metal sulfides (TMS) at the former Badische Anilin und Sodafabrik (BASF) in 1924 [8]. Their early findings showed that MoS2 and WS2 were effective hydrogenation catalysts, and this led to the future development of hydrotreating catalysts. The traditional industrial metal sulfides are Ni or Co-promoted Mo/W disulfides. The Ni or Co promotion in a fully sulfided catalyst gives the so-called active Ni/CoMoS phase [9]. It is postulated that the promotion weakens the Mo-S bonds via d-electron donation by Ni/Co producing a sulfur vacancy or so-called co-ordinately unsaturated sites (CUS), indicated by the green dotted oval in Fig. 1 [4,10]. In terms of deoxygenation, the electrophilicity of Mo thus attracts oxygen-bearing molecules [10,11]. On the other hand, the presence of H2 generates metal hydrides and sulfhydryl groups that play additional roles [12]. In other words, the catalytic activity of metal sulfides is thus governed by the type and composition of sites available to the substrate molecule which can be engineered by parameters such as promoters, support, additives, and activating conditions. The morphologies of the evolved catalyst also play a dominant role in determining the activity of the catalysts [13]. These Ni or Co-promoted MoS2 catalysts are depleted of sulfur during oxygen removal from bio-oils and require constant addition of sulfur sources like DMDS or H2S to maintain them in their active sulfide state [14]. Due to this fact, these sulfided catalyst systems are criticized in literature since they nullify the advantage of bio-oils which inherently contain no or low sulfur [15]. However, these sulfided catalysts have excellent hydrotreating selectivity and are low cost which makes them viable for commercial refinery operations. Metal sulfide catalysts will continue to have a significant role in the refining industries owing to the versatility of this type of catalytic material and also the transition toward cleaner and sustainable fuel production. This can be evident from the scientific publications regarding HDO which have increased by 10-fold since 2010 because of the escalating need for alternative sources of energy and also the increasing application of sulfided catalysts in deoxygenation applications ( Fig. 2). Existing review articles related to the application of sulfided catalysts focused on the understanding of the fundamental principles of the materials [16], and their applications in electrocatalysis [17,18], photocatalysis [18], and as supercapacitors [19].Although there are various reviews in the last decade focused on the upgrading of renewable feedstocks in the form of model compounds and/or real feedstocks over various types of catalysts (see Table 2); a comprehensive review solely dedicated to sulfided catalysts is still lacking. Given the significance of hydrotreating catalysts in the emerging field of advanced biofuel production, a comprehensive review of the use of sulfided catalysts for biomass conversion is needed. In this work, we have reviewed the catalytic upgrading of different important bio-feedstocks such as triglycerides, monomeric and dimeric phenolic compounds that are present in pyrolysis oil, biomass-derived pyrolysis oil, woody feedstocks, and waste lignin using metal sulfide catalysts. The reaction routes of the biomass-derived feed during the hydrotreatment have been emphasized. Furthermore, kinetics studies of upgrading the various feedstocks using sulfided catalysts and also the deactivation of catalysts are highlighted. Insights of the deoxygenation reaction and deactivation mode over metal sulfides from theoretical studies and computational approaches are also presented. Finally, the challenges and future possible research related to the valorization of different bio-feedstocks into liquid fuels employing the sulfided catalysts, and hurdles utilizing bio-feeds in the industry are also discussed in the current work.Out of all advanced biofuel technologies like – Hydrotreated Vegetable Oil (HVO), fast pyrolysis, catalytic pyrolysis, hydro pyrolysis, lignin depolymerization, and hydrothermal liquefaction (HTL), it is only HVO process that has been commercialized so far. There are already several companies like Neste, Preem, Diamond diesel, REG producing HVO biofuels at a commercial scale. The only other biofuel process which is close to commercialization is fast pyrolysis bio-oil (FPBO), where commercial fluid catalytic cracking (FCC) trials of FPBO are underway [45].HEFA/HVO (Hydroprocessed Esters and Fatty Acids) fuels are compatible with fossil-based diesel fuels due to their chemical resemblance. As a result, existing refinery facilities, and infrastructure can be used for production, transportation, and distribution for further use. Fig. 3 shows a parametric comparison between ultra-low sulfur diesel and HVO/HEFA fuels. Thus, standalone, or blended forms of HVO/HEFA fuels contribute toward a significant reduction of GHG and particulate matter emissions. As a result, tremendous research focus has been devoted to understanding the mechanistic insights in the core upgrading process, HDO. Such renewables can be obtained from variable feedstocks like animal fat, waste cooking/vegetable oil, and tall oil from wood and forest biorefinering. It is expected that refineries will face scarcity of these feedstocks in the next few years as the demand for sustainable fuels from the heavy transport and aviation sectors accelerates. Thus, more research and technology developments are needed in pretreatment methods for low-value, ubiquitous, and high-contamination feedstocks such that they can be hydroprocessed in the refinery.However, the turnover time of a hydrotreater is much shorter when renewable oils like used cooking oil, waste animal fat, tall oil, etc. are processed compared to a hydrotreater processing fossil feeds like vacuum gas oil (VGO), etc. There are three main reasons for the shorter lifetime of catalysts in hydrotreater processing HVO feedstocks. Firstly, since the oxygen content of renewable oils is higher than sulfur in VGO, larger quantities of hydrogen are required to remove oxygen as H2O. The second reason is that hydrodeoxygenation reactions are more exothermic compared to hydrodesulfurization reactions as can be observed from the differences in the enthalpy of formation of H2O(g) and H2S(g) respectively (−242 kJ/mol vs −21 kJ/mol) [47]. Last but one of the most important reason which has not received enough attention in research studies is the contaminants present in renewable feeds like iron, phosphorus, alkali metals, etc. [48–50]. So these three factors – depletion of hydrogen, high temperature, and contamination, combined result in accelerated catalyst deactivation and pressure build-up [51]. In this section, we will focus on the studies that have used sulfided catalysts since they are the most industrially relevant catalysts. The literature studies can be categorized in two segments based on the feedstocks studied – Model compounds which include - Fatty acids (FAs), Fatty acids alkyl esters (FAAEs), Triglycerides (TGs), and commercial feedstocks like UCO (used cooking oils), tall oil, etc. It should be noted that the HDO reaction mechanism for fatty acids, fatty acid alkyl esters, triglycerides is quite similar. Renewable oils like UCO, tall oil, etc. primarily contain free fatty acids and triglycerides so their reaction chemistry is similar as well. Typically, fatty acid alkyl esters and triglycerides undergo hydrolysis to produce fatty acids as the common intermediate in the overall reaction scheme. Deoxygenation of fatty acids over sulfided catalysts occurs in the following three ways [48,52]: a) A so-called direct-HDO in which oxygen is removed as a water (H2O) molecule b) Decarbonylation (DCO) in which oxygen is removed as carbon monoxide (CO) c) Decarboxylation (DCO2) in which oxygen is removed as carbon dioxide (CO2) A so-called direct-HDO in which oxygen is removed as a water (H2O) moleculeDecarbonylation (DCO) in which oxygen is removed as carbon monoxide (CO)Decarboxylation (DCO2) in which oxygen is removed as carbon dioxide (CO2)In the direct-HDO route, there is no loss of carbon as oxygen is removed in the form of a H2O molecule while in the two latter routes, oxygen is removed in the form of CO or CO2 such that the hydrocarbon product is formed with one less carbon.The term decarbonation (DCOx) will be used to refer to decarbonylation and decarboxylation together, otherwise, they will be separately specified in the following sections of this review. The hydrodeoxygenation or “HDO” is a broader term to define the removal of oxygen irrespective of the three routes. However, “direct-HDO” is referred to when deoxygenation occurs while producing water as the side product [48,52].The catalytic cycle for hydrodeoxygenation of fatty acid molecules (here stearic acid) over sulfided molybdenum catalysts is presented in Fig. 4 [53]. The following steps of this catalytic cycle, initiated with the creation of a sulfur vacancy have been also reported for phenol-like molecules, so this is also relevant for later sections discussing the phenolic models. Sulfur linked to Mo reacts with hydrogen to produce H2S and a “sulfur vacancy” is created on the MoS2 structure. It is postulated that there is always a dynamic equilibrium of such sulfur vacancies depending on H2/H2S partial pressure. Then a heterolytic dissociation of the hydrogen molecule occurs such that one hydrogen atom binds to sulfur to form a sulfhydryl (-SH) group while the other hydrogen atom forms a metal hydride bond with molybdenum. The carbonyl moiety of the fatty acid molecule binds at the sulfur vacancy. Then a proton from the acidic SH group attacks the hydroxy group of the stearic acid as an example. A water molecule is removed, and the charge is transferred to the neighboring carbon. This cation species extracts a hydride from the next Mo atom. In the final step of this catalytic cycle, a hydrogen molecule reacts to yield metal hydride (Mo-H) and sulfhydryl (-SH) species and results in the conversion of stearic acid to octadecanal. The patent literature has reported the use of Ni or Co-promoted molybdenum sulfided catalysts for the deoxygenation of fatty acid-based feedstocks for more than three decades now [54].Several studies have been explored for HDO of fatty acid to green diesel employing supported and unsupported sulfided catalysts. The sulfidation temperature influences the formation of different Mo species (Mo4+, Mo5+, Mo6+) catalyzing deoxygenation of palmitic acid to straight-chain hydrocarbons to varying degrees over NiMo/Al2O3-TiO2 catalyst [55]. The role of unsupported MoS2 and Ni-promoted MoS2 has been studied using hexadecenoic acid by Wagenhofer et al. [56]. It is concluded that fatty acid deoxygenation over MoS2 is primarily followed via C-O hydrogenolysis to an aldehyde, hydrogenation to a primary alcohol, dehydration, and hydrogenation to corresponding alkanes (Cn pathway). On the other hand, deoxygenation over unsupported Ni promoted MoS2 mainly proceeds through a ketene intermediate, and decarbonylation via the scission of its C-C bond to yield saturated hydrocarbons (Cn-1 pathway) via alkene intermediates. However, keto-enol tautomerism of intermediate aldehyde has also been demonstrated for the HDO of fatty acid over supported sulfided NiMo/Al2O3 catalysts [52,57]. Ni or Co promotion to base MoS2 typically promotes the decarbonation pathway [53]. Surfactant-modified and magnetically reusable unpromoted and Ni(Co) promoted MoS2 over greigite (G) have been evaluated for stearic acid HDO and the deoxygenation activity was ranked in the order of NiMo/G > CoMo/G > Mo/G [58]. Interestingly, an inhibiting effect of fatty acid which preferentially occupies the active site and hinders access for the intermediate aldehydes/alcohols has also been explained via experimental and DFT studies [10,59]. Hydrogen donor solvents and acidity imparted by BEA zeolite were found to enhance the conversion/deoxygenation of stearic acid at low pressure (0.8 MPa H2 and 350 °C) and give a high yield of alkanes (C17 being the major one) for NiMo supported over mixed oxide (γ-Al2O3-BEA-zeolite) [60].Alkyl esters like – fatty acid methyl esters (FAMEs) and fatty acid ethyl esters (FAEEs) have different decomposition pathways. As shown in Fig. 5, β-elimination occurs only in FAEEs (or esters with alkyl groups >C1) to produce free fatty acids [61]. The FAEEs are expected to have a similar reaction mechanism like TGs via β-elimination. This reaction is not possible for FAME molecules due to the absence of the β hydrogen.Another route through which such alkyl esters can produce fatty acids is hydrolysis. There is not a significant difference in the reaction mechanism for the hydrolysis of methyl and ethyl esters with a similar yield and product distribution [62,63]. Hydrolysis requires the presence of moisture (H+/H2O) and the Lewis acid sites (e.g., alumina as support). HDO reactions produce water so hydrolysis of alkyl esters is quite possible in the catalyst bed. Another possible route can be the direct deoxygenation of such alkyl esters.Laurent and Delmon tested sulfided CoMo and NiMo catalysts for hydrodeoxygenation of model compounds containing ester groups (Diethyl sebacate) [64]. During the reaction, a group of products has been identified demonstrating that the major pathways for deoxygenation are hydrogenation and decarboxylation. De-esterification to carboxylic acid occurs to a limited degree. The activation energy for the hydrogenation reaction was found lower for NiMo than CoMo catalysts while no appreciable differences in the decarboxylation activation energy were discerned although a higher decarboxylation degree was observed for NiMo catalysts.Krause and coworkers [65,66] explored HDO activities of methyl esters (methyl heptanoate and methyl hexanoate) in flow and batch reactors over sulfided NiMo/γ-Al2O3 and CoMo/γ-Al2O3. Methyl ester conversion was found higher over NiMo/γ-Al2O3 and the formation of the corresponding deoxygenated products (n-heptane/heptenes and n-hexane/hexenes etc.) required more hydrogen than that with CoMo/γ-Al2O3. Analysis of the reaction products revealed that primary alcohols are produced from the methyl ester via hydrogenolysis of the C-O, σ-bond of the carboxyl group which upon dehydration yields alkenes, and further hydrogenation forms the n-alkanes. A second path proposed is the de-esterification to carboxylic acid and methanol which takes water from the alcohol dehydration reaction. The carboxylic acid can either further be reduced to alcohol under the reaction conditions or can be decarboxylated to alkenes. A third path can be direct decarboxylation of esters to one carbon atom with fewer product hexenes/pentenes with additional methane/carbon dioxide as shown in Fig. 6.Coumans et al. [52,67] investigated HDO of methyl oleate using sulfided NiMo over a few supports (γ-Al2O3, activated carbon, SiO2, and SiO2-Al2O3) in a fixed bed reactor under trickle flow conditions at 260 °C, 30 or 60 bar and WHSV of 6.5 h−1. Hydrogenation of the double bond in methyl oleate produces predominantly methyl stearate during the early stage (∼10 min) of the reaction. Besides, oleic acid and stearic acid are also observed in the reaction mixture. Initial hydrolysis of methyl oleate to oleic acid was found higher over NiMo/Al2O3, and NiMo/Al2O3-SiO2 catalysts than others due to the presence of surface Al3+ which acts as Lewis sites as mentioned earlier. Blockage of such sites by carbonaceous deposits deactivates Al-containing catalysts while others show stable deoxygenation activity. High hydrolysis, high stability (168 h on stream), and selectivity to C18 hydrocarbons over NiMo/C were attributed to the activity of evolved metal sulfides and/or to surface acidic moieties.Triglycerides have the following three main decomposition pathways: Route 1: β-elimination; Route 2: γ-hydrogen migration; and Route 3: Direct deoxygenation (DO), as shown in Fig. 5. In Route 1, the removal of hydrogen at the β position and then hydrogenation occur alternatively in a stepwise manner from triglyceride to diglyceride to monoglyceride to glycerol, with the potential to release three free fatty acid molecules per triglyceride molecule. Also, most of the studies report the evolution of fatty acids during the deoxygenation of triglycerides [61]. However, the subsequent β-elimination of di-fatty acid ester is not possible without hydrogen and active sites. γ-hydrogen migration (Route 2) is reported to occur only at a higher temperature of 450 °C, so it is not likely at the typical HDO reaction conditions. Route 3 of direct deoxygenation (DO) occurs only in the presence of a highly active NiMo catalyst [61]. So it could be concluded that the triglycerides-based renewable feeds can yield fatty acids only through Route 1. Indeed, facile hydrolysis of a triglyceride molecule (triolein, glyceryl tioleate) to fatty acid intermediates has been demonstrated over sulfided NiMo/Al2O3 through route 1 [52].As mentioned earlier, oxygen from feedstocks like triglycerides, and fatty acids/alcohols are eliminated via water and carbon oxides as deoxygenation proceeds and yields hydrocarbons as mainly n-alkanes in the diesel range with high cetane values. Table 3 presents the state-of-the-art of sulfided catalysts for HDO of triglyceride-based feedstocks like waste vegetable/cooking oil and tall oil that have been studied exclusively via the catalytic HDO process. Kubička and co-workers investigated the deoxygenation of rapeseed oil over sulfided Ni, Mo, and NiMo over Al2O3 in a fixed bed flow reactor [68–71]. Bimetallic NiMo yields a higher amount of hydrocarbon than the monometallic catalysts for a given conversion. Ni and Mo-containing catalysts promote decarboxylation and direct-HDO respectively. Fatty acids are the only intermediates over Ni/Al2O3, thus no fatty esters form, while over Mo and NiMo containing catalysts fatty alcohol and fatty ester formation proceeds due to the rapid disappearance of fatty acids. Esterification of fatty alcohols and fatty acids was observed higher over Mo/Al2O3 [70]. The authors also detected that the hydrocarbon phase obtained (over 310 °C) was mostly composed of n-alkanes, n-C18, n-C17, and i-alkanes of varying amounts (based on the reaction conditions and catalyst). Such an organic liquid product is compatible with mineral diesel, thus meeting or exceeding the required quality. However, it suffered from poor low-temperature properties necessitating further processing. Furthermore, NiMo sulfides over SiO2, TiO2, and Al2O3 have been evaluated to elucidate the interaction of the active phase and support [71]. It was found that NiMoS over SiO2 enhances hydrogenolysis of triglycerides to fatty acids at low deoxygenation degree while over TiO2 fatty ester formation increases. As the deoxygenation progresses n-C17 yield increases over NiMo/SiO2 demonstrating its preference for the decarbonation route, while the other two showed HDO preferences. Observed reactivity was thus ascribed to the differences in the support properties despite the active phase dispersion variation in the order of SiO2 > Al2O3 > TiO2.M. Toba et al. [72] studied different grades of waste oils which can be converted to paraffinic hydrocarbons over NiMo/γ-Al2O3, CoMo/γ-Al2O3, and NiW/γ-Al2O3. Modification of the alumina support by B2O3 promoted the formation of i-paraffins over the bimetallic catalysts. NiMo/B2O3–Al2O3 and NiW/Al2O3 showed high HDO and hydrogenation activity for a longer period (∼80 h) while hydrogenation activity of CoMo/B2O3–Al2O3 decreases (approximately after 10 h on stream) resulting in high olefin formation. Compared to Mo-based catalysts, tungsten-based catalysts accelerated deoxygenation by decarboxylation/decarbonylation. H. Wang et al. [73] also investigated HDO of waste cooking oil over supported CoMoS, elucidating the deoxygenation activity, deactivation, and regeneration of the catalyst. Crude tall oil, distilled tall oil, and tall oil fatty acid (TOFA) were hydrotreated employing a commercial NiMo/γ-Al2O3 in a trickle-bed reactor at 5 MPa and 300–450 °C. Hydrocarbon fractions with 45 wt% paraffins were obtained at the most favorable tested conditions for crude tall oil. TOFA hydrotreating yielded more than 80% n-alkanes where the decarboxylation route was dominant over the direct-HDO route at high temperatures greater than 400 °C [74,75]. Soybean oil has been studied using sulfided NiMo/γ-Al2O3, and CoMo/γ-Al2O3 [76]. HDO yielded higher amounts of straight-chain alkanes (66%) over the NiMo catalyst than the CoMo (43%) catalyst due to isomerization and cracking enhancement in the latter. NiMoS over Mn-modified Al2O3 was reported to enhance triglyceride conversion and subsequent deoxygenation during the HDO of waste soybean oil [77]. Refined cottonseed oil has been hydrotreated with desulphurized petroleum diesel under refinery conditions and it was found that such treatment increases the cetane number of the final product [78]. Kubička et al. also investigated CoMo sulfides over mesoporous MCM-41 for hydrotreatment of refined rapeseed oil and the observed deoxygenation activity is lower than that for CoMo/Al2O3 [79]. Al incorporated MCM-41 showed better deoxygenation activity towards hydrocarbon formation. Withdrawal of Al from the MCM-41 framework favored the formation of fatty esters instead of fatty acids with MCM-41. Jatropha oil has been hydrotreated using sulfided NiMo over acidic SAPO-11 and Al2O3 support. It was claimed that the higher amount of total and strong acidic sites in SAPO-11 affects the formation of different active phases of NiMo and in combination they promote decarbonation, hydrocracking and isomerization reactions [80]. Lower acid sites with alumina supported sulfided NiMo on the other hand shows high selectivity to diesel range (C15–18) hydrocarbon fractions.Depolymerization and deconstruction of the complex structure of biomass can be performed via various processes for example enzymatically [81], thermally using fast pyrolysis [82], and catalytically. In all these processes, the operating conditions largely affect the composition and yield of final products. For example, hydrothermal liquefaction (HTL), is a technology where bio-oils are produced using water as a medium under supercritical or subcritical conditions [43]. Depolymerization and liquefaction of biomass can also be performed under oxidative or reductive conditions producing renewable-based oils [83]. For instance, the oxidative depolymerization reaction has the benefit to produce a pool of high-value and functionalized green chemicals. On the other hand, the reductive depolymerization gives a considerable yield of deoxygenated monomers, such as BTX (benzene, toluene, xylenes) products which are of interest as platform fuel precursors. The heterogenous structure of the biomass components like lignin impacts the selectivity for linkage cleavage and consequently on the selectivity for oligomeric, dimers, and monomeric products. One of the similarities of these processes is that the depolymerized lignin fragments contain different functional groups like methoxy (CH3O-), hydroxyl (-OH), benzyl alcohol (C7H8O-), ketone (R-CO-R), and aldehyde (-CHO) groups which contribute to the high oxygen content of the bio-oils. Similar functionalities and product spectrum can be seen in bio-oils derived from the pyrolysis of biomass. The oxygen content in bio-oil contributes to negative properties like poor heating value, high viscosity, corrosiveness, thermal and chemical instability [84]. Due to such detrimental characteristics, an upgrading process like hydrotreatment which includes hydrodeoxygenation, hydrocracking, hydro-decarbonylation and decarboxylation, and hydrogenation is required before application as a biofuel. In a lab-scale experiment, the reaction atmosphere for hydrotreatment involves a temperature range of 300–450 °C and a hydrogen pressure of 50–200 bar mimicking the operating conditions of a refinery hydrotreating process. The hydrotreatment of the lignin feedstocks aims to first cleave the recalcitrant linkages such as the carbon-carbon (C-C) and ether (C-O-C) bonds present in the lignin chemical structure. Then the produced alkylphenols are further reacted to form deoxygenated aromatics and cycloalkanes. This section will present an overview of the use of sulfided catalysts in supported and unsupported form for the hydrotreatment of bio-oil model monomer compounds such as guaiacol, phenol, anisole, and cresol. More attention will be dedicated to the discussion on the reaction schemes of the hydrodeoxygenation of the oxygenates that are present in the bio-oils over sulfided catalysts. The catalytic mechanism such as the active sites of sulfided catalysts and reaction network when using different bio-oil model compounds will also be discussed. Table 4 presents the state-of-the-art of metal sulfide catalysts for hydrotreating phenolic monomers. Various catalytic systems employing mixed oxide supported and sulfided catalysts have been reported for the HDO of phenolics. Garcia-Mendoza et al. have studied the activities of NiWS supported on TiO2, ZrO2, and the mixed oxide TiO2-ZrO2 for the HDO of Guaiacol at 320 °C [85]. Their results show that the NiWS supported catalysts system shows remarkable influence in shifting the distribution of the product towards deoxygenated products with NiWS supported on TiO2 showing an 80% HDO product selectivity at full guaiacol conversion [85]. The authors also speculated that the synergistic effect of NiWS and TiO2, and also the NiWS phase were responsible for the high catalytic and deoxygenation ability [85]. In a similar catalyst system, Hong et al. have shown that a 2 wt% Ni loading and 12 wt% W loading on such mixed oxide sulfided catalysts can give full guaiacol conversion and a 16% cyclohexane yield under different reaction conditions [86]. The study also mentions that nickel (Ni) performs better than cobalt (Co) as a promoter in catalyzing the HDO of guaiacol [86]. Another study using CoMoS supported on the mixed oxide Al2O3-TiO2 for the HDO of phenol has also shown that the mixed oxide improved the HDO activity with a better metal-support interaction than the conventional CoMoS supported on Al2O3 [87]. The use of activated carbon as catalyst support has also been reported in the literature [88–90]. Mukundan et al. have prepared an amorphous highly dispersed and disordered nanosized MoS2 single-layer on activated carbon by a microemulsion technique for guaiacol HDO and found that the single-layer MoS2/C promotes deoxygenation and hydrogenation better than a multi-layered MoS2/C in the production of phenol [90]. The conclusion was made based on the ratio of phenol/catechol produced using single or multi-layered MoS2/C, where the single-layered catalyst gave a higher ratio. This result inferred the importance of the morphology of the MoS2 catalyst in affecting product selectivity. Moreover, Mukundan et al. proposed a reaction pathway for HDO of guaiacol based on the detected compounds over the course of 5 h as shown in Fig. 7 [90].Templis et al. studied hydrotreatment of phenol over a NiMo/γ-Al2O3 catalyst in reduced and sulfided form [91]. Results demonstrated that the reaction routes for phenol hydrodeoxygenation occurred via two main parallel routes, the first one is direct deoxygenation (DDO) of phenol, and the second is the hydrogenation of the phenol ring forming cyclohexanol and followed by the hydrogenolysis removing the hydroxyl group producing cyclohexene and cyclohexane. The main difference between both catalysts was that the sulfided catalyst had a high cyclohexane selectivity of more than 90% while the reduced catalyst had higher activity for phenol conversion. Their results indicated that the sulfided catalyst favored the DDO route while giving high cyclohexane selectivity. Fig. 8a) shows the general reaction network for the HDO of phenol using a sulfided NiMo catalyst [91]. Adilina et al. also studied the classical NiMo catalysts supported on pillared clays (PILC) in reduced (NiMoPR) and sulfided (NiMoPS) forms for the HDO of guaiacol [104]. They have used techniques like quasielastic neutron scattering (QENS) and inelastic neutron scattering (INS) measurements to understand the interaction between guaiacol and the reduced and sulfided NiMo catalysts with the clay support [104]. Their results revealed that guaiacol adsorbed on these types of catalysts via two interaction modes, as illustrated in Fig. 8b): the first interaction is guaiacol adsorbed with the Ni-Mo-S site via an H-bonding interaction for sulfided catalysts and the second interaction is chemisorption of guaiacol on both the Ni-Mo site and also the clay support via phenate formation as can be observed in the reduced catalysts [104]. Their results with the sulfided catalysts also showed high activity and selectivity for guaiacol HDO.The promoters play a role in conventional hydroprocessing catalysts. Badawi et al. have demonstrated that cobalt promotes both DDO (Direct cleavage of the hydroxyl group) and HYD pathways (Hydrogenation of the aromatic ring and followed by the cleavage of the hydroxyl group) in the HDO of phenol to different extents [92]. They have performed DFT calculations and have shown that both DDO and HYD pathways occur on sulfur vacancy sites (CUS) [92]. Romero et al. have also reported the same findings [95]. Using 2-ethylphenol as a model compound [95], they have found that both Ni and Co improve the deoxygenation rate, while Ni only facilitates the HYD pathway. The reaction mechanism for DDO and HYD is illustrated in Fig. 9, respectively [95]. The main difference between these two pathways is that HYD originated from flat adsorption by the aromatic ring while the DDO pathway adsorption occurred through the oxygen atom.In addition to Ni and Co, a study conducted by Yang et al. has demonstrated that phosphorus (P) was able to promote the phenol HDO activity over a CoMoS-supported MgO catalyst, and they proved that DDO is the major pathway in phenol deoxygenation [96]. A non-conventional hydrotreating catalyst like supported ReS2 has been reported in several studies [100,99,101,103,102]. For instance, ReS2 supported on SiO2 or γ-Al2O3 supports was applied in the processing of dimethyl dibenzothiophene and guaiacol [102]. Both Re-based catalysts showed high HDS and HDO activities; ReS2 supported on SiO2 showed high HDO rates giving 40% HDO products [102]. In addition to inexpensive transition metals used as promoters, research has examined the use of rare earth and noble metals as promoters for metal sulfide catalysts in phenolics HDO [98,99]. For instance, Ir and Pt have been incorporated into RuS2/SBA-15 and used in the HDO of phenol [98]. The results have demonstrated a higher conversion rate of phenol (37–41%) and better cyclohexane selectivity (62–63%) than with the non-promoted RuS2/SBA-15 [98]. It is important to note that the use of noble metals involves high costs for catalyst production, which limits their industrial application. The sulfur content in some bio-feedstocks, such as Kraft lignin, may act as a poison to such noble catalyst systems, nevertheless, studying such a system facilitates better insight into the reaction pathways of the HDO of phenolics.Jongerius et al. studied a pool of lignin model compounds using CoMoS supported on Al2O3 under the same reaction parameters (300 °C, 50 bar H2, 4 h, and batch system) for comparison [94]. Their main findings suggest that the mono-aromatic oxygenates underwent three distinct pathways that included HDO, demethylation, and methylation. This provided invaluable products like phenol, benzene, cresols, and toluene [94]. Less than 5% of hydrogenated products were detected in the reaction medium, indicating that hydrogenation is the least preferred reaction network for this catalyst system [94].It is commonly found in the considerable number of studies on the HDO of phenolic compounds that sulfiding agents, such as dimethyl disulfide (DMDS) or carbon disulfide (CS2), were co-fed during an experiment to create H2S to maintain the sulfidation degree of the sulfided catalyst. Results show that adding a sulfiding agent during the HDO process had a negative effect on the HDO activity of phenolics but promoted the HDO of aliphatic oxygenates such as vegetable oils and animal fats [66]. As a result, one should note that the addition of a sulfiding agent also plays a role in affecting the effectiveness of the catalyst other than the type of reactant being used. Ferrari et al. have studied the effect of H2S partial pressure and sulfidation temperature on the conversion and selectivities of phenolics [88]. It was found that the increase in H2S partial pressure reduced the formation of deoxygenated products from the HDO of guaiacol over CoMoS supported on carbon [88].Over recent decades, these traditional TMS catalysts have been tested by omitting the use of catalyst support, resulting in unsupported TMS. There are several methods to prepare unsupported TMS, that can be used in the hydrotreatment processes. One of these is a hydrothermal synthesis with synthesis parameters, such as moderate synthesis temperature (150–250 °C) and the absence of hydrogen pressure [112,123,122,124,105,125]. The synthesis method involved simple operation and also controllable catalyst morphology, allowing easy scale-up for industrial application. Wu et al. prepared a series of hydrophobic unsupported MoS2, NiS2-MoS2, and CoS2-MoS2 using hydrothermal synthesis with the aid of silicomolybdic acid for the HDO of 4-ethylphenol [105]. The CoS2-MoS2 catalyst achieved a 99.9% 4-ethylphenol conversion with a 99.6% ethylbenzene selectivity after 3 h. The catalyst showed good recyclability after 3 runs at 225 °C [105]. Cao et al. also presented results for highly efficient unsupported Co/MoS2-x (x is the molar ratio of Co/(Co + Mo)) catalysts with high dispersion, rich in defects, and curvy slabs in deoxygenation of p-cresol under mild HDO condition [121]. During the hydrothermal synthesis of catalysts, the accommodation of Co atoms in the coordinative unsaturated sites (CUS) of MoS2 and also edge sites resulted in the formation of a Co-Mo-S active phase that enhanced the HDO activity [121]. Their results further demonstrated that the Co/(Co + Mo) molar ratio of 0.3 provided the highest HDO activity and toluene selectivity, on the other hand, excessive Co introduction, causing the formation of large Co9S8 particles that hinder the HDO active sites with decreased HDO activity [121].A two-step strategy for the synthesis of Co-MoS2−x catalysts involved first the hydrothermal and then subsequently the solvothermal method as explored by Wu et al. [122]. They concluded that the reducing ability of MoS2−x induced by the sulfur vacancy was able to reduce the Co precursor to Co metallic while decorating the edges of MoS2−x, leading to the formation of the metal-vacancy interfaces that catalyze the HDO reaction [122]. They further performed Gibbs free-energy calculations ( Fig. 10) and clarified that DDO of 4-methylphenol underwent a two-step, hydrogenation and dehydration with CH3C6H4OH+ and CH3C6H4 - being the transition states [122]. As can be seen in Fig. 10, the Gibbs free energies for the formation of the transition state species were decreased for Co-MoS2−x in comparison to MoS2−x indicating that the metal-vacancy interface favored the adsorption of 4-methylphenol and lowered the reaction energy barriers, hence enhancing the HDO activity [122]. Another study by Wang et al. has proposed a reaction network for p-cresol HDO using a hydrothermally prepared CoMoS catalyst, as shown in Fig. 11a) [112]. Two different deoxygenation routes for p-cresol have been proposed: the first is the DDO route, where the partially hydrogenated dihydrocresol is attacked and the dissociated H+ and the OH2 + species are cleaved in the form of H2O producing toluene [112]. The second route involves HYD where the partially hydrogenated p-cresol is fully hydrogenated to 4-methylcyclohexanol and then dehydrated to 3-methycyclohexene. The product, 3-methylcyclohexene then undergoes hydrogenation and forms methylcyclohexane [112]. The study also described a p-cresol adsorption scheme on an unsupported CoMoS catalyst [112], as shown in Fig. 11a). p-cresol could adsorb via its vertical orientation and coplanar position in relation to the DDO and HYD routes, respectively [112]. It can also be considered that there is a difference in the CoMoS properties which determine the adsorption orientation and thus the reaction route [112]. For instance, it was mentioned in their study that an increase in the number of MoS2 layers and also reduced slab length can enhance the toluene selectivity and p-cresol conversion [112].The use of a template like zeolitic imidazolate framework-67 (ZIF-67) was used to prepare a self-supported defect-rich CoMoS-x catalyst for HDO of p-cresol [120]. The main finding in this work highlighted the importance of a pre-reduction of the catalysts in decalin (300 °C, 30 bar H2, and 6 h), and such pre-treatment promoted the formation of sulfur vacancies on the MoS2 surface and facilitated the surface restructuring of Co-Mo interfaces resulting in the in-situ generation of an abundance CoMoS active sites favoring the DDO route, as shown in Fig. 12 [120]. A hard template like mesoporous silica SBA-16 has also been used to synthesize an unsupported NiMoW sulfide catalyst for the HDO of guaiacol in a fixed-bed reactor [111]. The NiMoW sulfide unsupported catalyst gave a 99.6% guaiacol conversion with minimal coke formation at 400 °C [111]. Adapted from the reference, shown in Fig. 12 b), guaiacol underwent HDO via demethylation (DME), demethoxylation (DMO), and transalkylation [111]. Phenol was formed by either the direct demethoxylation of guaiacol or the dehydroxylation of catechol; both reactions resulted in the production of benzene [111]. It is worth noting that phenol was first obtained from the HDO of guaiacol as a reaction intermediate caused by the higher bond dissociation energy for the hydroxy group on the aromatic ring than the methoxy group [84].Furthermore, the cleaving and HDO of dimeric phenols are of interest because the depolymerized lignin streams and bio-oils can contain not only monoaromatic compounds but also various aromatic dimers and oligomer fragments. These intermediates are present in the reaction feed, and they should be cleaved during the deoxygenation process. Hence it is necessary to study the HDO of model fragments that can mimic specific structural linkages that can be found in lignin under HDO conditions. Metal sulfides have been explored for cleaving lignin model dimers having C-O-C ether and C-C linkages. The cleavage of lignin dimers containing etheric linkages (α-O-4 or β-O-4, 4-O-5, etc) is faster due to their low bond dissociation energy while C-C linkages (5–5′, β-1, β-β, etc.) are quite recalcitrant. The bond dissociation energy of the etheric linkages is in the order of 4-O-5 (ca. 330 kJ/mol) > β-O-4 (ca. 289 kJ/mol) > α-O-4 (ca. 218 kJ/mol) while for the 5–5′ linkage it is ca. 490 kJ/mol [126–128]. The review for model lignin dimers ( Table 5) is discussed in this section while Section 4 of this paper is focused on real lignin feed for hydrotreatment. These types of model dimer compound-related studies can provide a means of screening the effectiveness of catalysts and understanding reaction mechanism schemes that occur during lignin depolymerization. Various studies involving lignin dimers revealed that sulfided catalysts can effectively break down the dimers to monomers however to a different extent. Koyama et al. showed the hydrocracking of benzyl phenyl ether, diphenyl ether, benzyl phenols, diphenylmethane, and dibenzyl over sulfided NiMo/alumina, Fe2O3/alumina, and MoO3/TiO2 in the range of 340–450 °C [129]. Monomer yield increased with increasing temperature for model dimers having a phenolic hydroxyl group and C-O-C linkages while the C-C biphenyl bond was quite recalcitrant even up to 450 °C. Additional dimers were also found to be formed due to dehydroxylation and subsequent hydrogenation reactions.As mentioned earlier, Jongerius et al. reported that with a commercial sulfided CoMo/Al2O3 the β-O-4 bonds of phenyl coumarin alkyl ether could be cleaved to monoaromatics while 5–5′ linkages could not be broken at 300 °C and 50 bar of H2 pressure [94]. Shuai et al. reported a substantial yield of aromatic monomers from the selective cleavage of -CH2- linked C-C phenolic dimers over a commercial CoS2 catalyst at 250 °C and 50 bar of H2 following 1.5 h of reaction [130]. However, β-1 and 5–5′ C-C linked dimers could not be cleaved but rather transformed into hydroxyl dimers via demethoxylation. Surface engineering of unsupported Co-promoted MoS2 nanosulfides by Song et al. showed that diphenyl ether (4-O-5) could be cleaved efficiently to produce benzene selectively [106]. The same author also claimed that in-situ exsolution synthesized CoMoS on a sulfated zirconia support enhanced the 4–0–5 (diphenylether) C-O cleavage and the subsequent deoxygenation to benzene, toluene, and ethylbenzene [133]. Their work also compared the catalysts prepared by a conventional impregnation and physically mixed sulfated ZrO2 supported CoMo sulfide catalysts, and showed that the exsolution method represented the best method based on the activity and characterization tests. The activity enhancement was mainly attributed to the better interaction between the CoMo sulfide phase and support, Lewis acid sites of sulfated ZrO2, and highly-dispersed CoMo sulfide [133]. Ji et al. showed that nanocrystalline pyrite and marcasite (FeS2), supported over activated carbon was highly active and selective for the hydrodeoxygenation of dibenzyl ether into toluene at 250 °C under an initial H2 pressure of 100 bar for 2 h [134]. The high activity of such a catalyst was claimed to be due to the transformation of surface FeS2 into Fe(1−x)S [134]. A proposed reaction pathway for the chemical transformation of dibenzyl ether to benzaldehyde and toluene is shown in Fig. 13a) [134]. The proposed chemical pathway involves the formation of phenylmethylium, and subsequent cleavage of the ether bond. Then hydride transfer takes place from phenylmethanolate to phenylmethylium, resulting in the formation of targeted products, benzaldehyde, and toluene. In another study, Zhang et al. proposed that the cleaving of lignin β-O-4 ether bonds can occur through a dehydroxylation-hydrogenation reaction over the acid-redox site of a NiMo sulfide catalyst which significantly lowers the bond dissociation energy and subsequently facilitates the formation of styrene, phenols, and ethers with H2 and an alcohol solvent [135]. A potential main route for the conversion of 2-phenoxy-1-phenylethanol (β-O-4-A) over the NiMo sulfide catalyst was proposed in their work which involves the adsorbed β-O-4-A firstly losing a hydroxy group and generating carbocation intermediates (PhCHδ+CH2OPh) at the weak and medium acid sites of the catalyst. The carbocation intermediates were then transformed into radical intermediates (PhCH·CH2OPh) by obtaining an electron from the catalyst redox cycle (Fig. 13b). Due to the lower bond dissociation energy (BDE) of the Cβ-OPh ether bond in the radical intermediate (66.9 kJ/mol) than the one in β-O-4-A (274.0 kJ/mol), the Cβ-OPh bond cleaves easily, and then the generated radical species react with activated H2 or methanol to form various arene products like phenols and ethers [135].A similar strategy of peroxidation via O2/NaNO2/2,3-Dichloro-5,6-dicyano-1,4-benzoquinone (DDQ)/N-Hydroxyphthalimide (NHPI) and subsequent hydrogenation over a NiMo sulfide catalyst was found beneficial for the overall cleavage of β-O-4 model compounds [136]. Recently our group studied lignin dimer hydrotreatment involving sulfided NiMo over Al2O3 and ultra-stable Y zeolite (USY). It has been found that the NiMoS-USY combination can give a high yield of mono-aromatics including deoxygenated aromatics, mono/alkyl phenols, and cycloalkanes via hydrogenolysis of etheric C-O bonds and subsequent cleavage of C-C intermediates formed by transalkylation reactions [131]. Interestingly, with the NiMoS-USY combination, recalcitrant β-1, -CH2 linked dimers and 5–5′ linkages could be significantly cleaved due to a suitable balance between Brønsted acidity and Ni-promoted MoS2 redox sites. This study led to a further study investigating USY with different silica/alumina ratios (SAR) as catalyst supports for NiMoS in valorizing lignin dimers [132]. Among all the studied catalysts, NiMoS on USY with an intermediate SAR of 30 was found to be most effective in cleaving β-O-4 and C-C linkages with a high degree of hydrogenolysis, hydrocracking, and HDO reactions owing to an apparent optimal balance between acidic and deoxygenation sites. Moreover, it was found that a higher surface acidity promoted the initial conversion of 2-phenethylphenylether (PPE) and 2,2-biphenol (5–5′) via transalkylation and isomerization [132]. These results suggested that tuning the acidity and pore sizes of the USY-zeolite and further impregnating with NiMoS was an effective way to create a catalyst that is efficient in both breaking recalcitrant lignin linkages and achieving a high level of deoxygenation.Pyrolysis oil produced from the fast pyrolysis of solid biomass contains a variety of compound groups like sugars, alcohols, phenols, ketones, aldehydes, furans, and acids. These compounds contribute to various detrimental properties of pyrolysis oil which need to be addressed before the full utilization of pyrolysis oil as a fuel or other applications by hydrotreatment in petroleum refineries. Shumeiko et al. performed a series of screening tests of lab-synthesized and commercial sulfided NiMo catalysts for long-term hydrotreatment of wheat/barley (50/50 wt%) straw-derived pyrolysis oil in a fixed-bed reactor aiming to produce a hydrotreated pyrolysis oil that is compatible with the petroleum refinery fraction for coprocessing [137]. The assessment of their results was based on the HDO and HDS activity. Their results showed that the catalysts synthesized by co-impregnation were better than the catalysts prepared by a two-step impregnation procedure despite having the same active NiMo phase loading and the same commercial alumina support. While the commercial catalysts performed worst among all the catalysts in terms of HDO activity, they showed the best HDS performance [137]. The difference in activity may be attributed to the different physicochemical properties of the catalysts and also preparation methods. These results suggest that the HDS activity of a sulfide catalyst is not suitable for indicating its HDO performance for pyrolysis oil. The long-term experiments (80 h) in their work were useful in understanding the deactivation of the sulfided catalysts [137]. Their experimental results showed that the product quality changes, which was indicated by a gradual loss of catalyst activity with increasing time-on-stream [137]. Thus, their work demonstrated the feasibility of using biomass-derived pyrolysis oil to obtain a compatible feedstock for co-processing in a refinery, however, the stability of the sulfided catalysts needs to be fully addressed before achieving a successful deployment of the technology. Another study conducted by Zhang et al., upgraded a pyrolysis oil produced by the fast pyrolysis of forest residues with light cycle oil (LCO) as a reaction medium using a dispersed unsupported MoS2 catalyst [6]. The use of the dispersed unsupported catalysts was considered to allow better interaction between the active sites of the catalysts, hydrogen, and the heavy feedstock resulting in less solid yield. The low solid yield ranging from 0.8 to 1.8 g/100 g bio-oil at the end of their experiment showed that the use of dispersed unsupported materials can suppress the side reactions such as polymerization and re-polymerization of the large molecular weight compounds and reactive species that result in solid residues.Priharto et al. studied the hydrotreatment of pyrolysis oil derived from lignin-rich digested stillage over commercial sulfided NiMo and CoMo catalysts [138]. They demonstrated the feasibility of utilizing solid waste residues from bioethanol processes for the production of pyrolysis oil. The further hydrotreatment of the pyrolysis oil also resulted in an appreciable oil yield of 60–64 wt% [138]. It should be noted that the nitrogen content in such feed should be refined employing hydrodenitrogenation (HDN), as, from their GCMS analysis, nitrogen-containing aromatic heterocyclic compounds present in the feed like indoles were converted to pyrroles. Hence, the removal of nitrogen content in pyrolysis oil by the means of hydrodenitrogenation (HDN) should be addressed in any future study with the aid of sulfided catalysts. For instance, Izhar et al. studied HDN of fast-pyrolysis oil derived from sewage sludge over a phosphorus-promoted sulfided NiMo/Al2O3 catalyst [139]. The main finding from their work showed that dissolving pyrolysis oil using a non-polar solvent like xylene improved nitrogen removal compared to using protic solvents due to the competition between denitrogenation and deoxygenation reactions [139].Lignin (an amorphous solid) is a renewable and sustainable future source of aromatic compounds for the chemical industry [140]. Lignocellulosic biomass contains up to 33 wt% of lignin. Softwoods (coniferous woods, e.g. spruce), contains 27–33 wt% lignin. Hardwoods (deciduous species, e.g. birch) have 18–25 w% lignin and grasses constitute 17–24 wt% of lignin [141,142]. Chemically, lignin is a 3-dimensionally complex biopolymer composed of three basic structural units; sinapyl alcohol, coniferyl alcohol, and coumaryl alcohol. The composition of these structural units is different among the lignocellulosic biomasses. In softwoods, the coniferyl alcohol units form 90–95 wt% of the lignin with the remaining being only the sinapyl alcohols (10–5 wt%). In hardwood lignin, an equal amount of (50–50 wt%) coniferyl and sinapyl alcohols are observed. These woods are devoid of the coumaryl alcohol structural unit. Grasses contain 0–5 wt% of the coumaryl alcohol units, but the major contributions are from the coniferyl alcohol (75 wt%) and the sinapyl alcohol (20–25 wt%) units. The structural units in lignin are connected by various C-O-C ether linkages (α-O-4, β-O-4, 4-O-4) and C-C linkages (β-β, β-1, β-5, 5–5) (see Fig. 15) to generate the three-dimensional structure of lignin. Among these two kinds of linkages, the ether linkages (65%) dominate in lignin, of which the β-O-4 alone accounts for 50% of all ether linkages. The composition of the chemical linkages in softwood, hardwood, and grass lignin are also dissimilar. For instance, softwood contains more C-C bonds than hardwood. Also, the C-O-C and C-C linkage patterns vary among the softwood/hardwood/grasses plants classes too. Lignin structure is very complex among and within the various lignocellulosic biomasses. This necessitates the exact structural determination of every isolated lignin which is very challenging to accomplish [143].The natural sources of lignins are the many agricultural residues (grain dust, sunflower stalk, bagasse, etc.) and forest residues [3]. The lignins from these sources can be utilized only after their extraction. Otherwise, they need to be co-processed along with the cellulose and hemicellulose fractions of the biomass. On the other hand, commercial lignins (also known as the technical lignins: Kraft lignin, soda lignin, hydrolysis lignin, etc.) are already extracted lignins almost devoid of the cellulose and hemicellulose fractions. Commercial lignins are usually generated as the byproducts of various commercial pulping/hydrolysis processes. For instance, Kraft lignin is generated as the byproduct of the Kraft pulping process, likewise, the soda pulping process generates soda lignin as a byproduct, and hydrolysis lignin is produced during the enzymatic hydrolysis of cellulosic biomass to ethanol [3]. The severity of the pulping processes and the chemical reagents used in the processes (NaOH, Na2S, H2SO4, etc.) [140] adversely affect the lignin structure. Bonds are broken and new ones are generated during the processes. Hence, the structures of commercial lignins are different from their natural counterparts. Currently, most of the commercial lignins produced are utilized as a combustion fuel in the pulping process to regenerate heat energy for the pulping process. This gives rise to a low value-addition of the lignin ($70–150 per ton). On the other hand, the conversion of commercial lignins to chemicals (e.g.: phenols, benzenes, toluene, xylenes, etc.), and fuels significantly improves its value-addition (approximately $1300 per ton) [7].The sole way to obtain the different monomeric aromatic compounds from lignin is through its depolymerization. In general, lignin depolymerization can be achieved by techniques such as pyrolysis, gasification, hydrogenolysis (H2), chemical oxidation (O2), and hydrolysis (H2O) [144]. Other methods include microwave-assisted lignin depolymerization, biological depolymerization, and a so called lignin-first approach, as summarized in Fig. 15. There are many excellent reviews available in the literature on various lignin depolymerization techniques and their different advantages [3,140,141,143,144–151]. Anyways, the produced bio-liquids needs further upgrading to produce bio-fuels. For instance, the liquid bio-oil obtained from the pyrolysis of lignin is corrosive and high in oxygen content. To produce value-added compounds from this bio-oil, an additional step involving catalytic hydrotreatment is necessary. On the contrary, chemical depolymerization especially using heterogeneous catalysts has the advantages of high product selectivity to either value-added oxygenates or deoxygenates in a single step, high efficiency of the reagents used, and moderate reaction conditions, and the ease of reaction control [140]. This part of the review focuses on the chemical reductive method (H2) of lignin valorization using sulfided catalysts.The first step in lignin depolymerization is its thermal degradation to oligomeric lignin fragments ( Scheme 1) [152] . The macromolecular structure of lignin is a major hurdle for employing solid catalysts during this step, since, in most of the porous solid catalysts, their pore dimensions do not match with the dimensions of the lignin macromolecules. However, homogeneous catalysts may better facilitate the first step. In the second step, the formed smaller fragments undergo further degradation to form oxygenated monomeric molecules. Solid catalysts can perform this step. In the third step, deoxygenation (catalytic HDO) of these monomers to liquid aromatic products ensues. Further catalytic hydrogenation of aromatics to cyclic aliphatic compounds follows in the final step (Scheme 1). The gaseous products of lignin depolymerization are mainly, CO2, CO, CH4, and C2-C4 alkanes. At any stage of the lignin depolymerization, the repolymerization of the depolymerized lignin fragments, aromatic oxygenates, and aromatics may occur. This repolymerization occurs mainly through C-C bond formation (on the other hand, the C-O-C bonds may undergo further breakage) leading to the formation of a solid phase composed of polycondensed aromatic structure, usually called as lignin depolymerization char [152] . The main challenge in lignin depolymerization is to reduce the char formation and concurrently to increase the formation of the liquid products [153].The conversion in lignin depolymerization in the literature is expressed in contrasting ways by different research groups. Some authors separate the unconverted lignin from the char through solvent extraction to calculate the actual lignin conversion to liquids and gases. While others consider only the total amount of solid products (also containing unconverted lignin) obtained after the depolymerization in their calculation. The liquid product yields are also represented in different ways in the literature. The weight of bio-oil produced after depolymerization (sometimes expressed as the wt% of a particular solvent-soluble fraction) is the common method found in the literature. In the very early reports on lignin depolymerization, both the conversion and the composition of bio-oils are rarely mentioned. Whenever the composition of the bio-oil was specified in the literature, it is briefly stated in this section of the review. However, to obtain more detailed information about the full composition of the bio-oil and conversion, it is recommended to follow the corresponding cited references. Moreover, some literature represents the char yield as ‘char’, while others refer to it as ‘solid residue’ obtained after the reaction. These terms can be synonymous, but in some cases, the ‘solid residue’ could also include char as well as unconverted lignin.Early reports on the use of sulfided catalysts in lignin conversion were mainly focused on examining to what extent lignin valorization could be achieved rather than the catalyst structure, and the activity correlations. The choice of the various sulfided catalysts for this was purely based on their reputation for sulfur tolerance and capability for desulfurization/hydrogenation/hydrogenolysis activity. One such early study was reported by Vuori et al. in 1988, where they compared the lignin liquefaction under mild thermolysis conditions (< 400 °C) and catalytic conditions [153]. A commercial sulfided CoMo/Al2O3 was employed as the hydrotreating catalyst. Kraft lignin was the feedstock and tetralin was chosen as a hydrogen donor solvent (tetralin can act both as a solvent and 4-hydrogen atom donor to the reaction medium). The liquid products of the reaction (345 °C, 20 bar of H2, 5 h) mainly constituted phenols and acids (ether soluble fraction) amounting to 11.5 wt% in yield. This yield was not very much higher than the thermolytic reaction where the liquid product yield was 8.1 wt%. In the gaseous products, CH4 (from lignin) contributed around 2.7 wt% yields from the catalytic reaction as compared to 1.1 wt% from the thermolytic conditions, suggesting an enhancement in reaction rate under catalytic conditions. Surprisingly, the authors found more char formation with the catalytic process (33 wt%) than the thermolytic process (21 wt%), leading to a conclusion that the catalyst could improve the reaction rate, however, was incapable of preventing the condensation reactions leading to char. Even the presence of both the hydrogen donor solvent (tetralin) and H2 pressure (20 bar) was unable to prevent condensation reactions from lignin fragments [153].In general, solvents in lignin depolymerizations can stabilize the reactive intermediates, enhance the catalyst-lignin interactions, and promote the solubility of lignin. The role of solvents in lignin depolymerization was reviewed in detail by Raikwar et al. [154]. To understand the effect of hydrogen-donor and non-donor solvents on the lignin liquefaction process, Schuchardt et al., employed a set of different solvents such as xylene, pyridine, cyclohexanol, isopropanol, and tetralin; the latter three being hydrogen-donor solvents [155]. The catalyst of choice was ferrocene in situ sulfided with CS2 or S. Among these solvents, tetralin gave the maximum heavy-oil yield of 44 wt% at 65 wt% of lignin conversion (400 °C, 270 bar at 400 °C, 0.5 h). This was attributed not only to the high hydrogen-donor ability of tetralin but also to its heavy-oil extraction ability.The C-C bonds in lignin are more difficult to cleave than the C-O bonds (vide supra). Therefore, catalysts for lignin depolymerization also need to be efficient for the C-C bond cleavage since a large amount of char (repolymerized lignin-containing a large amount of C-C bonds) formed during the depolymerization needs to be broken down as well for higher monomer yield. A comparative study using unsupported MoS2 and CoS2 on the model compound dimethylguaiacylmethane containing a methylene bridge and Kraft lignin showed the CoS2 catalyst to be superior in C-C bond cracking (11.7 wt% yields to the aromatic monomer from Kraft lignin at 250 °C, 50 bar H2, 15 h) than MoS2 [130]. However, the catalyst has a limitation, not all C-C bonds could be cleaved by CoS2. It is efficient only if there are hydroxy groups on any of the benzene rings related to the ortho position of the methylene/C-C bond. In the absence of these -OH groups, no C-C bonds were cleaved. The influence of the -OH group was more effective than the -OMe group in the C-C bond cleavage. Further studies showed that the active phase of the catalyst for C-C bond breakage is not CoS2 but CoS generated by the reduction of CoS2 in the H2 atmosphere. Complementary to this, the surface composition of the catalyst after the reaction showed a larger contribution of CoS ( Fig. 16a). The recycling studies showed a decrease in catalytic activity. The Co:S atomic ratio was reduced from 1:1.7–1:1 in the third run and the presence of the CoO phase was noticed. Conditional experiments with CoO showed less activity, indicating that the most active phase of the reaction was its sulfided form. The decrease in activity during the recycle runs can thus be attributed to its oxidation to the oxide phase. Both the supported catalysts sulfided CoMo/Al2O3 and unsupported CoS/S2 are effective for the breakage of the C-C bonds in lignins.Lignin depolymerization behavior in the presence of a sulfided catalyst during the heating period of the batch reactor is often overlooked. This information gives an idea about the lignin fragmentations happening in the presence of the sulfided catalysts in the early hours of the reactions where there is a large temperature variation. This was investigated by Joffers et al., using Protobind 1000 lignin and a commercial sulfided NiMo/Al2O3 catalyst [157]. The time taken for the reactor to reach the desired temperature of 350 °C was only 14 min. Immediately following the 14 min, the lignin conversion was 27 wt%. This lignin conversion resulted in 24 wt% yields of liquids (mainly monomeric phenols), and 3 wt% of gases (CH4, CO, CO2, and C2-C5 hydrocarbons). The remaining was the lignin residue (solid). Gel permeation chromatographic (GPC) analysis of the tetrahydrofuran soluble fraction of this lignin residue (oligomers) showed a molecular weight of 3575 g/mol, corresponding to 20 phenylpropane units (parent lignin had 26 units). These monomer units were further reduced to 6 after 28 h of reaction [156]. The THF-soluble lignin residue still contained stronger C-C linkages between monomer units, which were difficult to break. This could be due to the reaction conditions or perhaps due to the choice of a NiMo/Al2O3 catalyst than a CoMo/Al2O3 catalyst. Characterization of the catalyst after the reaction confirmed coke formation. Moreover, the sulfur content in the catalyst was decreased to 7 wt% as compared to the 9 wt% in the freshly sulfided catalyst (Protobind 1000 had only 0.1 wt% of sulfur in it). Fig. 16b) shows the changes in the BET surface area, pore volume, and pore diameter of the catalyst as a function of reaction time. All these textural properties decreased in the first hour of the reaction and became almost constant after 28 h of the reaction. The sulfided catalyst showed good stability for longer reaction runs. The main cause of its deactivation is due to the loss of sulfur. Therefore, an additional sulfidation step is necessary to regenerate most of the initial activity.To increase the monomer yield during lignin depolymerization, i.e., effectively removing the products from the catalyst preventing their further transformation, and competition for active sites, the lignin depolymerization process was attempted in a semi-continuous mode using sulfide catalysts. In the case of both a batch reactor and a fixed-bed reactor, the semi-continuous mode involves the continuous withdrawal of the reaction products under the flow of the H2. The lignin and the solvent (if any) are already placed in the reactors; they are not in a continuous feeding mode. In one such study, a NiMo/Al2O3 catalyst was mixed with lignin (hydrolysis lignin) and packed into a tubular reactor [158]. No solvents were used in the depolymerization. At the reaction temperature (380 °C, 40 bar of H2, 4 h), the lignin underwent thermal degradation to smaller fragments, which were then subsequently transformed at the catalyst active site. The H2 gas which was continuously flowing through the reactor enabled the mass transfer. The liquid products were collected in a gas-liquid separator during the depolymerization, and it consisted of an aqueous phase and an organic phase. The liquid products were composed of hydrocarbons, oxygenates, and phenols. The gaseous products were CO, CO2, CH4, etc. The solid product was mainly the lignin condensation product, char, which was separated from the catalyst by sieving. The catalyst was sulfided either in situ in the presence of the lignin in the reactor using dimethylsulfide at a temperature range of 200–220 °C (this temperature range is lower than the decomposition temperature of the lignin), or ex-situ (a pre-sulfided catalyst mixed with lignin and loaded into the reactor). The elemental analysis of the in situ and ex situ sulfided catalysts showed a Mo/S ratio of 1.86 and 1.49 (w/w) respectively, indicating a lower degree of sulfidation in the in situ sulfided catalyst. The pre-sulfided catalyst was more efficient (higher amount of liquid and gaseous products, and a lower amount of char) than the in situ sulfided catalysts. The sulfidation state of the catalyst is the crucial factor for its activity. Increasing the catalyst-to-lignin ratio increased the liquid and gaseous product yield, and consequently decreased the solid residue. This behavior was attributed to the proximity effect between the catalyst active sites and the reactants, which increased with an increase in catalyst amount. Similarly, an increase in H2 pressure favored a higher rate of hydrogenation and deoxygenation during the hydrocracking process. The study showed the adaptability of commercial sulfided catalysts for process modifications.A slightly modified semi-continuous batch process with a sulfided CoMo/Al2O3 catalyst was reported by Pu et al. [159]. A constant flow of H2 to maintain the reactor pressure and a reflux system to extract continuously the light aromatic products and H2O from the reactor were the main features of the semi-continuous setup. The semi-continuous approach helped to remove the gases formed during the reaction so that their contribution to catalyst properties/deactivation could be avoided. However, the contribution of H2O which was not completely removed until the reaction temperature was reached, to catalyst deactivation, could not be avoided. When the catalyst was characterized after the reaction (350 °C, 80 bar H2, 13 h), significant changes in its composition and textural properties were observed ( Fig. 17). Coking had started at the early hours of the reaction. The carbon content in the catalyst after the first hour of the reaction was 12 wt% and remained the same until the end of 13 h (Fig. 17a). The sulfur content was also reduced from 7.5 to 6 wt% after 13 h of reaction (Fig. 17b). However, based on XPS analysis, the S/Mo atom ratio had decreased significantly when comparing the freshly sulfided catalyst to that after 13 h reaction (2.2 and 1.7 respectively, Fig. 17b). Sulfur loss during the reaction appears to be inevitable for the sulfided catalysts. Other notable changes were observed in the textural properties of the catalyst. The surface area and pore diameter decreased from 193 to 185 m2/g, and from 0.47 to 0.29 cm3/g respectively in the first hour and remained the same (Fig. 17a), while the average pore diameter decreased from 7.9 to 6.2 during the first hour and remained almost the same until 13 h. The changes in both the composition and textural properties of the catalyst occurred during the heating period where lignin started to undergo depolymerization (the 0th hour is immediately following the heating period in Fig. 17). Nevertheless, with all these changes in its properties, the catalyst was active for oligomer cracking and deoxygenation reactions during the 13 h (the liquid fraction increased from 44 to 82 wt%). Perhaps, the most important aspect influencing the activity of the catalyst is its sulfur content. If there was a continuous source of sulfur, it can be presumed that the catalyst could have maintained its activity for longer reaction times.Instead of the ordinary organic solvents used for lignin depolymerization, the use of slurry-oils with sulfided catalysts was also reported. Meier et al. studied the performance of a sulfided NiMo/Al2O3 catalyst in 5 different slurry-oils [160]. They were, (1) light fraction oil from bitumen and lignin coprocessing (S = 2 wt%), (2) heavy fraction oil of the same bitumen and lignin coprocessing (S = 4 wt%), (3) a recycled residual oil from (2), (4) standard vacuum gas oil, and (5) lignin-derived slurry oil. The performance of the catalyst in different slurry oils is compared in Table 6. Since the lignin-derived slurry oil already contained phenols, the calculation of phenolic yields solely from the lignin feedstock resulted in negative values because the amount of the initial phenol in the lignin slurry oil was subtracted from the total phenols produced after the reaction. The highest yield for solid residue (coke, 10.7 wt%) was obtained when the heavy fraction oil was used as the slurry oil. This amount was about 1.5 times lower than without the catalyst. The minimum yield to the solid residue (0.3 wt%) was obtained with lignin oil. Although the total amount of oil obtained from lignin with different slurry oils was in the high range of 68–83 wt%, the highest being in lignin slurry oil, the amount of phenol obtained ranged only between − 1.4–4.1 wt%. The phenolics yield without the catalyst was only 2.9 wt%. Hence, the catalyst had only a small effect in improving the phenolic yield during the hydrocracking process. When the heavy fraction slurry oil was used in its 3rd recycle run, a higher yield to total lignin oil (79 wt%), with a low yield to the solid residue (4.9 wt%) was obtained. The results of using slurry-oils containing sulfur in combination with a commercial sulfided NiMo/Al2O3 catalyst appeared highly promising for maintaining the sulfidation state of the catalyst and the commercialization of the process as other noble metal catalysts would normally undergo sulfur poisoning during the reaction.Sulfided catalysts were also used for solvent-free hydrotreatment of lignin. The solvent-free attempt was aimed at alleviating the techno-economic issues resulting from solvent recovery and recycling that would arise for large-scale production. Meanwhile, solvents have the advantages that they can impart good heat and mass transfer properties which are poorer in a solvent-free reaction [161]. The earliest study on a solvent-free depolymerization process using sulfided catalysts was reported by Oasmaa et al., who mainly focused on the process and influence of lignin types rather than the catalysts [162]. Five technical lignins; 3 pine Krafts, 1 birch Kraft, and 1 organocell, over a combination of two commercial catalysts; a sulfided NiMo/aluminosilicate catalyst and 20 wt%Cr2O3/Al2O3 catalyst (1:1) was used for the process. In general, the product oil yields (395–400 °C, 100 bar of H2, 0.5 h) followed the trend as organocell (71 wt%) > pine Kraft (63 wt%) > birch Kraft (49 wt%)). Out of the produced bio-oil, the detectable aromatic yield was in the range of 19 wt% for organocell, 21 wt% for pine, 14 wt% for birch Kraft lignins. The amount of solid residue produced after the reaction was however lower for pine Kraft lignin (4 wt%) than organocell and birch Kraft lignins (7 wt%). The study demonstrated that the sulfided catalyst could efficiently depolymerize the lignins to monomers under solvent free conditions at least in laboratory scale (70 g of lignin).Later, a comparative study of sulfided NiMo on two different supports including activated carbon (AC), and MgO-La2O3 oxide was attempted under solvent-free conditions [161]. The sulfided NiMo/AC gave 55 wt% yields (350 °C, 100 bar of H2, 4 h) to the dichloromethane soluble fraction (average molecular weight being 700 g/mol) with only 9 wt% yields to the solid residue. The monomers in this fraction were largely composed of alkyl phenolics and aromatics. The NiMo/MgO-La2O3 gave 48 wt% yields of the dichloromethane soluble fraction having an average molecular weight of 660 g/mol, however with 12.7 wt% of the solid residue. The NiMo/AC catalyst was slightly more effective in producing low molecular weight fragments from the lignin during the depolymerization. The XRD of NiMo/MgO-La2O3 after the reaction showed NiS, Ni3S4, and MoS phases on the catalyst, indicating its sulfidation state after the reaction ( Fig. 18). Only a negligible decrease in the surface area (from 29 to 23 m2/g) and pore volume (from 0.16 to 0.14 cm3/g) was observed between the fresh and used catalysts. However, an increase in particle size from 4.3 to 15.7 nm of the supported NiMo occurred after the reaction (Fig. 18), probably due to the severity of the experimental conditions and due to the low heat dispersion effect under solvent-free conditions.The main active component of the supported sulfided Ni(Co)Mo/Al2O3 catalyst is the MoS2 phase wherein Ni and Co act mainly as promoters and Al2O3 acts as a dispersing medium. The unsupported form of sulfided catalysts has the advantage that they could be synthesized in different morphologies and compositions. Moreover, the unsupported MoS2 can offer more sulfur vacancies at the edge of its slabs. Unsupported sulfided catalysts were also studied in lignin depolymerization. Li et al., synthesized MoS2, and MoS2-based composite catalysts (MSx/MoS2, M = Ni, Co, Ag) with a flower morphology ( Fig. 19a-d) for lignin (corn stover) depolymerization [163]. A bio-oil yield of > 78 wt% (310 °C, 25 bar H2, 1 h) was obtained over MoS2. The performance of other sulfided catalysts (NiS2, CoS2, and Ag2S) was inferior (< 65 wt% yields) to that of MoS2. A significant improvement in bio-oil yield (>85 wt%) was obtained when the composite catalysts NiS2/MoS2, and CoS2/MoS2 were used (5 wt% of MS2). The enhancement in the catalytic activity of MoS2 when other metal sulfide components were present was explained by the Edge Decoration (ED) model. Characterization studies showed that the parent MoS2 had the typical layer structure where the edges of the layers are wedge-shaped providing the hydrogenation sites. When NiS2 and CoS2 were present, the layer structure of MoS2 became more curved and less stacked, increasing the amount of potential surface-active sites. This was in conjunction with the enhancement in the surface area of MoS2 (5 m2/g) when NiS2 and CoS2 were present (6 and 15 m2/g for NiS2/MoS2 and CoS2/MoS2 catalysts, respectively). According to the ED model, the NiS2 weakened the Mo-S bond, thereby promoting the breakage of the Mo-S bond for the generation of S vacancies (active site for hydrogenation). Further comparison of MoS2 with FeS2 and CuS showed the activity trend for bio-oil production as (250 °C, 1 h, no H2 pressure) MoS2 (82 wt%) > CuS (65 wt%) > no catalyst (53 wt%) > FeS2 (37 wt%) [164].Another unsupported sulfided catalyst reported for lignin depolymerization is VS2 where different morphologies of catalyst (sheets and nanoflowers) were compared (Fig. 19 e-h) [165]. When VS2 sheets were used, the lignin conversion was at about 77 wt% with 59 wt% yields to the bio-oil (250 °C, 20 bar of H2, 1.5 h). The solid residue amount accounted for nearly 22 wt% yields. The use of VS2 nanoflowers decreased the conversion (65 wt%) and bio-oil yield (50 wt%) and increased the solid residue amount (35 wt%). The flower morphology imparted steric hindrance to the reactant and eventually decreased its catalytic performance. According to the proposed mechanism, the H atoms from thermally broken H2 molecules adsorbed on the VS2, leading to the formation of -VH and -SH bonds. These bonds were unstable and could undergo breakage to form the VS2 catalyst, meanwhile transferring the hydrogen to the unsaturated reactant molecule.Narani et al. [166] found better results with sulfided a NiW/AC catalyst than NiMo/AC catalyst for the hydrotreatment of Kraft lignin in supercritical methanol (320 °C, 35 bar of H2, 8 h). Two fractions (methanol and dichloromethane soluble fractions) of liquid products were identified. The methanol soluble oil was composed of aromatic monomers and low molecular weight (500 g/mol) oligomers, and the dichloromethane soluble fraction was composed of solely high molecular weight (2725 g/mol) oligomers. When sulfided NiMo/AC was used, a 57 wt% yield to the methanol soluble oil was obtained. Only a trace amount of char was formed over the catalyst. The performance of sulfided CoMo/AC was inferior to that of the NiMo/catalyst (41 wt% yield to methanol soluble oil with 9 wt% yield to char), possibly due to the increase in acidity of the catalyst when Ni was replaced with Co. Sulfided NiW/AC increased the methanol soluble oil yield to 82 wt% with no concurrent char formation. The acidity of the sulfided NiW/AC catalyst was even lower than NiMo/AC and was ascribed to the reason for the increase in the product yield (18.4 versus 44.5 µmolg−1 of NH3 adsorption). Over the non-sulfided NiW/AC catalyst, substituted guaiacols were the predominant product. Sulfidation improved the activity towards deoxygenation (involving removal of -OCH3) leading to more phenols. By prolonging the reaction time from 8 to 24 h up to 35 wt% yield to monomers was obtainable. Nonetheless, this longer reaction time did not lead to over-hydrogenated compounds, highlighting the selectivity of the catalyst for phenols. The effect of different supports such as acidic ZSM-5, and basic MgO–La2O3, MgO–CeO2, and MgO–ZrO2 were also investigated for the depolymerization. Their performance was inferior to that of AC. The MgO–La2O3 (291 μmolg−1 of CO2 adsorption, the highest of all basic supports) gave similar results like NiW/AC. Different characterization techniques were used to analyze the structure and composition of the used catalysts. XRD of the spent NiW/AC catalyst showed peaks corresponding to WS2 and Ni3S2 phases, indicating its sulfided state. In support of this, the EDX analysis of the spent catalyst showed 2 wt% of sulfur on it, which was homogeneously distributed. Morphological analysis of the spent catalyst by TEM confirmed the preservation of its spherical morphology ( Fig. 20a), however, particle agglomeration was observed. The catalyst was in its active state even after a long reaction time of 24 h, indicating its high stability for longer reaction runs.Another important factor that affects the activity and stability of a sulfided catalyst during lignin hydrotreatment is the impurities in the lignin. Recently, our group has reported the role of inorganic impurities of a commercial Kraft lignin (Na, K, Ca, Fe, etc.,) on the activity of sulfided NiMo/Al2O3 catalyst [7]. These impurities in the lignin come from the Kraft pulping process which uses reagents NaOH and Na2S, and from the source wood. These inorganic impurities were deposited on the catalyst during the depolymerization, of which the major impurity element was the Na because of its higher amount in the lignin. Conditional studies using poisoned catalyst showed that at lower loadings of individual inorganic impurities, their promotor effect was prominent (the monomer yields were higher than in their absence on the catalyst, Fig. 20b). However, at their higher loadings, their poison effects were dominant. When all these elements were present together on the catalyst, their poisoning effect was much stronger. The number of moles of impurities, their strength, and their synergism were the main factors responsible for the catalyst deactivation.In summary, both supported and unsupported sulfided catalysts have been used for the hydrotreatment of various lignins. Supporting the metal sulfide active phase on various metal oxides (Al2O3, MgO–La2O3, etc.) is helpful for its high dispersion. On the other hand, the unsupported catalysts have the advantages that they can be synthesized in different morphologies and compositions to tune the activity. But, the unsupported catalysts are prone to particle agglomeration (increase in their crystallite sizes) under long heat treatments. Supported sulfided catalysts were employed to study process modifications in lignin hydrotreatments. The catalysts showed good adaptability in batch, semi-continuous, and solvent-free lignin hydrotreatments. The deactivation of the sulfided catalyst occurs mainly through the removal of lattice sulfur atoms, and through the deposition of impurities from the feedstock. The latter is a common mode of deactivation in most catalytic processes and can be solved by using impurity-free feedstocks. A crucial factor governing the stability of sulfided catalysts is the sulfidation state of the catalyst. During deoxygenation reactions, there is a high risk that the removed oxygen atoms could substitute the lattice sulfur atoms, decreasing the sulfur vacancies and sulfidation state of the catalyst. The H2 gas used in the hydrotreatment could also remove some fraction of S as H2S. In these circumstances, a simple re-sulfidation of the catalyst can regenerate its sulfidation state and restore the activity. Another method to maintain the sulfidation state of the catalyst is to provide a continuous supply of sulfur sources during the hydrotreatment. The use of lignins containing a significant amount of sulfur (e.g, Kraft lignin) has the advantage of maintaining the sulfidation state of the catalyst longer than the use of sulfur-free lignins (e.g, hydrolysis lignin).The kinetics for deoxygenation of oxygenates presented in biomass-derived oils have been studied using model compounds and real feedstocks in the currently reported literature. The subject remains important as it allows a better understanding of the reaction mechanisms, and the function of catalysts and it facilitates the upscaling of the reaction process. The approach for kinetic modeling studies, for lab-scale level research, involves the construction of mathematical expressions for the mass and heat transfer phenomena, in some cases phase equilibrium, and further develops the kinetic model based on a lumped or molecular-based approach depending on the complexity of the feedstocks. The following section discusses the kinetics of HDO reactions for triglycerides, phenolics, lignin, and biomass-derived oils over metal sulfides.A kinetic study based on a sulfided CoMo/Al2O3 catalyst for the hydrotreatment of a mixture of 10 wt% cottonseed oil with desulphurized diesel to produce renewable diesel has been reported by Sebos et al. [78] A plug flow approximation for the reactor was considered to study the kinetics of the HDO of the triglycerides of the feedstock. Overall kinetics were presented including the influence of internal mass transfer resistance. The first-order reaction kinetics was presented with the reaction rate constant ( k HDO ) calculated as: k HDO = − ln 1 − x ∙ m ̇ / m cat where x is the conversion, m ̇ is the mass feed (a mixture of 10 wt% of refined cottonseed oil in desulfurized diesel (S < 50 ppm)) and m cat is the mass of the catalyst. The experimental conversion values were plotted with the one proposed by the model to estimate the goodness of the model as shown in Fig. 21A).A pseudo-first-order lump-type kinetic model was developed by Sharma et al. to study the deoxygenation of triglycerides (jatropha oil) over mesoporous titanosilicate (MTS) supported sulfided CoMo catalyst to determine the triglyceride conversion pathway at 300 ℃ and 320 ℃ [167]. The best-fitted model showed that the triglycerides were converted not only to deoxygenated (C15 – C18) and oligomerized (> C18) products but also were directly cracked to lighter (< C9) and middle (C9 – C14) range hydrocarbons (as shown in Fig. 21B). Among all, the oligomerized product formation rate is the highest at these temperatures from the triglycerides ( Table 7).A recent study on the reaction kinetics based on the hydrodeoxygenation of stearic acid (SA) was reported by Arora et al. over a sulfided NiMo/γ-Al2O3 catalyst [59]. A Langmuir-Hinshelwood (LH) type kinetic model was developed that showed good agreement with the experimental variation of selectivities with different reaction conditions. In their work, a simplified pathway for HDO of SA is shown ( Scheme 2) and the LH type rate expressions used are presented in Table 8. The reaction scheme includes intermediates like octadecanal (C18 =O) and octadecanol (C18-OH) and explains the selectivity for the three major reaction routes (decarboxylation, decarbonylation, and direct-HDO). A single catalytic reaction site was considered here and hence an SA inhibition term is included in all the rate expressions. The presented results show that the model can predict well the increase in the conversion rate of stearic acid with the increase in temperature. The key in their work was that a phase equilibrium model was used to predict the saturation concentration of H2 in the liquid phase which depends on the reaction temperature. It was assumed in their study that the gas-liquid transport was fast compared to the rate of reaction and hence allowed the activation energies for the reactions to be predicted. The model could also predict well the conversion and selectivity with variations in residence time, pressure, and feed concentration [59].Hočevar et al. explored the kinetics of HDO using model compounds involving primary/secondary alcohol (1-hexanol), aldehyde (hexanal), methyl ester (methyl hexanoate), ether (dihexyl ether), carboxylic acid (hexanoic acid) [168]. Based on the kinetics constants and activation energies obtained from mathematical modeling and DFT calculation, it was observed that the primary alcohol is more resistant to HDO which undergoes a dehydration reaction to form ethers at the studied conditions. The secondary alcohol follows the typical path like that one highlighted in Scheme 2. Interestingly, experiments with a high initial concentration of aldehyde (hexanal) led to a parallel aldol condensation reaction (C-C coupling to C12 hydrocarbon) in addition to the deoxygenation reaction (C6). However, the products of the aldol condensation reaction were not noticed for HDO of any other model compounds owing to its very low concentration and high reactivity for hydrogenation [168].The kinetic analysis of the HDO of phenolic compounds dates back to 1987 when Gevert et al. [169] studied the reaction kinetics of 4-methylphenol HDO over a sulfided CoMo/Al2O3 catalyst and concluded that the HDO reaction of methyl-substituted phenols can occur through two independent pathways, one forming aromatic products (by direct deoxygenation of phenol, DDO) and the other naphthenic products (ring hydrogenation followed by deoxygenation of saturated or partially saturated phenols to cycloalkanes, HYD). No oxygen-containing products were detected. The rate-limiting step in path 1 has been considered as the C-O bond cleavage and that in path 2 as the hydrogenation of phenol’s aromatic ring [169].The reaction network is shown in Scheme 3 A) where (a) represents 4-methylphenol, (b) is toluene, and (c) represents methylcyclohexane plus methylcyclohexene. The authors also calculated that the adsorption constant for path 1 is twice as large as that of path 2. This indicated that the two reaction paths proceed on two different types of active sites. The poisoning effect of H2S on such HDO reactions was also studied which showed that H2S strongly suppresses toluene formation from path 1 whereas it hardly affects path 2, which supports their hypothesis that the two paths occur on two different catalytic active sites [169]. However, it was worth highlighting that authors tend to explain the deoxygenation pathways in different ways. For instance, in the study conducted by Wang et al. [112], they considered that the adsorption scheme of p-cresol on the sulfided catalyst surface differed between the deoxygenation paths.Later the effects of methyl substituents of methyl-substituted phenols on the HDO reaction over a sulfided CoMo/Al2O3 catalyst were studied by Simons and co-workers [97]. They also investigated the relationship between the relative reactivities of the methyl-substituted phenol species and the intrinsic properties of the reactant determined from electronic structure calculations. These properties include the electrostatic potential in the reactant molecules and also the electron-binding energies of various molecular orbitals. The reaction data were analyzed using Langmuir-Hinshelwood (LH) kinetics to determine the adsorption and rate constants leading to the two independent aromatic and cyclohexane paths.The following LH equations were considered here: (1) ∂ A ∂ τ = − k 1 K A A + k 2 K A A 1 + C 0 K A A n (2) ∂ B ∂ τ = k 1 K A A 1 + C 0 K A A n (3) ∂ C ∂ τ = k 2 K A A 1 + C 0 K A A n where the mole fractions of the methyl-substituted phenolic feed (A), the aromatic benzene (B), and the cyclohexane and cyclohexene (C) formed from two different pathways as discussed above [169]. K A = equilibrium constant when A is adsorbed on the catalyst surface. k 1 = rate constant for the formation of the aromatic product (DDO pathway). k 2 = rate constant for the formation of hydrocarbon products (HYD pathway). C 0 = feed concentration of A. τ = space-time variable and n = order of inhibition.The kinetic analysis of their work showed that the optimal n parameter in Eqs. (1), (2), and (3) was n = 2, which resulted in the best fit of the data. This result has been interpreted as the reactions for both pathways involving an adsorbed species and an active site. The assumption of having only one adsorption site, K A was also examined by applying two separate adsorption constants, K B and K C , for each reaction with modified Eqs. (1), (2), and (3). Their results from the regression analysis demonstrated that both K B and K C constants were identical and within the experimental errors, which is the equivalent to a single value ( K A ). Hence, a single catalytic site was considered to be present for both reaction pathways due to the same adsorption constant being calculated. Fig. 22A (a) shows the variation of the adsorption constant depending on the location of the methyl groups. The k 1 rate constant leading to the aromatic products is the lowest for phenols and highest for 3,5-DMP (Fig. 22A (b)). The k 2 rate constant for the aromatic ring hydrogenation path shows a different trend from k 1 . k 2 seems to drop significantly with the methyl group in position 2 (Fig. 22A (c)). Overall, from Fig. 22 the authors showed significant variations in the adsorption and rate constants as the location and number of the substituent methyl groups varied [97]. A correlation between the derived adsorption, rate constants, and molecular parameters is also studied in this work.A kinetic study of guaiacol (GUA) conversion over a ReS2/SiO2 catalyst using an LH kinetic model was studied by Leiva et al. [170]. They observed two different kinds of active sites for the guaiacol conversion over ReS2/SiO2 and dissociative adsorption of hydrogen was considered. The two active sites are the metal ion vacancy (M) and the sulfur stable ion (X2- ⎼M). The rate-determining step considered here was the H+ attack on the oxygen of the methoxy group. The reactions considered for the model are as follows:Adsorption of GUA: (4) GUA + M ↔ GUA − M Equilibrium K GUA Hydrogen dissociative adsorption: (5) H 2 + M + X 2 − − M ↔ H − − M + H + − X 2 − M Equilibrium K H 2 Addition of H⎼: (6) GUA − M + H − − M ↔ GUAH − − M + M Equilibrium K 3 Addition of H+: (7) GUAH − − M + H + − X 2 − − M → Ph + MeOH + 2 M Rate constant K 4 The rate expression used here is: (8) 1 r GUA = 1 k GUA C H 2 + 1 k GUA K GUA C H 2 C GUA Fig. 22B clearly shows the goodness of fit of the model (Eq. (8)) which follows a linear behavior. This result indicates the presence of two different active sites on the sulfided catalyst. The authors also showed that this kinetic model did not fit well the catalytic activity of ReOx/SiO2 (shown in the inset in Fig. 22B) suggesting that ReOx/SiO2 followed a different pathway for the conversion of GUA [170].The HDO of cresol isomers over sulfided Mo/Al2O3 and CoMo/Al2O3 was investigated by Gonçalves et al. [171]. They reported that over both catalysts the reactivity of cresols follows the order: m-cresol > p-cresol > o-cresol. The kinetic analysis of the HDO of m-cresol was studied in this work and the role of cobalt on the HDO of three cresols was investigated [171]. Scheme 3B shows that the HDO of cresols follows two independent pathways. The desired direct deoxygenation (DDO) pathway forms toluene (TOL) (strongly promoted by Co), whereas, the hydrogenation (HYD) route forms methylcyclohexene (MCHe) and methylcyclohexane (MCH) (not affected by Co). A pseudo-first-order reaction kinetic model was assumed to apply for the reactions. k DDO and k HYD are the kinetic rate constants for the DDO and the HYD pathways respectively, k HYD ′ is the intrinsic rate constant for the hydrogenation of methylcyclohexene to methylcyclohexane, C TOL , C MCHe and C MCH represent the molar concentrations of toluene, methylcyclohexenes, and methylcyclohexane [171]. (9) C CRE = C CRE , 0 . e − k HDO . τ (10) C TO L = C CRE , 0 . k DDO k HDO 1 − e − k HDO . τ (11) C MCHe = C CRE , 0 . k HYD k HDO − k HYD ′ e − k HYD ′ . τ − e − k HDO . τ (12) C MCH = C CRE , 0 1 − e − k HDO . τ − k DDO k HDO 1 − e − k HDO . τ − k HYD k HDO − k HYD ′ e − k HYD ′ . τ − e − k HDO . τ The selectivity of each product i (in mol%) can be calculated from Eqs. (10)–(12) as: (13) S i = C i C CRE , 0 − C CRE × 100 Fig. 23 shows that the experimental data points and those predicted by the model fit well for the selectivity of products as a function of the conversion of m-cresol validating the model. It has been shown from the values of the rate constants that the hydrogenation of methylcyclohexene to methylcyclohexane was 2.9 times higher over CoMo/Al2O3 when compared to Mo/Al2O3, showing that Co acts as a promoting agent improving the hydrogenating properties of molybdenum sulfide [171].Further, the promotional effect of isolated Co atoms decorated on monolayer MoS2 sheets (sMoS2) was studied by Liu et al. [172]. A kinetic study was developed for the conversion of 4-methylphenol to toluene and it was demonstrated that cobalt immobilization on the MoS2 monolayer (Co- sMoS2) showed a 34 times higher rate (396.4 ml s−1molMo −1) when compared to non-promoted sMoS2 (11.7 ml s−1molMo −1) at 30 bar and 300 °C ( Fig. 24a). The activity order shows: Co- sMoS2 > sMoS2 > FMoS2 > bulk MoS2. FMoS2 stands for few-layer MoS2. It was also shown that the incorporation of single Co atoms on the basal planes of sMoS2 facilitates the formation of more basal sulfur vacancy sites during hydrogen activation at 300 °C that enhances the activity of the Co-doped monolayer MoS2 for the HDO of 4-methylphenol. The rate of the HDO reaction was calculated considering a pseudo-first-order reaction (Eq. (14)) [172]: (14) ln 1 − x = − k C cat t k = pseudo first order reaction constant (ml s−1 mol−1). x = conversion of 4-methylphenol. C cat = concentration of catalyst under reaction system. t = reaction time (s).Recent work by Cheah et al. investigated the role of transition metals (Ni, Cu, Zn, Fe) on γ-Al2O3 supported MoS2 for the HDO of propylguaiacol (PG) [173]. In this work, the authors developed a kinetic model considering the reaction network to elucidate the reaction pathway of demethoxylation and dihydroxylation of PG. The experimental results were nicely fitted to the model, thus validating the model. Initially, the authors developed a simplified pseudo-first-order kinetic model to fit the kinetic data for the HDO of PG considering the route: A = 4-propylguaiacol → B = 4-propylphenol → C = propylbenzene → D = propylcyclohexane as shown in Fig. 24b). However, this simple kinetic model could not fit the experimental results well as shown in Fig. 24b), because it did not take into consideration any of the side reactions. Thereafter, another model was proposed taking into consideration all the main side reactions, including intermediates and reactants. The fit for the reaction kinetics for all the catalysts improved with this modified model with over a 90% coefficient of determination [173]. More importantly, their work also provided a means of evaluating how the promoters (Ni, Fe, Zn, and Cu) for MoS2/Al2O3 influenced the product selectivity for different pathways and eventually the products of the reaction with the aid of the kinetic model. Their results suggested that Ni is a promoter for the Mo catalyst while doping metals such as Fe, Zn, and Cu acted as inhibitors for the formation of deoxygenated cycloalkanes. On the other hand, both Zn and Fe had a negative impact on the HDO activity for PG but changed the selectivity towards aromatics like propylbenzene at full conversion.Due to the complex nature of lignin, a typical hydrotreatment experiment produces a broad spectrum of especially liquid phase products. The kinetic modeling for the hydrotreatment of a feedstock like lignin can be important to understand the complex series of reactions involved in the transformation of lignin, how they are influenced by operating conditions and catalyst properties, and eventually aid in an effective scale-up of a lignin hydrotreatment process. Pu et al. developed a kinetic model for lignin hydrotreatment over a CoMoS-supported catalyst in a semi-batch reactor using a lumped approach [174]. There were three main lumps divided further into product groups: lignin oligomeric residues (solid): (i) THF-insolubles, THF-solubles, and solubilized oligomers; (ii) liquid product lumps: dimethoxyhenols, methoxyphenols, alkylphenols, catechols, alkanes ( < C13), alkanes ( ≥ C13), aromatics, naphthenes, and H2O; (iii) gas products lumps: CO2, CO, CH4 and C2-C6 (light hydrocarbon) with v i j , the overall stoichiometric coefficient for component i in reaction j ( Scheme 4). The model accounted for gas hydrodynamics which was characterized by Residence Time Distributions (RTD), liquid-gas mass transfer resistance, and vapor-liquid equilibrium effects with reactions (1) to (10) from Scheme 4. This resulted in a model that fitted well with the obtained experimental data [174]. The model results were able to well describe the main overall lignin depolymerization reactions and further deoxygenation reactions of phenolic monomers in the liquid phase [174].Reaction (1): THF-insolubles → k 1 v TSB 1 ·THF-solublesB.Reaction (2): THF-solublesA + v H 2 2 ·H2 → K 2 v TSB 2 ·THF-solublesB +  v C H 4 2 ·CH4 +  v H 2 O 2 ·H2O +  v C 2 C 6 2 ·C2 – C6.Reaction (3): THF solublesB +  v H 2 3 ·H2 → K 3 v SO 3 ·Solubilized oligomers +  v C H 4 3 ·CH4 +  v H 2 O 3 ·H2O +  v C 2 C 6 3 ·C2 – C6 +  v AP 3 ·Alkylphenols +  v AK 1 3 ·Alkanes (< C13) +  v AK 2 3 ·Alkanes (≥ C13).Reaction (4): Solubilized oligomers → K 4 v AP 4 ·Alkylphenols.Reaction (5): Dimethoxyphenols +  v H 2 5 ·H2 → K 5 Alkylphenols + 2·H2O + 2·CH4.Reaction (6): Methoxyphenols +  v H 2 6 ·H2 → K 6 Alkylphenols + H2O + CH4.Reaction (7): Methoxyphenols +  v H 2 7 ·H2 → K 7 Catechols + CH4.Reaction (8): Catechols +  v H 2 8 ·H2 → K 8 Alkylphenols + H2O.Reaction (9): Alkylphenols +  v H 2 9 ·H2 → K 9 Aromatics + H2O.Reaction (10): Alkylphenols +  v H 2 10 ·H2 → K 10 Naphthenes + H2O.Grilc et al. screened a series of catalysts covering the commercial NiMo catalysts in sulfided, oxide, and reduced form and other catalysts in the hydrotreatment of a solvolyzed biomass-derived oil [175]. A complex reaction pathway was proposed and followed by constructing a lumped kinetic model based on the quantified functional groups by Fourier transform infrared spectroscopy (FTIR) [175]. Among all the tested catalysts, the commercial sulfided NiMo catalyst was found to be suitable for yielding bio-oils with high gross calorific value. The authors also discovered that the unsupported bulk MoS2 resulted in high HDO activity and selectivity which is worth further investigation [175]. The same authors then extended their work by comparing the selectivity and activity of several synthesized and commercial unsupported Mo catalysts in oxide, carbide, and sulfide form, and also unsupported WS2 nanotubes [176]. A similar lumped kinetic model was developed giving apparent kinetic constants that correspond to main reactions like hydrodeoxygenation (k1), decarboxylation (k4), decarbonylation (k3), dehydrogenation (k2), and hydrocracking of a solvolytic oil as shown in Scheme 5a) [176]. One of the observations was that the urchin-like MoS2 possessed the highest k1 value among all other unsupported sulfided catalysts which corresponds to the removal of the hydroxyl group in the form of water. While the decarbonylation, decarboxylation, and hydrocracking reactions occurred to a lesser extent using the unsupported materials which could be explained by the absence of the use of an acidic catalyst support that can cleave the C-C linkages [176].Grilc et al. also studied the simultaneous liquefaction and hydrotreatment of biomass (Sawdust samples like beech, fir, and oak) over a sulfided catalyst (NiMo on alumina), a reduced Pd on alumina catalyst, and zeolite Y. Emphasis was placed on studying the effect of different process parameters like time, pressure, temperature, wood, and solvent type. The yield and product composition from the simultaneous reactions were identified and correlated to a lumped kinetic model accounting for liquefaction, decarboxylation, decarbonylation, HDO, and char formation reactions as shown in Scheme 5a). Their modeling results showed that reaction temperature played an important role in the liquefaction and HDO of biomass. The increase in reaction temperature from 300° to 350°C resulted in a 2.5-fold higher yield for HDO products, while the solid residue yield decreased by 39%. However, when the reaction temperature is increased over 350 °C, a lower oil yield was achieved which was mainly attributed to an increased formation of char [177]. Also, in their work, sulfided NiMo on alumina was found to achieve higher oil yield as compared to the noble metal Pd on alumina.Common pathways for catalyst deactivation include poisoning, coking, sintering, fouling/physical blockage, leaching or vapor formation, and solid-state transformations [178]. Based on the literature, it can be deduced that metal sulfide catalysts used for upgrading renewable feedstocks can lose activity due to loss of sulfur, impurities present in renewables, evolved products, coking, and sintering of the active phase [48,179–182].Since bio-based renewables have high oxygen contents, often sulfides catalyst loses sulfur through a sulfur-oxygen exchange during HDO. Such an exchange more easily takes place on sulfur edges over unpromoted MoS2 than the promoted sites (in the case of CoMoS) in the presence of a high H2O partial pressure as observed via combined CO adsorption and IR studies during HDO of 2-ethyl phenol [179]. The authors also demonstrated via DFT that in the presence of a large amount of water the exchange is stronger and irreversible over MoS2 while Co promotion makes the catalyst more water tolerant and enables the poisoning to become more reversible. Hence, a continuous supply of sulfiding agents (e.g., H2S) in the feed at a sufficiently low concentration can restore the catalytic activity, while an inhibition can occur at a higher H2S concentration [183]. Resulfidation of the catalyst can also restore its initial activity as has been demonstrated for a sulfided CoMo/Al2O3 catalyst while hydroprocessing 2-hydroxydiphenylmethane (250 °C, 155 bar, WHSV = 0.49 h−1) [184]. It is often criticized that the addition of such agents will however contaminate the product oils. Since TMS catalysts, e.g. Ni/Co-promoted Mo/W sulfides are also highly active in hydrodesulfurization, the final product typically contains only traces of sulfur.Bio-oil impurities depend on their source of production and prior pretreatment processes. Typical impurities include alkali, alkaline-earth metals (Na, K, Ca, Mg, etc.), phosphorus, sulfur, and nitrogen. Irreversible K deactivation of NiMoS2/ZrO2 (K impregnated as KNO3 to the catalyst at a K/(Ni+Mo) molar ratio of 1) was attributed to the preferential occupation of edge vacant sites of the promoted MoS2 during HDO of phenol and octanol [180]. Trap grease phospholipids also containing alkali metals were shown to cause severe deactivation of a commercial CoMoS/γ-Al2O3 catalyst via coking and severe pore plugging while upgrading rapeseed oil [185]. Fe was found to preferentially block the Ni-promoted sites in NiMoS/γ-Al2O3 while upgrading fatty acids [48]. Bio-oil phospholipids having phosphate and choline moieties have also been shown to lower the activity of NiMoS/γ-Al2O3 during HDO of oleic acid [49]. Nitrogen-bearing compounds (e.g. amines, pyridines, quinolines, etc.) have been shown to cause the deactivation of TMS catalysts [186–188].The presence of water in bio-oil or produced during HDO may influence the activity of TMS-based catalysts via oxidation of the sulfide phase or deterioration of the structure of the active phase [65]. Couman and Hensen et al. [52] reported that water had little influence on sulfided NiMo/ γ-Al2O3 below a concentration of 5000 ppm during HDO of fatty esters. Krause and co-workers reported that the inhibition effect of water can be compensated for by H2S during HDO of an aliphatic ester (methyl heptanoate) [65]. The decarboxylation route has been reported to be affected by H2O perhaps via keto-enol isomerization during HDO of fatty esters [52]. Support material used can also be affected by the water e.g. γ-Al2O3 reportedly transforms to its boehmite phases or poisons the acidic sites [179]. The interaction of water with active sites may also lead to the formation of an inactive sulfate layer [179]. However, water may affect the HDO of phenolic compounds over metal sulfides to a varying extent [64,179]. The presence of a small amount of water (water/p-cresol molar ratio < 1) was found to increase the direct-HDO route of p-cresol to toluene while a high amount of water (water/p-cresol molar ratio > 1) significantly reduces the deoxygenation rate and toluene selectivity due to preferential occupancy of active sites [124]. Additionally, CO formed during HDO under reduction conditions can strongly inhibit the direct-HDO pathways owing to the lower adsorption energy of CO over metal sulfides (e.g., MoS2, CoMoS) [183,189].Coking is one of the leading challenges to deal with for hydrotreating catalysis. TMS-based HDO catalysts may deactivate via carbonaceous deposits (reactive/soft or refractory/hard) formed either by physical or chemisorbed processes [190]. Such deposits arise from the undesired side reactions (e.g. cracking, aromatization, dehydrogenation, cyclization, condensation, etc.) involving adsorbed species/precursor molecules (alkenes, aromatics, oxygenates, etc.). Condensation and rearrangement reactions typically play a major role in low-temperature coke deposition (<200 °C) while dehydrogenation and hydrogen transfer reactions lead to the formation of polyaromatic hydrocarbons (>350 °C) [191]. On the other hand, carbon species can be bonded to the active site via weak interaction [192] or can partially replace sulfur atoms at the edge sites of MoS2 in the form of a Mo-S-C bond in a MoS2−xCx phase [193]. The latter can be synthesized via the thermal treatment of MoS2 with a mixture of dimethyldisulfide in N2/H2 which is reported to stabilize the MoS2 crystallites with a smaller size (i.e., carbon species restricts the crystal growth) with lower stacking, thus enhancing the activity instead [194]. In addition, metal carbides (Co-C and Mo-C bonds) may form which may be observed by EXAFS analysis [192].Catalyst/support combination is another critical parameter in determining the activity, selectivity, and coke formation. Unsupported TMS like MoS2 may undergo agglomeration [181,182] under reaction conditions which can be prevented in the presence of hydrogen and hydrocarbon feed [193]. It is important to note that sintering or agglomeration may occur through Ostwald ripening or coalescence to obtain higher thermodynamic stability which lowers the surface area and catalytic activity [195]. Segregation of the promoter (e.g. Ni/Co) [196,197] as sulfides (Co9S8/Ni3S2) may influence the catalyst deactivation [182]. Acidity/basicity of the active phase or support materials can also play an important role [198] in aiding coke-forming reactions. Both Lewis acid sites (LAS) have a high affinity for basic precursors and Brønsted acid sites (BAS) via proton donation and carbonium cation formation may contribute to coke formation. BAS, in addition, promotes coupling/isomerization reactions [199]. The deactivation of MoS2 by phenolic compounds is indirect. The basicity of the phenolic compounds and their substituent nature gives different adsorption mechanisms as in Scheme 6a). Due to their basicity, they adsorb as phenolates on the acidic alumina support. If they are adsorbed closely to the MoS2, they could block the accessibility of other reactants for deoxygenation to the MoS2 active site, resulting in the deactivation of the whole catalyst for deoxygenation reactions [200]. In addition, the oxygenates that adsorbed on the sulfide phase reduce the active phases and cause catalyst poisoning [201]. Thermal instability and repolymerization reactions of phenolic compounds in bio-oil derived from fast pyrolysis and lignin lead to premature deactivation of the catalyst [202]. Recently, Kraft lignin impurities (Na, K, etc.) have been demonstrated to deactivate a sulfided NiMoS/Al2O3 catalyst especially at high loading as discussed above in Section 4.4.2 [7].With a representative model compound, 2-hydroxydiphenylmethane, and using a sulfided CoMo/Al2O3 catalyst, a decline in the catalytic activity was noticed after 20 h of reaction (250 °C, 155 bar, WHSV = 0.49 h−1). But, a simple re-sulfidation of the catalyst restored its initial activity. The mode of deactivation of the CoMo/Al2O3 catalyst could be ascribed due to the loss of sulfur. Another interesting aspect of the CoMo/Al2O3 catalyst was revealed when 4-methyl guaiacol was used as the model compound. The char formation from this experiment was around 40 wt%, far less than the non-catalytic thermal reaction indicating that the CoMo/Al2O3 catalyst had generated products with less char forming tendency. This observation is somewhat contrary to the previous reports where more char was formed in the presence of the catalyst [153]. However, the role of temperature/pressure in char formation could not be neglected (vide supra). The deactivation of TMS catalysts while upgrading lignin/lignin-derived phenolic compounds has been discussed in Section 4 . Apart from the experimental approaches described in the previous section, computational approaches based on density functional theory (DFT) calculation, can also provide a better insight into the physiochemical properties of the sulfided-based catalytic materials and their relationship with their observed reactivity in deoxygenation. The deactivation mode of the metal sulfides can also be understood through DFT calculation. For instance, recent work by Liu et al. investigated the detailed mechanism of the in situ and ex situ substitution of sulfur atoms in the active phase (Co(Ni)MoS edge) by oxygen atoms under the presence of water via the DFT [203]. The calculation was also performed for the energy change, Gibbs free energy evolution, and reaction coordination to gain an understanding of the water deactivation of Co(Ni)MoS under HDO conditions [203]. It was outlined in their work that, for in situ oxygen substitution, the oxygen atoms from water molecule occupied the positions of the edge sulfur atoms; while for the ex situ substitution, the water molecules occupied directly the unsaturated sites after the desorption of edge sulfur as hydrogen sulfide [203]. It was found that the substitution of sulfur atoms depends on the ratio of partial pressure between hydrogen and water, and also between hydrogen and hydrogen sulfide. Moreover, it was found that the CoMoS is more prone to water deactivation than NiMoS, and when compared to MoS2, both promoted Co(Ni)MoS showed better water resistance. Another recent study by Diao et al. used DFT calculations to demonstrate and validate that the Co-doped MoS2 and Mo-doped Co9S8 enhanced the vertical adsorption of oxygenates which further undergoes C-O cleavage of diphenyl ether (DPE) and were also able to avoid benzene ring hydrogenation [204]. It was further shown that the Mo-doped Co9S8 surface promoted the adsorption and C-O bond activation in DPE. Besides, there are also excellent reviews on the summary and studies of the application of the computational approach to study other catalyst systems such as transition metal phosphides (TMP) [205], transition metal catalysts [206,207], and transition metal sulfides (TMS) [208]. With this in mind, there is still a need to engage in theoretical studies that can disclose the reaction mechanisms in the complex catalytic reaction and strive to improve the existing catalyst systems, eventually aiding the selection and implementation of these catalysts into the complex refinery.Recently, the European Commission (EC) has published its ambitious legislative package ‘Fit for 55 targeting a 55% reduction in GHG emissions by 2030 compared to the 1990 level [209]. The presented package provides tools to tackle the climate crisis, and different solutions taking sustainability into account must be endorsed and applied. Advanced biofuels play a key role in decarbonizing the transport sector. Therefore, many research efforts have been dedicated to the development of heterogeneous catalysts for application in advanced biofuel production in the past few decades. Metal sulfides remain the core catalysts in hydroprocessing industries as they are effective in the removal of heteroatoms such as sulfur, nitrogen, oxygen, halides, and metals. This review has emphasized the use of hydrotreating catalysts, metal sulfides in the valorization of triglyceride feeds, oxygenates in monomer and dimeric form, biomass-derived pyrolysis oil, and lignin feed. Besides the type of catalysts, the catalytic performance and progression of reactions during hydroprocessing also depend largely on the reaction parameters like reactor type, reaction temperature, pressure, residence time, and solvent system. These aspects have been discussed in this work.The major challenges associated with the hydroprocessing of various renewable feedstocks and the possible future research and development in respective areas are listed in Table 9. Numerous research papers report the use of alumina as a catalyst support for hydroprocessing catalysts due to its good textural and mechanical properties, and low relative cost [210,211]. The acidic nature of alumina is found to be beneficial in breaking the C-O bond in anisole which is also found in the lignin structure [210]. The sulfur vacancies located at the edges of the metal sulfides act as unsaturated sites and Lewis acids sites and are found to be active in cleaving C-O linkages [44]. Other supports like silica and activated carbon were also used as supports for NiMo hydrotreating catalysts and studied in vacuum residue hydrotreating reactions [211]. Carbon as catalyst support has also gained attention owing to its high surface area, inert nature, high thermal stability, and low cost [212]. An important conclusion made is that the effectiveness of the hydrotreating catalysts depends on the pore size diameter, pore-volume, and metal dispersion that ultimately improves the efficiency of hydroconversion. As discussed in this work, the unsupported or self-supported metal sulfides have also gained interest because of their higher activity per catalyst mass as compared to the supported sulfide catalysts [208]. In addition, the direct use of the active metal sulfides phase allows the elimination of the transport resistance interference due to the support during the reaction. With this, ExxonMobil and Albemarle Catalysts developed an unsupported catalyst by so-called NEBULA technology that claims to show superior activity as compared to the conventional hydrotreating catalysts [213,214]. Another great example is the Eni Slurry Technology (EST) process which uses highly dispersed MoS2 nanoparticles and has proven the feasibility of using unsupported catalyst materials in hydrotreating [215]. Apart from developing stable, active, and cost-effective metal sulfide catalysts, the issue related to sulfur leaching and replenishment when using sulfided catalysts remains a central research topic. Some studies have been communicated in this regard exploring the benefits of applying sulfiding agents to compensate for sulfur loss during the process [14,62,66]. There is also a need and an interest in the research community in designing metal sulfide catalysts and also understanding the mode of sulfide deactivation aided by DFT tools. A better understanding of the physicochemical properties of metal sulfides aided by the first principles approaches can eventually benefit the tailored synthesis of metal sulfides and improve desired product selectivity. A review by Raybaud in 2007 and extended by others provided insights into the understanding of the sulfide active phases, the localization and role of promoters, electronic properties, and morphological changes influenced by the operating parameters, synthesis methods, or addition of promoter [10,12,216–220].In addition, when dealing with hydrotreating of complex feedstocks like lignin, the diffusion of depolymerized lignin oligomeric fragments into the catalyst pores to access active sites is likely to be limited by pore transport resistance, and therefore, the self-supported sulfide catalysts may be seen as beneficial to gain better active site accessibility. Depolymerized lignin fragments that do not undergo deoxygenation reactions due to the inaccessibility of active sites, may instead repolymerize to form solid residues like char, which is usually undesirable. Studies related to process improvement could also be another way to ensure an efficient hydroconversion of solid lignin. A recent example shows that a modification to a semi-batch reactor operating mode by injecting a lignin slurry into a reactor that has reached the desired reaction temperature can effectively avoid the repolymerization and recondensation reactions resulting in better lignin conversion [221]. Supported and unsupported versions of NiMoS catalysts was also reported effective in this regard [222,223].The hydroprocessing of various renewable feedstocks such as triglycerides, model compounds for bio-oil, and lignin-derived oils are discussed in this work. There are different challenges involved while using these different feedstocks for the scale-up of a hydrotreatment process. One of the common challenges when dealing with bio-feedstocks is the deactivation of the catalyst caused by the presence of inorganic impurities in the feedstocks. These inorganic elements can act as poisons to the catalytic sites, causing a decrease in the catalyst lifetime during the time-on-stream. Future research should focus on understanding the role of these impurities on the catalytic activity of a typical hydrotreating catalyst and also the in-depth deactivation mechanism. For instance, one recent study revealed that low concentrations of impurity elements like Na, K, Ca, and Fe promotes the deoxygenation ability of a NiMoS/Al2O3 catalyst. However, when they are present in higher concentrations, these impurities are deposited on the catalyst, ultimately leading to the poisoning of the catalyst [7]. More of these types of studies should be pursued using different bio-feedstocks as they can be seen as highly relevant for operation in refineries in terms of the stability of the catalysts and also it could possibly provide a way to regenerate, recycle and reuse the catalyst. Issues like catalyst pore plugging due to coking and inorganic impurities present in bio-feed should also be addressed and investigated in the future by studying different guard bed materials like catalysts or adsorbents to improve the catalyst lifetime. Research related to catalytic material development that is more resistant to deactivation and can be easily restored following deactivation is of high interest. In addition to these, the pretreatment of these bio-feedstocks for the removal of inorganic elements that are responsible for the catalyst deactivation is required to better improve the properties of the feedstock. The pretreatments and enhancement methods to improve the quality of the feedstocks is desirable to achieve efficient refining of bio-feedstocks. The removal of nitrogen content in bio-feedstocks also remains an area that is less explored and requires more attention. For instance, pyrolysis oil derived from sewage sludge contains high nitrogen and sulfur-bound polyaromatics compounds which reduces the quality of the product fuels and also generates toxic emissions upon combustion.To sum up, we have comprehensively reviewed the use of industrially-relevant metal sulfide catalysts for the upgrading of biomass feedstocks like triglycerides, monomeric and dimeric phenolic compounds, pyrolysis oil, and waste lignin. Various aspects such as sulfide deactivation, reaction kinetics, and mechanisms have been discussed. The challenges and future research opportunities concerning the efficient upgrading of bio-feedstocks to liquid fuel were explored. Metal sulfides will remain as a core in the processing of renewable feedstocks in existing refinery infrastructures and both the research community and industries play a significant role in realizing future biorefineries.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This research is a collaboration work between Chemical Engineering, Competence Centre for Catalysis (KCK) at Chalmers, Preem AB, and RISE Research Institutes of Sweden Energy Technology Center (ETC). The authors would like to acknowledge the Swedish Energy Agency (2017-010890 and 2018-012459) and Preem AB for financial support.

Human activities such as burning fossil fuels for energy production have contributed to the rising global atmospheric CO2 concentration. The search for alternative renewable and sustainable energy sources to replace fossil fuels is crucial to meet the global energy demand. Bio-feedstocks are abundant, carbon-rich, and renewable bioresources that can be transformed into value-added chemicals, biofuels, and biomaterials. The conversion of solid biomass into liquid fuel and their further hydroprocessing over solid catalysts has gained vast interest in industry and academic research in the last few decades. Metal sulfide catalysts, a common type of catalyst being used in the hydroprocessing of fossil feedstocks, have gained great interest due to their low cost, industrial relevance, and easy implementation into the current refining infrastructures. In this review, we aim to provide a comprehensive overview that covers the hydrotreating of various bio-feedstocks like fatty acids, phenolic compounds, pyrolysis oil, and lignin feed using sulfided catalysts. The main objectives are to highlight the reaction mechanism/networks, types of sulfided catalysts, catalyst deactivation, and reaction kinetics involved in the hydrotreating of various viable renewable feedstocks to biofuels. The computational approaches to understand the application of metal sulfides in deoxygenation are also presented. The challenges and needs for future research related to the valorization of different bio-feedstocks into liquid fuels, employing sulfided catalysts, are also discussed in the current work.

The increasingly strict environmental regulations on exhaust emissions of gasoline-powered transportation vehicles are compelling refineries around the world to produce gasoline with lower sulfur content [1–4]. S-Zorb gasoline adsorption desulfurization technology [5–10] is a typical ultra-low sulfur gasoline production technique, which has achieved a large-scale industrial application. However, S-Zorb technique still faces the problem of loss of octane number, due to the hydrogenation of the olefins present in gasoline.Aromatization and isomerization reactions are two effective approaches to recover or even enhance the octane number of gasoline because they can transform paraffins, linear olefins and naphthenes to higher octane compounds, such as isoparaffins, isomeric olefins and aromatics. Naphthenes are important components of fluid catalytic cracking (FCC) gasoline, representing 5%–10% (by volume fraction) of FCC gasoline fraction. Naphthenes are primarily distributed between C6 C8, so methylcyclohexane is a representative of naphthenes in FCC gasoline fraction. Transformation of naphthenes to aromatics and isomerization products through dehydrogenation aromatization as well as isomerization of alkanes can greatly improve the gasoline RON [11–13]. Generally, dehydrogenation of naphthenes is complex because a series of competing reactions, such as ring contraction, ring opening, cracking, hydride transfer and isomerization, can occur simultaneously. Although dehydrogenation aromatization and isomerization of naphthenes can be accomplished over either acid [14,15] or metal catalysts [16,17], the catalytic activity of their combination is better than any single catalyst due to the synergetic effect of acid and metal [18,19].The typical catalyst for these processes is a bifunctional heterogeneous catalyst consisting of a noble metal supported on an acidic support [20–22]. The catalytic hydro-conversion of cyclohexane over Pt/HY catalysts were studied by Onyestyák et al. recently [19]. Gopal et al. demonstrated that the maximum isomer yield for a Pt/H-zeolite catalyst could be obtained when the metal and acidic functions of the catalysts were well-balanced [23]. Choudhury et al. noted that the hydroisomerization yield could be enhanced by the elimination of strong acid sites from the micropores [21]. Belatel et al. investigated the MCH reaction on PtIr/sulfated zirconia catalysts and showed that no isomerization was found to take place in the absence of metals [13]. However, the scarcity and high cost of noble metal has impeded their widely application in the aromatization and isomerization of gasoline. In this work, we studied the non-noble metal Ni supported on different supports for transformation of methylcyclohexane (MCH). Fe ion-exchange ZSM-5, ZSM-5, alumina and silica were utilized to illustrate the effect of acidity and support interaction on the dehydrogenation, isomerization and cracking of MCH. To investigate whether these RON recovery catalysts can be integrated into the S-Zorb process, the reactions were carried on the S-Zorb reaction condition (1.5 MPa H2 and 673 K).The raw material ZSM-5 was supplied by Shanghai ShenTan Environmental and Advanced Materials Corporation. Inert silica and γ-Al2O3 were purchased from Evonik Degussa (China) Co., Ltd and Across respectively. Fe-ZSM-5 samples were prepared through ion-exchange of ZSM-5 with iron nitrate solution (Aldrich Chemical, > 99.99% pure) at 343–353 K for 2 h. Then the resulting samples were filtered, washed with distilled water, dried at 423 K for 12 h, and then calcined in air at 873 K for 3 h. In order to investigate the effect of molecular sieve structure and acid properties on catalytic performance, the inert silica and γ-Al2O3 were used for reference supports. The catalysts were prepared by spraying a saturated solution of nickel nitrate on the above four supports and then dried at 423 K for 3 h and calcined at 873 K for 1 h. The obtained samples were designated as NiO/ZSM-5, NiO/ZSM-5-Fe, NiO/SiO2, and NiO/Al2O3, respectively.The crystal structure of catalysts was characterized by X-ray diffraction (X'Pert SW, SIEMENS) using Cu Kα radiation operated at 40 kV and 40 mA. Diffraction lines of 2θ from 5° to 70° with a scanning speed of 5° min−1 were taken to determine the crystalline phase of the catalyst.The chemical compositions of the catalysts were measured using a X-Ray Fluorescence spectrometer from Rigaku Corporation.Nitrogen sorption measurements were carried out over a Quantachrome Autosorb-6B unit. The isotherms were measured at 77 K after degassing samples below 1.3 Pa at 573 K for 8 h. The BET specific surface area was estimated using adsorption–desorption data as per the ASTM 4365 standard applicable for microporous materials. Total pore volume was equal to the amount of N2 adsorbed at a relative pressure of 0.99.NH3 temperature-programmed desorption (NH3-TPD) experiments were conducted on a Micromeritics AutoChem II 2920 analyzer equipped with a thermal conductivity detector (TCD) to determine the density and strength distribution of the acid sites. Typically, a 0.1 g sample was housed in a quartz U-shaped tube and pretreated in flowing helium (50 mL min−1) at 823 K for 1 h. After the pretreatment, the sample was cooled down to 393 K and ammonia-saturated in a stream of 10% NH3/He flow (50 mL min−1) for 0.5 h. Subsequently, the physically adsorbed NH3 was removed by flowing helium (50 mL min−1) at 393 K for 1 h. Finally, the chemically adsorbed NH3 was desorbed, with the temperature of the sample being raised from 393 to 823 K at a heating rate of 10 K min−1 and maintained at 823 K for 10 min.The acidity properties of samples were determined using the pyridine FT-IR method, which was carried out on a NICOLET 6700 Fourier-transform infrared spectrometer (Thermo Fisher Scientific Corp.). Self-supporting wafers (13 mm in diameter) were made from ca. 10 mg of catalysts. The sample was initially evacuated to 1.0 × 10−3 Pa at 723 K for 2 h and then cooled to 363 K to be saturated with pyridine for 5 min. Then, pyridine desorption was performed under vacuum for two consecutive periods (0.5 h each) under isothermal conditions at 473 and 623 K, followed by IR measurements. Infrared spectra were measured at a 4 cm−1 resolution. The sample was examined in the range of 1350–1800 cm−1. For Py-IR spectra, the bands located at 1540 cm−1 and 1450 cm−1 can be assigned to pyridine adsorbed on Brönsted (B) and Lewis (L) acid sites, respectively [24,25]. Total B acid sites and total L acid sites, and medium and strong B acid sites and medium and strong L acid sites can be obtained from the Py-IR measurement results at 473 and 623 K, respectively. The quantitative calculation of B and L acidity by Py-IR analysis is based on the integrated Lambert–Beer Law [26]: C SW  = AS/mε where C SW (μmol g−1) is the concentration of B or L acid sites in reference to a unit weight of dry sample, A (cm−1) is the integrated absorbance, S (cm2) is cross sectional area of the sample wafer, m (g) is the weight of the dry sample, and ε is the integrated molar extinction coefficient determined by Corma et al. (εB = 0.059 + 0.004 (cm2 μmol−1)*A, εŁ = 0.084 + 0.003 (cm2 μmol−1)*A) [27].Temperature-programmed reduction (TPR) was measured with the same apparatus as that of NH3-TPD. Prior to the TPR experiments, the catalysts were dried in flowing He at 773 K for 1 h. A mixture of 10% of H2/Ar was used as the reducing gas at a flow rate of 30 mL min−1. The rate of temperature rise in the TPR experiment was 10 K min−1 up to 923 K.Adsorption properties of MCH on different samples were determined by MCH pulsed adsorption and TPD using the same apparatus as that of NH3-TPD. Approximately 20 mg of catalyst was pretreated to remove water and reduce the supported nickel oxide. Then, pulses of MCH were injected repeatedly until the TCD signal showed no further adsorption. Following the pulsed adsorption experiments, the system was purged for 1 h in a He flow (30 mL min−1) to remove residual MCH. Subsequently, TPD of MCH from 373 to 823 K was performed at a heating rate of 10 K min−1.MCH dehydrogenation reaction was performed in a bench-scale high-pressure fixed bed reactor with a 25 mm inner diameter. Activity tests were carried out with 16 g of the pelletized catalyst, particle sizes ranged from 0.10 to 0.30 mm. Prior to activity test, the catalyst was activated by in situ reduction in a flow of H2 (1.5 MPa) at 673 K for 1 h. Then, the temperature was maintained at 673 K. The experiment was carried out in a mass space velocity 5 h−1 for 12 h with sampling interval of 2 h. The effluent from reactor was condensed, and then the liquid samples were taken and analyzed using a GC with an OV101 capillary column (30 m) and an FID.The textural properties of the catalysts were obtained by analyzing the BET surface areas and total pore volume (Table 1 ). As shown in Table 1, the BET areas follow the trend: NiO/ZSM-5-Fe > NiO/ZSM-5 > NiO/SiO2 > NiO/Al2O3. The reference sample NiO/SiO2 presents a similar pore volume as the two ZSM-5 supported samples. All of the samples have the same content of nickel oxide, which implies that the method of introducing nickel by spray impregnation is repeatable.N2 adsorption isotherms of the four samples are shown in Fig. 1 . The isotherms of ZSM-5 supported samples belong to a combination of type I and IV patterns with H4 hysteresis loops, which are indicative of micropores and mesopores according to the IUPAC classification [28]. The shape of the hysteresis loop relates to the shape of meso- and macropores, namely, the horizontal loop implies ink bottle-shaped pores, whereas a vertical hysteresis loop implies cylindrical pores [29]. With Fe modification, the hysteretic loops change slightly from horizontal to vertical, suggesting a portion of the pores shifts to cylindrical pores connected to the external surface, which is favorable for the diffusion of molecules [30]. The reference sample NiO/SiO2 and NiO/Al2O3 exhibited type IV patterns, corresponding to a mesoporous structure. In addition, the hysteretic loops of these two samples seem to be of type H3 of IUPAC classification, which is associated with porous solid having slit-shaped pores [31].XRD patterns of NiO/ZSM-5, NiO/ZSM-5-Fe, NiO/SiO2 and NiO/Al2O3 are illustrated in Fig. 2 . All of the samples display distinct diffraction peaks of NiO phase at 2θ = 37.0°, 43.1° and 62.8°, representing the existence of crystalline phase nickel oxide on the catalyst. The XRD patterns of ZSM-5 supported samples show the intact MFI zeolite structure, which has 2θ values of characteristic diffraction peaks of approximately 7.9°, 8.9°, 23.3°, 23.9° and 24.4°. This indicates that the ferric nitrate treatment does not significantly damage the long-range order of the original zeolite framework. XRD patterns of NiO/ZSM-5-Fe show much weaker intensity of NiO diffraction peaks, suggesting that the active phase is dispersed better than other samples. In addition to the diffraction peaks of NiO phase, NiO/SiO2 exhibits a broad peak centered at 2θ = 22.0°, which is the characteristic peak of amorphous SiO2, and diffraction peaks of NiO/Al2O3 at 2θ = 37.4°, 46.07° and 66.9° are related to alumina phase.In order to investigate the acid properties of the synthesized samples, the NH3-TPD experiment was carried out. The profiles are presented in Fig. 3 . Usually, NH3 desorption is temperature-dependent and can be classified in three stages, viz.: weak (< 473 K), moderate (473–723 K) and strong (> 723 K). The area of a specific peak corresponds to the amount of desorbed NH3 and can be taken as the standard to quantify the acid amount [32]. The total acid amounts are reported in Table 2 . The acid amounts were found to follow the order: NiO/ZSM-5 > NiO/ZSM-5-Fe > NiO/Al2O3 > NiO/SiO2. The NH3-TPD profile of NiO/SiO2 displays no peak of NH3 desorption, which is in accordance with the Pyridine FTIR result (Table 2), while NiO/Al2O3 possessed moderate acid amount owing to the surface hydroxyl ions of alumina. It implies that introducing nickel by spraying onto inert silica cannot create acid sites. The profiles of the NiO/ZSM-5 sample exhibit one main peak centered at 480 K, which is attributed to NH3 bound to weak acid sites. The tailing phenomenon suggests that moderate and strong acid sites are present with much less amounts. Interestingly, two distinct NH3 desorption peaks are observed for the profiles of NiO/ZSM-5-Fe. One centered at 470 K corresponds to weak acid sites and the other at 645 K to moderate acid sites. It can be deduced that iron exchange decreased the acid amounts but enhanced the acid strength. Moreover, the decreased acid amount can be ascribed to the interaction between acid sites and Fe species [33].To further probe the strength and the nature of the acid sites (L vs. B), FTIR analyses of pyridine adsorbed catalysts were carried out at 473 and 623 K, respectively. The results are listed in Table 2. Silica shows no acidity. Alumina contains mainly L acidity. Meanwhile two ZSM-5 supports have both L and B acid sites. The acidity results at 473 K involve all acidic sites, but only acidity with moderate and strong strength is included in the results of 623 K. From Table 2 and Fig. 3, it is found that ZSM-5 supported sample has the highest L and B acidic amounts in which weak acid sites are dominated. On the contrary, Fe modified ZSM-5 sample has the most L and B sites with moderate acidic strength.The TPR characterization was carried out to find out information about the interaction between the Ni active metal and supports. TPR patterns of oxide catalysts are shown in Fig. 4 . All four samples have an obvious peak with Tmax identified at 650–680 K, which was attributed to the reduction of NiO particles. And the peak temperature of NiO reduction is in the order of NiO/ZSM-5-Fe < NiO/ZSM-5 < NiO/SiO2 < NiO/Al2O3. This reflects the trend of interaction between NiO and support. NiO/Al2O3 shows a wider reduction peak with the highest temperature at approximately 680 K, which corresponds to the expected reduction temperature for NiO in strong interaction with alumina supports [34]. The shoulder peak detected at around 780 K appears to be due to the reduction of NiAl2O4 spinel. The reduction of NiO/SiO2 occurred at temperature around 660 K, suggesting the decreased interaction of Ni with support. The reduction peak of the ZSM-5 supported samples shifts to lower temperatures (approximately 650 K), which implies that the metal-support interaction in NiO/ZSM-5 was weaker than those of NiO/Al2O3 and NiO/SiO2. Fe modified NiO/ZSM-5 catalysts show the lowest reduction temperature of NiO, suggesting some interaction of Fe ions with nickel oxide. Two shoulder peaks appear over the NiO/ZSM-5-Fe at approximately 736 K and 810 K respectively, which can be ascribed to the reduction of Fe2O3 to FeO and Fe gradually [35]. This suggests that ion-exchanged iron is in the form of iron oxide in NiO/ZSM-5-Fe.The heterogeneous catalytic reaction is comprised of different steps, such as surface and pore diffusion, adsorption and surface reaction [36]. Adsorption affects the reaction performance to some extent. MCH adsorptions were investigated by TPD after MCH saturation. As illustrated in Fig. 5 , MCH adsorption capacities on different Ni-based catalysts decrease in the order of NiO/ZSM-5-Fe > NiO/ZSM-5 > NiO/Al2O3 > NiO/SiO2, which is consistent with the order of the number of the moderate acid sites. This implies that the MCH adsorption capacity might be directly related with the moderate acid sites of the samples. Namely, as for acidic supports, MCH mainly adsorbs on the moderate acidic centers. It indicates that this experiment is carried under oxide form of catalyst, thus the MCH adsorption on Ni surface is excluded.The catalytic transformation of MCH over NiO/SiO2, NiO/ZSM-5, NiO/Al2O3 and NiO/ZSM-5-Fe was conducted at 673 K and hydrogen pressure of 1.5 MPa, and the conversion of MCH and yield of various products versus reaction time are presented in Fig. 6 . As Scheme 1 depicts, the reactions of MCH over the bifunctional catalysts involve dehydrogenation aromatization, isomerization, and cracking to aromatics, linear chain isoparaffins, cyclicisoparaffins and small alkenes. Over the bifunctional catalysts, MCH can continue to lose hydrogen to give toluene mainly on the nickel sites. Due to the introduction of B acid sites, protonation of the tertiary hydrogen in MCH forms hydrogen and methylcyclohexyl carbocation. The methylcyclohexyl carbocation can isomerize, lose a proton to produce methyclohexene (MCHE) or crack by the β-scission mechanism to give 2-methyl-1-hexene, which could occur further cracking to obtain products with less than seven carbons, primarily C4s and C3s. Methylcyclohexyl carbocation can also isomerize to alkyl cyclopentyl carbocations, which can desorb to generate alkylcyclepentanes. The RON of above products are all higher than that of MCH. As shown in Fig. 6a, significant differences in activity and product distribution are observed among the four samples. As for MCH conversion, NiO/ZSM-5-Fe provides the highest catalytic activity. The NiO/ZSM-5-Fe (86.7% initial conversion) is much more effective than NiO/ZSM-5 (40% initial conversion) for MCH conversion. But, it is surprised that the activity of NiO/Al2O3 is lower than that of NiO/SiO2 although NiO/Al2O3 has a certain amount of L acid sites and moderate adsorption capacity of MCH.From the reaction results over four catalysts, we deduce that the main active centers of acidic catalysis for MCH conversion should be the B acid sites with moderate acid strength. It can be seen that the conversion of MCH follows the sequence of NiO/ZSM-5-Fe > NiO/ZSM-5 > NiO/SiO2 > NiO/Al2O3. And the B acidity obtained from pyridine adsorption at 623 K is also in the order of NiO/ZSM-5-Fe > NiO/ZSM-5 > NiO/SiO2 = NiO/Al2O3 = 0 (Table 2). It is indicated that among the supported catalysts, NiO supported over ZSM-5-Fe exhibits the highest MCH conversion and the least cracking activity, although the total amounts of acids over NiO/ZSM-5-Fe is lower than NiO/ZSM-5. NiO/Al2O3 catalyst with moderate adsorption capacity of MCH, which total acid amounts are lower than that of of NiO/ZSM-5-Fe, but is higher than that of NiO/SiO2, has the lowest MCH conversion activity. This reaction results illustrates that other acidic sites, such as weak acidic centers and moderate L acidic centers, may provide the adsorption of MCH, but cannot efficiently catalyze MCH conversion.A decrease of MCH conversion with the reaction time is observed which is attributed to the coke formation, blocking the active sites. Because the S-Zorb process is operated in the way of fluidized fluid beds with continuous regeneration, thus the coke can be removed in the regeneration of deactivated catalysts if this catalysis system is integrated into the S-Zorb technique.The product distributions over the four samples are illustrated in Fig. 6b–d. The yield of aromatic product over NiO/ZSM-5-Fe is significantly higher than those over NiO/ZSM-5, NiO/Al2O3 and NiO/SiO2 (Fig. 6b). The dehydrogenation is supposed to mainly occurs on the Ni active centers. The TPR shows that the temperature of reduction peak is in the sequence of NiO/ZSM-5-Fe < NiO/ZSM-5 < NiO/SiO2 < NiO/Al2O3. The decrease of interaction of support with Ni makes the reduction of NiO more easier. Due to the interaction between Fe ions and NiO and the weakened interaction of ZSM-5 with NiO phase, the sufficient Ni reduction was achieved on the NiO/ZSM-5-Fe. Meanwhile, the enhanced dehydrogenation aromatization of MCH performances for NiO/ZSM-5-Fe and NiO/ZSM-5 catalysts could also be attributed to the synergetic effect between active Ni components and the B acid sites of ZSM-5.Hydrocracking and isomerization of MCH take place in several steps, firstly the saturated cycloalkanes are transformed to the olefins via dehydrogenation on metal sites, then the olefin intermediates protonated to carbenium ions on the B acid sites, which either isomerizes to the dimethyl substituted cyclopentene or cracks to fragments. At last, these unsaturated products can be further hydrogenated to the saturated 5-membered ring cycloparaffin or the saturated lower molecular paraffins (principally propane). The hydroisomerization activity of MCH over these four catalysts is shown in Fig. 6c. It can be seen that NiO/ZSM-5-Fe catalyst exhibits the highest isomerization yield, followed by NiO/ZSM-5 catalyst, while NiO/SiO2 and NiO/Al2O3 show almost no isomerization activity. From the results in Table 2, the isomerization activity is supposed to be mainly related with the B acids at 623 K, namely the B acid with moderate strength. It is postulated that weak B acid sites cannot effectively transfer the proton onto olefins to form the carbenium ions. On the other hand, olefins adsorb on moderate B acid sites by protonation, leading to the isomerization reaction. Thus NiO/ZSM-5-Fe with more moderate strong B acid sites shows excellent isomerization activity.Although isomerization can improve the RON, another acidic catalytic cracking reaction may decrease the liquid products yield. As shown in Fig. 6d, the cracking yields is in the order of NiO/ZSM-5 < NiO/ZSM-5-Fe < NiO/Al2O3 < NiO/SiO2, which is agreement with that of total acidic amounts. It is reasonable that the cracking reaction occurs both on the L and B acid sites. The acidic strength and types have less influences on the cracking activity than acid amounts.From above reaction results, NiO/ZSM-5-Fe obviously exhibits an appropriate balance between the high yield of aromatic and isomerization products and less cracking. The modification of ZSM-5 with iron can raise the amount of medium strong B acid sites and decrease the total amount of acid sites, which leads to a high yield of aromatic products by enhancing the conversion of MCH and inhibiting the ring opening and cracking of MCH simultaneously. These results show that NiO/ZSM-5-Fe can effectively catalyze the aromatization and isomerization of MCH under the S-Zorb reaction conditions. Moreover, modification of NiO/ZSM-5 with iron creats more medium strong B acid sites, which is favorable to improve aromatization and isomerization products during gasoline desulfurization, with simultaneous enhancement of the RON of fuels and desulfurization efficiency. Thus, the NiO/ZSM-5-Fe sample may be introduced as specific modifier onto S-Zorb catalysts, and further study will be investigated through the combination of NiO/ZSM-5-Fe with S-Zorb catalyst so as to recover the RON of gasoline during the S-Zorb reactive adsorption desulfurization.In summary, NiO/SiO2, NiO/Al2O3, NiO/ZSM-5 and NiO/ZSM-5-Fe catalysts were developed and evaluated for the conversion of MCH under the S-Zorb catalytic adsorption desulfurization conditions. The results indicated that the catalytic activity and the distribution of main products were significantly influenced by the interaction between NiO and supports and the acid–base properties of the catalysts (including acid amount, strength and types). TPR results showed that the locations of reduction peaks gradually shifted to higher temperature in the order of NiO/ZSM-5-Fe < NiO/ZSM-5 < NiO/Al2O3 < NiO/SiO2, as well as the onset temperature of reduction. This indicated that NiO/ZSM-5-Fe and NiO/ZSM-5 was more easier to be reduced than NiO/Al2O3 and NiO/SiO2 due to the weaker interaction between NiO and ZSM-5. On the other hand, the Ni active centers over the ZSM-5 surface would be expected to catalyze the dehydrogenation of MCH into methylcyclohexene and so on. The total acid sites, especially the number of B acid sites with mediun strong acidity, also played a critical role for MCH conversion. The medium strong B acid sites were the main active sites for aromatization and isomerization reaction and it was improved significantly after modification with iron metal. So, the best result was obtained by using NiO/ZSM-5-Fe as the catalyst, with 86.7% MCH conversion and remarkable aromatization and isomerization yields of 82.1% and little cracking products (4.6%).The authors have declared that no conflict of interest exists.The authors would like to acknowledge the financial support from the National Natural Science Foundation of China (21433001, 21406251 and 21403265), Science and Technology Development Projects of SINOPEC, China (No. 113138, 112008 and 110099), and The Young Taishan Scholars Program of Shandong Province (tsqn20161052).

In this work, nickel metal supported on different supports (SiO2, Al2O3, ZSM-5) were prepared by spraying nickel nitrate on the supports and calcined at 873 K. Then, they were characterized by XRD, XRF, N2 adsorption–desorption, NH3-TPD, MCH-TPD, H2-TPR, and pyridine-FTIR, and tested as catalysts for the dehydrogenation aromatization and isomerization of methylcyclohexane (MCH) under the conditions of S-Zorb catalytic adsorption desulfurization (T = 673 K, P = 1.5 MPa, WHSV = 5 h−1). The H2-TPR results showed that the interaction of NiO with support decreased in the order of NiO/ZSM-5-Fe < NiO/ZSM-5 < NiO/Al2O3 < NiO/SiO2. The decrease of the interaction appeared to facilitate the reduction of Ni and therefore to promote the dehydrogenation aromatization of MCH. It was found that a direct correlation existed between the gasoline components yields, cracking activity and the total number of different supports acid sites measured by NH3-TPD tests. Higher total acidity of ZSM-5 resulted in gasoline loss because of higher cracking activity of MCH. The number of total acid sites of NiO/ZSM-5-Fe decreased and the medium strong Brönsted acid sites necessary for MCH isomerization increased after the modification of ZSM-5 by iron metal. So, NiO/ZSM-5-Fe exhibited enhanced MCH conversion, aromatic and isomerization yields when compared to NiO/ZSM-5 and other Ni-based catalysts. This study shows that NiO/ZSM-5-Fe catalyst may be possible to be integrated into the S-Zorb system achieving the recovery of the octane number of gasoline.

CH4 is a major contributor to global warming. A reaction like the decomposition of methane has great environmental importance as CH4 is consumed and its concentration is depleted in the environment. Again, this reaction has great economic feature as it produces clean energy source H2 and high-quality carbon without the formation of CO/CO2 (CH4 → C + 2H2). The separation of H2 gas from solid carbon is easier than the separation of two gases in other reforming processes (stream reforming; CH4 + H2O ⇌ CO + 3H2, dry reforming; CH4 + CO2 ⇌ 2CO + 2H2). However, the decomposition of methane through C–H cleavage occurs at very high reaction temperature (up to 1200 °C). To decrease the reaction temperature or bond dissociate energy of C–H prominently, the high temperature sustainable Fe, Co, Ni, Cu-based catalyst is required (Gajewski and Pao, 2011; Wu et al., 2009). When the electronic promotor La2O3 was added with physical mixture of Ni-Cu alloy with a mass ratio of 0.107, the rate of C–H dissociation was increased (Figueiredo et al., 2010). The interaction of deposited carbon (after CH4 decomposition) with lattice oxygen at high temperatures cannot be neglected which forms CO in low quantity (Choudhary et al., 2001). The catalyst community is trying to develop a catalyst for COx-free H2 production through CH4 decomposition.Among the supported catalyst systems, Ni supported on activated carbon had drawn attention due to the in-situ generation of the active site (metallic Ni) through the reducibility of activated carbon in the carbonization process. Coal char derived from lignite coal showed minute CH4 decomposition in micropores. C–H bond dissociation energy of CH4 over coal char was found four times (89–105 kJ/mol) less than in uncatalyzed reaction (Bai et al., 2006). Prasad et al. (Sarada Prasad et al., 2011) prepared activated carbon from a coconut shell and impregnated Ni over it. Ni supported on activated carbon-showed an initial decrease of CH4 conversion in the first two hours and thereafter an increase in CH4 conversion. 10 wr% Ni supported on carbon (derived fromcoal liquefactionresidue) (Zhang et al., 2013) showed a continuous rise of CH4 conversion (13 to 60%) at 850 °C during 9 h time on stream. Another thermally stable support that can hold Ni during high temperature reaction was tried. Nanosized Ni (prepared by citric acid at pH controlled condition) supported over high silica ZSM (Si/Al = 300) showed about 45% CH4 conversion for a small time (14 min) at 700 °C (Michalkiewicz and Majewska, 2014). Among Ni/HY, Ni/SiO2, Ni/H-ZSM; CO contamination was found lowest on Ni/SiO2. Ni supported on showed ∼ 10% CH4 conversion during the entire range of reaction temperatures 500 °C – 800 °C for 5 h (Dong et al., 2015). 30 wt% Ni loading over SiO2 support showed 15% CH4 conversion up to 8 h time (Venugopal et al., 2007). Longevity of Ni-supported catalyst was found in the following order Ni/MgO > Ni/SiO2 > Ni/LiAlO2 > Ni/ZrO2 (Bonura et al., 2006) in which Ni supported on MgO showed>30% CH4 conversion up to 210-minutes. Ni/MgO catalyst can be regenerated in the O2 stream and utilized again for the reaction without any prior reduction step. 10-40 wt% Ni-MgO catalyst prepared by hydrothermal method showed the presence of NiO-MgO solid solution with mesoporosity (Bai et al., 2021). CH4 and N2 gas feed (1:2 vol ratio) over 30–40% Ni-MgO reached above 45% H2-yield within 3 h at 600 °C. Karimi et al. studied the decomposition of CH4 (in CH4: N2 = 3: 17) over Ni supported on MgSiO3-(prepared by the coprecipitation method) (Karimi et al., 2021). The catalyst showed 64% CH4 conversion at 600 °C. Low La/Ni ratio in LaNiO3, La4Ni3O10, La3Ni2O7 and La2NiO4 was known for bulk carbon decomposition with high degree of graphitisation (Li et al., 2001). Ni-incorporated hydrotalcite was derived from 2: 0.7: 0.3 mol ratio of Ni: Al: La metal precursors respectively. It had strong metal support interaction and showed ∼ 30% CH4 conversion up to 24 h (Anjaneyulu et al., 2015).The Co-Al mixed oxide had Co3O4 phase and Co2AlO4 (spinel) phases (Calgaro and Perez-Lopez, 2019; Zardin and Perez-Lopez, 2017). The catalyst reduced under CH4 (than under H2) had lower particle size and showed 75% CH4 conversion at 750 °C reaction temperature. In Co-Al mixed oxide, the Co3O4 phase favoured graphene formation. 20 wt% Co-impregnated Al2O3-coated silica fabric has strong metal support interaction and showed 90% CH4 conversion up to 11.6 h at 700 °C (Italiano et al., 2010). Cu and Ni supported on Alumina-was found better than Cu supported onalumina-catalyst because of the formation of Ni-Cu alloy. After reduction, 70%Ni–10%Cu–10%Fe/Al2O3 catalysts showed the formation of Ni-Cu-Fe alloy (Chesnokov and Chichkan, 2009). Alloy formation caused a decrease in the number of contacts between metal particles and thus sintering was prevented. Upon iron addition in 70%Ni–10%Cu/Al2O3 catalyst, H2 concentration remained between 71 and 77% and the diffusion coefficient of the carbon atom was increased three times (carbon nanofiber yield 136 g/g). At 15 ml/min methane flow rate, 65%Ni-10%Fe-25%SiO2 catalyst showed 20% CH4 conversion at 550 °C reaction temperature (Wang et al., 2012). However, the presence of Ni-Fe redox (in Ni2-xFexAl; x = Fe/Al) also functions as oxygen carrier (Huang et al., 2018) which can mitigate the target of COx-free H2 production.In the mean of Ni, Co, and Cu free catalyst, a mechanochemical activation of LaFeO3 and CeO2 mixture had drawn attention. It caused an accumulation of oxygen vacancy about Fe+3 which became the sites of oxygen exchange between O2 form air to surface to bulk CeO2 (Pinaeva et al., 2013). However, in presence of oxygen; COx-free hydrogen production from CH4 was not possible over mechanochemical mixture of LaFeO3 and CeO2. If 60 wt% Fe supported on alumina catalyst was reduced under the H2 stream, iron oxide was reduced into metallic Fe (Ibrahim et al., 2015). The metallic Fe is an active site for CH4 decomposition. Fe supported on Al2O3- generated multiwalled nanotube and 77.2 % H2 yield up to 4 h at 700 °C. Decomposition of CH4, C2H4, and C2H2 over Iron-based catalysts was reported (Maroto Valiente et al., 2000; Qian et al., 2008). Jin et al. prepared activated carbon from coconut shell and impregnated the 40 wt% iron oxide and alumina (Fe/Al = 24/16) over activated carbon (Jin et al., 2013). Here, activated carbon brought in-situ reduction of Fe(NO3)3 to metallic. During N2 pre-treatment process at 870 °C, the carbon wall was burned off by Fe and created mesopores. The catalyst showed 35% CH4 conversion up to 100 h.By literature review, we come to know that the widely available and cheap Fe can be utilized for the generation of COx-free H2 through CH4 decomposition. The activated carbon as support had the additional benefits as it had in-situ generation capacity of catalytic active sites (metallic Fe) by carbon reducibility. Tungsten had appealing redox chemistry and WC had high thermal stability (Mounfield et al., 2019). In the presence of W, additional CH4 decomposition sites were previously claimed also (Patel et al., 2021). Herein, waste date pits were utilized for the preparation of activated carbon. The WO3-activated carbon support was prepared by hydrothermal method and thereafter iron was impregnated over the WO3-activated carbon support. It is expected that if tungsten oxide is used as support along with activated carbon, Ni supported on WO3-activated carbon catalyst system would be benefited by high thermal stability, in-situ reducibility, and enhanced CH4 dissociation. The prepared catalyst was investigated for CH4 decomposition reaction and characterized through X-ray diffraction, N2-physiosorption, and porosity measurement, H2-temperature programmed reduction, thermogravimetric analysis, O2-temperature programmed oxidation and X-ray photoelectron spectroscopy. The fine correlation of catalytic activity and characterization results will add a step up in the development of an industrially suited catalyst for CH4 dissociation.The following materials were used in the preparation of the newly designed catalysts; Sodium tungstate dehydrate (Na2WO4·2H2O, ≥ 99% Sigma Aldrich), sodium chloride (NaCl; ≥ 99.0%, Sigma Aldrich), hydrated iron nitrate (Fe(NO3)3·9H2O; 99%; Loba Chemie), hydrochloric acid (HCl; 37%, Sigma Aldrich) and waste of date pits (collected from Albaha region, Saudi Arabia).The waste of date pits was cleaned, sieved, and washed several times by deionized water. Further, it is carbonized on heating at 250 °C under an electrical oven for 24 h. The black carbonized pits were obtained, ground and sieved. Finally, black carbon powder is obtained. To activate the black powder, concentrated H2SO4 was added and the mixture was heated at 250 °C in an oven for 24 h. The obtained material was washed several times with deionized water until pH 7 is not attained. The activated carbon material was abbreviated as “Ac”.The support WO3 nanoparticles were synthesized by hydrothermal process. 1.067 g of Na2WO4·2H2O and 0.038 g of pure NaCl were dissolved in 20 ml distilled water in stainless steel autoclave and stirred the solution in the dark for 30 min. Further, 5 ml HCl solution was added dropwise in this solution. The mixture (in an autoclave) was placed in the oven at 150 °C for 10 h. The precipitate in the autoclave was washed several times with distilled water until pH 7 was not reached. Finally, sample was calcined in air at 450 °C for 5 h. The material was used for support further and abbreviated as W.The support Ac-doped WO3 nanoparticles were prepared by the following procedure. Appropriate amounts of “x” wt% Ac and 100-x wt% WO3 (x = 5–95) were added in 20 ml distilled water under the stirring conditions in the autoclave. Further, HCl solution was added to the solution, kept for 30 min at room temperature, and then placed in an autoclave under the oven at 150 °C for 12 h. The precipitate in the autoclave was washed several times with distilled water until pH 7 was not reached. Finally, the sample was calcined in air at 450 °C for 5 h. The material was used as support further and abbreviated as xW(100-x) Ac (x = 0–100).30 wt% Fe loading was obtained from dissolving the specified amount of hydrated iron nitrate in 30 ml water and followed by impregnated of this solution over Ac or W or xW(100-x)Ac (x = 0–100) support at 80 °C for 3 h. Further, the slurry was dried overnight at 120 °C and calcined at 600 °C for 3 h sequentially. Fe supported on activated carbon, Fe supported on tungsten oxide, Fe supported on “tungsten oxide-activated carbon” catalysts were abbreviated as 30Fe100AC, 30Fe100WO3,and 30FexW(100-x) Ac (x = 0–100) respectively.X-ray diffraction (XRD) study of catalyst samples was carried out by Rigaku diffractometer using Cu Kα radiation source operated at 40 kV and 40 mA. 0.01 step size and 5–100 scanning range were set for analysis. Phase analysis was carried out by using X’pert high score plus software and JCPDS database. N2-physiosorption isotherms study of catalyst sample was carried over Micromeritics Tristar II 3020. Surface area was estimated by Brunauer-Emmet Teller (BET) method whereas pore volume and pore diameter were estimated by Barrett-Joyner-Halenda (BJH) method. The reducibility of the catalyst sample was studied by H2-temperature-programmed reduction (TPR) over Micromeritics Auto Chem II 2920, USA. 70 mg of the sample was subjected to a heat treatment at 10 °C/min up to 900 °C under 30 ml/min gas flow of 10% H2/Ar mixture gas. The thermogravimetric analysis (TGA) was carried out over 0.015 g of spent catalyst sample in the temperature range (room temperature to 1000 °C) at heating ramp 20 °C by using Shimadzu TGA-51. The TGA analysis was carried out under oxidizing gas O2. The weight loss/weight gain of catalyst sample against temperature was monitored continuously. O2-Temperature programmed oxidation (TPO) was carried out over spent catalyst system in 50–800 °C temperature range by using a 10% O2/He mixture through by Micromeritics AutoChem II. Before analysis, the spent catalyst was treated under high purity Argon at 150 °C for 30 min and subsequently cooled to room temperature. The morphology of the catalyst sample was investigated by using a field emission scanning electron microscope (FE-SEM, model: JEOL JSM-7100F) and transmission electron microscope (TEM, model: 120 kV JEOL JEM-2100F). Element valance state and binding energy of electron were determined by X-ray photoelectron spectroscopy (XPS) (Themo Fisher Scientific, USA) operated through AlKα excitation source and 20 eV pass energy.The detailed reaction set up for the CH4 decomposition reaction is shown in Fig. 1 . Catalytic decomposition of methane was carried out over 0.15 g catalyst packed in fixed-bed stainless steel tubular micro-reactor (PID Eng & Tech micro activity reference company; L = 30 cm, I.D = 9.1 mm) at atmospheric pressure. The reactor temperature was monitored by an axially positioned K-type stainless steel sheathed thermocouple at the centre of the catalyst bed. Prior to the reaction, the catalyst was activated under 40 ml/min flow of H2 for 60 min at 600 °C. Futher reactor is purged by N2 for 15 min to remove the remnant of H2. Now, the temperature of the reactor was raised to 800 °C under flow of N2. 15 ml/min CH4 and 5 ml/min N2 (total flow rate of feed gas 20 ml/ min) was allowed to pass through the catalyst bed at 800 °C with 8000 ml/hgcat space velocity of. GC-2014 SHIMADZU (Column: Shin carbon C20380 for gases and Haysepe Q AC0209 column for water analysis; carrier gas: Argon) equipped with conductivity detector was used to analyse the feed and output gas composition. The expression for CH4 conversion, H2 yield and Carbon yield (%) are given as C H 4 c o n v e r s i o n = C H 4 , i n - C H 4 , o u t C H 4 , i n × 100 % H 2 Y i e l d % = M o l e o f H 2 i n P r o d u c t 2 x m o l o f C H 4 i n x 100 C a r b o n y i e l d % = W p - W c a t × 100 Wcat (Where Wp is the weight of the product after reaction and Wcat is the weight of the fresh catalyst).The X-ray diffraction pattern of 30FexW(100-x) Ac (x = 0–100%) catalysts are shown in Fig. 2 and Fig. 3 . 30 wt% Fe supported over activated carbon had only phases related to iron oxide at Bragg angle (2θ) 24.07°, 33.12°, 35.60°, 39.19°, 40.80°, 49.38°, 54.10°, 57.56°, 62.33°, 63.99°, 69.55°, 71.80° (JCPDS reference number 00–024-0072) (Fig. 2 A-C). As activated carbon amount is substituted by tungsten oxide up to 5–10 wt% W, the diffraction peak intensity for iron oxide is suppressed greatly in 30FexW(100-x) Ac (x = 5–10) catalyst. However, upon the 1:3 ratio of W and Ac (25 wt%WO3-75%Ac), the many peaks related to Fe2O3 again appeared with variation in intensity over 30Fe25W75Ac catalyst (Fig. 2 D). The 30Fe25W75Ac catalyst also shows crystalline carbon phases at 2θ values of 27.38°, 36.10°, and 62.7° (JCPDS reference number: 00–018-0311) (Fig. 2 B-C). When “50 wt% WO3-50 wt% Ac” support is prepared for 30 wt% Fe dispersion, orthorhombic tungsten oxide phases at 22.97°, 24.02°, 33.22°, 35.70°, 49.55°, 54.24°, 62.59° (JCPDS reference number: 00–020-1324), tungsten carbide phase at 35.70°, 64.11° (JCPDS reference number 01–073-0471) and Fe2(WO4)3 phases at 20.33°, 22.45°, 22.97°, 25.45°, 29.98° (JCPDS reference number 00–038-0200) are appeared additionally ( Fig. 2 D). Upon 3:1 ratio of W and Ac respectively, the 30Fe75W25Ac catalyst shows the most intense diffraction peak patterns along with additional diffraction peaks for tungsten oxide (monoclinic phase) at 23.61°, 28.70°, 30°, 34.11°, 38.94°, 56.29°, 68.81° (JCPDS reference number 01–072-0677) ( Fig. 2 E-F). It may be expected that on further increasing the weight ratio of W and Ac (W/Ac = 75/25, 90/10, 95/5), the peak intensity of tungsten-related phases should be increased but the opposite diffraction results are noticed ( Fig. 3 A-C). It indicates either the addition of “5-10 wt% activated carbon in tungsten oxide matrix” or “addition of 5-10 wt% WO3 in activated carbon” brings a drop of the crystallinity of the catalyst sample. Finally, on complete substitution of activated carbon by tungsten oxide, 30Fe100W catalyst shows iron oxide, tungsten oxide and the intense peak intensity for Fe2WO6 mixed oxide (at 20.50°, 27.41°, 31.08°, 33.21°, 35.99°, 39.09°, 49.59°, 54.19°; JCPDS reference number 00–015-0688) ( Fig. 3 D-F). The support “activated carbon” has 0.8348 m2/g surface area, 0.002550 cm3/g pore volume, and 393 Å pore diameter. For 30FexW(100-x) Ac (x = 0–100) catalyst system, adsorption isotherm, pore size distribution, surface area, pore volume, and pore diameter are shown in Fig. 4 and Fig. S2 . The catalyst system belongs to the type IV isotherm having an H3 hysteresis loop. It indicates the presence of non-rigid aggregate-like mesopores. Upon incorporation of 5-10 wt% tungsten oxide, the surface area of 30FexW(100-x)Ac (x = 5, 10) catalyst is 2.5 times than 30Fe100Ac indicating expansion of framework (Kumar et al., 2016). Upon 25 wt% tungsten oxide incorporation, the surface area is noticed to decrease to 30% but pore volume is increased by 46%. However, upon further loading up to 50 wt% W; the surface area and pore volume of 30Fe50W50Ac are increased to 3 times and 1.8 times (with respect to 30Fe100Ac) respectively. The pore size distribution plot (dV/dlogW vs W) indicates that up to 50 % incorporation of tungsten oxide, pore size distributions remain bimodal. In the 30Fe50W50Ac catalyst, the intensity of the low pore-width range is more pronounced than the higher pore-width range. In 30FexW(100-x)Ac (x = 0–100) catalyst systems, when support is made up by major WO3 than Ac (upon > 50 wt% W incorporation), the pore size distribution becomes multimodal. In 30FexW(100-x)Ac (x = 75–100),the surface area decreases suddenly to 11–22 m2/g (against 59.15 m2/g in 30Fe50W50Ac) due to deposition of various crystallite inside the pore (Rahman et al., 2015).The H2-Temperatured programmed reduction profile of 30FexW(100-x) Ac (x = 0–100) catalyst systems are shown in Fig. 5 A and Fig. S3 . The total H2-consumption during the H2-TPR experiment is shown in Table S4 . H2-TPR of Fe2O3 is constituted by two sharp peaks at 355 °C and 577 °C and a broad peak between 624 °C and 934 °C ( Fig. S3 ). The three peaks are correlated with the sequential reductions Fe2O3 → Fe3O4 → FeO → Fe respectively (Ibrahim et al., 2015; Jozwiak et al., 2007). The activated carbon-supported Fe (30Fe100Ac) catalyst shows shifting of lower temperature reduction peak to relatively higher temperature (at 385 °C) indicating interaction of Fe2O3-species with support. The peak at 385 °C belongs to reduction of interacted-Fe2O3-species into Fe3O4. Upon incorporation of just 5 wt% WO3, a merged peak maximum at 420 °C for reduction of Fe2O3 → Fe3O4 → FeO and a broad peak at a higher temperature for the reduction of FeO → Fe are observed. Further incorporation of 10 wt% WO3, the reduction peak of 30Fe10W90Ac catalyst is shifted towards more higher temperature (444 °C). It is noticeable that the amount of reducible iron species had decreased upon providing support as well as increasing the proportion of tungsten oxide (up to 10 wt%) in the support. It indicates that the total reducible quantity has decreased due to the interaction of Fe-species with the new support composed of xW(100-x) Ac (x = 0, 5, 10) catalyst. As well as WO3 incorporation is increased to 25 wt%, the reduction peak maxima of 30Fe25W75Ac catalyst are shifted to a higher temperature and the amount of reducible iron-species is increased to ∼ 34% with respect to 30Fe10W90Ac catalyst ( Table S4 ). Shifting of reduction peak to a higher temperature also indicates increased metal support interaction upon tungsten oxide loading. 30Fe50W50Ac catalyst has the lower temperature reduction peak (for reduction of Fe2O3 → Fe3O4 → FeO at 470 °C and broad higher temperature reduction peak with comparable amount of reduceable species than 30Fe25W75Ac catalyst. 30FexW(100-x) Ac (x = 90–100 wt%) showed the reduction peak about 530–560 °C. It again shows a general trend of increasing metal-support interaction upon tungsten oxide loading. The peak pattern of 30FexW(100-x)Ac (x = 90–100 wt%) has also an additional peak in the temperature region of 593 to 720 °C for reduction of WO3 crystallite or “WO3 interacted species”(Ramanathan et al., 2013). It is noticeable that the total concentration of reducible species at the catalyst surface is decreased sharply above 50 wt% tungsten oxide incorporation. 30Fe50W50Ac, 30Fe90W10Ac, 30Fe95W5Ac, and 30Fe100W catalysts had 174.2 cm3/g, 72.28 cm3/g, 68.24 cm3/h, and 46.32 cm3/g consumption of hydrogen. The H2-TPR pattern of the 30Fe25W75Ac catalyst is needed to address separately. It has the highest amount of reducible species over the surface (183 cm3/g H2 consumption in H2-TPR result) among other tungsten oxide incorporated catalysts. The H2-TPR peak pattern is constituted by five peaks enveloping each other at 433 °C, 506 °C, 664 °C, 816 °C and 929 °C. It indicates the 30Fe25W75Ac catalyst had the highest concentration of “Fe-related” reducible species which interacted with the support to different extents.The thermogravimetric analysis (TGA) of spent catalysts is shown in Fig. 5 B. The weight gain over the catalyst system in TGA analysis may be due to oxidation of “reduced metal species” (like iron or tungsten-related metal oxide) over the catalyst surface (Ibrahim et al., 2015). In the case of spent 30Fe25W75Ac and spent 30Fe50W50Ac, weight loss due to oxidation of deposit carbon is optimum and so weight gain due to oxidation of “reduced metal species” over these catalyst systems is not evident on TGA analysis. The carbon yield % over the different catalysts is found in the following order; 30Fe25W75Ac (140%) > 30Fe50W50Ac (107%) > 30Fe10W90Ac (120) 30Fe95W5Ac (93.3%) > 30Fe90W10Ac (66.6%) > 30Fe5W95Ac (13.3 %) > 30Fe75W25Ac (6.7%) ( Table S5 ). Clearly, carbon yield % over 30Fe25W75Ac and 30Fe50W50Ac are higher than other catalysts. Previously, tungsten species were claimed to generate additional CH4 decomposition sites (Patel et al., 2021). It seems that the presence of 25-50 wt% of WO3 in the catalyst system cultivates the optimum amount of catalytic active sites which leads potential dissociation of CH4 into carbon and H2. It resulted in an excellent carbon yield and severe weight loss. These findings also give the sign of higher activity toward CH4 decomposition reaction over 30Fe25W75Ac and 30Fe50W50Ac catalysts. Severe weight loss over spent 30Fe25W75Ac and spent 30Fe50W50Ac catalysts also indicates that the carbon deposits over these catalysts are not inert, it is oxidizable under O2 stream. Inert carbon deposit may shade the catalytic active site permanently and causes fast deactivation. The non-inert carbon deposit over 30Fe25W75Ac and spent 30Fe50W50Ac catalysts may cause slower deactivation than other catalysts.O2-TPO of spent 30FexW(100-x) Ac (x = 0–100) catalyst system is shown in Fig. 5 C-5D. In the literature, the O2-TPO peak profile is differentiated into three regions 300–500 °C for easily oxidizable α-carbon (amorphous carbon) species (Al-Fatesh et al., 2021), 500–600 °C for moderately oxidizable β-carbon species (Patel et al., 2021) and > 600 °C for higher crystallization degree of carbon species (Zhang et al., 2015). In our catalyst system, TPO peak maxima is found about ∼ 600 °C in spent-30Fe5W95Ac, spent-30Fe50W50Ac, and spent-30Fe90W10Ac catalysts whereas, for spent-30Fe10W90Ac, spent-30Fe25W75Ac, spent-30Fe95W5Ac and spent-30Fe100W catalysts, TPO peaks are at about ∼ 650 °C. This observation indicates that a particular amount of tungsten oxide (spent-30Fe5W95Ac, spent-30Fe50W50Ac, and spent-30Fe90W10Ac catalyst) in the support induces less crystallization degree of carbon (than30Fe10W90Ac, 30Fe25W75Ac, and 30Fe100W catalyst).The morphology of catalyst samples is shown by SEM images in Fig. S6 . The catalyst morphology of fresh low tungsten-containing samples (30Fe10W90Ac) or high tungsten-containing samples (30Fe90W10Ac) is not differentiable. In the case of spent 30Fe10W90Ac catalyst, carbon tubes are easily observed than in 30Fe90W10Ac catalyst. The morphology of the catalyst and carbon tube is evident in TEM images under Fig. 6 . TEM image indicates particle size over the 30Fe25W75Ac has grown from 5.57 nm to 5.8 nm after the reaction (Fig. 6 A-D). Fig. 6 E shows the presence of carbon nanotubes of varying diameters. A typical multiwalled carbon tube having wall width of 3.81–4.74 nm and total tube width of 13.13 nm is evident in Fig. 6 F.The X-ray photo-electron spectra of 30Fe25W74Ac catalyst is shown in Fig. 7 . Fe (2p3/2) peak at 711 eV and Fe (2p1/2) peak at 725 eV and O (1 s) peak at 530.1 confirms the presence of Fe+3 oxidation state (Allen et al., 1974; Konno and Nagayama, 1980) (Fig. 7 A- B). The presence of W(4f7/2) peak at 35.4 eV and W(4f5/2) peak at 37.6 eV confirm the presence of WO3 or W+6 oxidation state (Fig. 7 C) (Barreca et al., 2001). The C(1 s) XPS spectra is observed at 284.7 (Barreca et al., 2001; Grünert et al., 1987) ( Fig. 7 ). The 30Fe25W75Ac catalyst has both carbon and WO3 but absence of carbidic carbon peak at 282.7 eV indicates that WC like species are not formed over the catalyst surface (Katrib et al., 1994). Overall, from the XPS spectra presence of Fe (III) (as Fe2O3), W (IV) (as WO3) species are confirmed. Fe2O3 and WO3 phases are already confirmed during the XRD analysis of sample.The CH4 conversion, H2-yield and ratio of H2-yield/CH4 conversion ( Y H 2 / C C H 4 ) are shown in Fig. 8 A-8C. The initial conversion of CH4 and H2-yield at 30wt.%Fe supported over activated carbon (30Fe100Ac) is just 4.43% and 1.96% respectively whereas 30wt.%Fe supported over tungsten oxide (30Fe100W) shows 20.5% initial CH4 conversion and 19.43% initial H2-yield. Markedly tungsten-oxide supported iron has a higher catalytic activity as well as ratio of Y H 2 / C C H 4 is > 0.94 whereas activated-carbon-supported iron has low catalytic activity and a ratio of Y H 2 / C C H 4 is just half (Fig. 8 C). During the entire time on stream (420 min), Y H 2 / C C H 4 ratio remains close to 0.9 over 30Fe100W whereas Y H 2 / C C H 4 ratio drops to ∼ 0.2 over 30Fe100Ac at the end of 420-minutes time on stream. A high Y H 2 / C C H 4 ratio indicates a higher accumulation of CH4 or CHx over the catalyst surface followed by higher H2 release to the gas phase during the reaction (Łamacz and Łabojko, 2019). The higher activity of 30Fe100W than 30Fe100Ac catalyst toward methane decomposition reaction can be explained by X-ray diffraction and H2-TPR results. 30Fe100Ac catalyst had only reducible iron oxide as surface active species whereas 30Fe100W has iron oxide, tungsten oxide, and Fe2WO6 mixed oxide phases. Patel et. al found “additional CH4 decomposition sites” over tungsten oxide-zirconia in CH4-temperature programmed surface reaction experiment (Patel et al., 2021). It was reported that Fe2WO6 may also be reduced to respective Fe and W under the hydrogen stream at reaction temperature (Pak et al., 2009). In H2-TPR results, we found the reduction peak for WO3 at 593 °C to 720 °C over 30Fe100W catalyst. That means 30Fe100W has a variety of reducible surface-active species that markedly influence the CH4-decomposition reaction. Overall, it can be said that tungsten oxide is promising support for Fe-based catalyst for CH4 decomposition reaction.Up to incorporation of 5–10 wt% WO3 in 30FexW(100-x) Ac (x = 5–10) catalyst, X-ray diffraction peaks are suppressed and surface area is increased up to 2.5 times (with respect to 30Fe100Ac catalyst). It indicates the suppression of crystallinity upon expansion of the surface (Khalid et al., 2013). H2-TPR result of 30Fe10W95Ac showed increased metal-support interaction over than 30Fe5W95Ac catalyst. Increased metal-support interaction over an expanded surface is the ideal condition for exposing more active sites for CH4 decomposition. Overall, it can be said that 30Fe5W95Ac catalyst has an expanded surface (than 30Fe100Ac) and 30Fe10W90Ac catalyst has an expanded surface (than 30Fe100Ac) as well as increased interaction of reducible Fe-species (Fe2O3, Fe3O4, FeO) with support (than 30Fe5W95Ac). 30Fe5W95Ac catalyst has 29.5% initial CH4 conversion (against 4.43% in 30Fe100Ac) and 22.5% initial H2-yield (against 1.96% in 30Fe100Ac). 30Fe10W90Ac catalyst shows 50.2% initial CH4 conversion and 48.39% initial H2-yield. The Y H 2 / C C H 4 ratio over 30Fe10W90Ac also remains ∼ 0.9 up to 250 min whereas, over 30Fe5W95Ac, Y H 2 / C C H 4 ratio falls to 0.5 within 70 min time on stream. Here, the role of tungsten in CH4 decomposition, higher surface area, and more “surface interacted reducible Fe-species” are evident over 30Fe10W90Ac catalyst which achieves higher CH4 conversion and higher Y H 2 / C C H 4 ratio.Up to 25 wt% W incorporation; prominent iron oxide phase is evident over 30Fe25W75Ac catalyst. The presence of Fe2O3 and WO3 phases are also confirmed by Fe(2p), W(4f) and O (1 s) XPS spectra. The decrease in surface area (up to 30%) of 30Fe25W75Ac (with respect to 3010W90Ac) is compensated by an increase in average pore diameter (up to 46%). 30Fe50W50Ac has again expanded surface area (three times than 30Fe100Ac) and other tungsten-related phases like tungsten oxide, tungsten carbide, and Fe2(WO4)3 over the surface. 30Fe25W75Ac catalyst has the highest amount of reducible species (183 cm3/g H2 consmption in H2-TPR result) over the surface among the rest tungsten oxide incorporated catalysts. These reducible iron species have interacted with the support along a wide range of temperatures as per the extent of interaction with support. 30Fe50W50Ac catalyst has also good number of reducible species (174 cm3/g H2 consumption in H2-TPR result) after 30Fe25W75Ac catalyst. Previously, tungsten species were claimed to generate additional CH4 decomposition sites (Patel et al., 2021). 30Fe25W75Ac and 30Fe50W50Ac catalysts show severe weight loss and higher carbon yield. The catalyst activity of 30Fe50W50Ac and 30Fe25W75Ac towards the CH4 decomposition reaction are also close to each other initially. It indicates that the presence of 25-50 wt% of WO3 in the catalyst induces an optimum amount of catalytic active sites leading to potential CH4 dissociation, excellent carbon yield, severe weight loss and optimum H2-yield over 30Fe25W75Ac and 30Fe50W50Ac catalysts. The initial CH4 conversion of 30Fe25W75Ac and 30Fe50W50Ac catalysts are 66.04% and 64.82 % respectively. Again, the initial H2-yield over 30Fe25W75Ac and 30Fe50W50Ac catalyst is found 63.12% and 59.81 % respectively.Fe supported over “xW(100-x)Ac (x = 10–50)” are able to show > 50% CH4 conversion, ≥50% H2-yield and ∼ 0.9 Y H 2 / C C H 4 ratio initially. Severe weight loss in TGA profile is obtained over 30Fe25W75Ac and spent 30Fe50W50Ac catalysts. It indicates that carbon deposits over these catalysts are oxidizable/not inert/active. The non-inert carbon deposit deactivates 30Fe25W75Ac and spent 30Fe50W50Ac catalyst slowly than the rest catalysts. After 160 min, CH4 conversion and H2 yield of the 30Fe25W75Ac catalyst drop to 37.61% (against 66% initial CH4 conversion) and 35.2% (against 63.12% initial H2 yield) respectively. 30Fe10W90Ac catalyst is found the second best as the CH4 conversion and H2-yield don’t fall below 25% after 160 min time on stream. The Y H 2 / C C H 4 ratio of both 30Fe25W75Ac and 30Fe10W90Ac catalyst is also ≥ 0.9 up to 240 min. At the end of 160 min 30Fe50W50Ac catalysts showed ∼ 19% CH4 conversion, ∼11% H2 yield and 0.56 Y H 2 / C C H 4 ratio. Overall, at the end of 420 min time on stream, 30Fe25W75Ac is found best. It has ∼ 14% CH4 conversion, ∼6% H2-yield and > 0.4 Y H 2 / C C H 4 ratio at 420 min time on stream.TGA results indicate severe coke decomposition over 30FexW(100-x) Ac (x = 25, 50) catalyst. O2-TPO result indicates that coke over the 30Fe25W75Ac catalyst has a higher crystallization degree than the 30Fe50W50Ac catalyst. The carbon yield calculation of the spent 30FexW(100-x) Ac (x = 5, 10, 25, 50, 90, 95) catalyst system is shown in Table S5 . Here also, the carbon yield over the spent 30Fe25W75Ac catalyst is greater than the 30Fe50W50Ac catalyst. Interestingly, the catalytic activity of the 30Fe25W75Ac catalyst is less affected by severe coke deposition but the activity of 30Fe50W50Ac drops suddenly on increasing time on stream. Initially, the activity of both catalysts is close to each other but after the end of 200 min, 30Fe50W50Ac has only ∼ 16% CH4 conversion (against 30% in 30Fe25W75Ac), 9% H2-yield (against 29% in 30Fe25W75Ac) and 0.56 Y H 2 / C C H 4 ratio (against 0.97 in 30Fe25W75Ac). It indicates that coke decomposition affects the performance of 30Fe50W50Ac to a great extent but not the performance of 30Fe25W75Ac catalyst. It seems that over 30Fe25W75Ac, rate of CH4 decomposition (carbon formation) is well matched with the rate of diffusion of carbon species from metal-gas interface (where decomposition of CH4 took place) to the metal-nanofiber interface (where carbon precipitates to form carbon nanofibers). Over highly crystalline 30Fe50W50Ac catalyst (than 30Fe25W50Ac), the rate of carbon formation may not properly match the rate of carbon diffusion. So, carbon species isn’t able to be transferred away in time and would cover the catalyst’s active sites leading to catalyst deactivation (Chen et al., 1997).Tungsten oxide incorporation of>50 wt% in 30FexW(100-x)Ac (x = 75, 90, 95) causes a fast drop of surface area due to the deposition of various crystallite inside the pore (Rahman et al., 2015) constituted by major-tungsten oxide and minor-activated carbon. These catalysts systems have also low density of reducible reducible-species over the catalyst surface. Low surface area catalyst and few catalytic active sites on the surface conveys less initial CH4 conversion. 30Fe75W25Ac has the highest crystallinity among rest catalyst systems. 30FexW(100-x) Ac (x = 75–95) catalysts has low initial CH4 conversion (14–36%) and initial H2-yield (13–29%).Tungsten oxide incorporated activated carbon is found to be an excellent support for Fe based catalyst towards CH4 decomposition reaction (than activated carbon incorporated tungsten oxide catalyst) due to enhanced surface area, a higher concentration of various types of reducible surface-active species as iron oxide, tungsten oxide, and Iron tungstate. The research outcome over 30FexW(100-x) Ac (x = 0–50) catalyst can be pointed as follow: • 30Fe10W90Ac catalyst has a comparable surface area but higher metal support interaction than the 30Fe5W95Ac catalyst. So, the earlier one has higher activity than latter. • 30FexW(100-x) Ac (x = 10–50) catalyst shows > 50% initial CH4 conversion, ∼50% initial H2-yield and ∼ 0.9 initial Y H 2 / C C H 4 ratio. • 30Fe25W75Ac catalyst has the highest concentration of reducible surface-active species (compared to the rest tungsten incorporated catalysts). It shows 66.04% initial CH4 conversion and 63.12% initial H2 yield and > 0.9 initial Y H 2 / C C H 4 . • 30Fe50W5pAc catalyst has a comparable concentration of reducible surface-active to 30Fe25W75Ac catalyst. Both catalysts have severe carbon deposits, higher carbon yield, higher initial CH4 conversion and higher initial H2 yield than other catalysts due to the potential dissociation of CH4 into carbon and H2. • On longer time on stream, the activity of the 30Fe50W50Ac catalyst drops fast than 30Fe25W75Ac catalyst due to improper matching between the rate of carbon formation and the rate of diffusion over highly crystalline 30Fe50W50Ac catalyst (compared to 30Fe25W75Ac catalyst). Even after 160 min, CH4 conversion and H2 yield over 30Fe25W75Ac catalyst does not drop below 35%. • Inferior catalytic activity over 30FexW(100-x)Ac (x = 75, 90, 95) is due to low surface area catalyst and few catalytic active sites. 30Fe10W90Ac catalyst has a comparable surface area but higher metal support interaction than the 30Fe5W95Ac catalyst. So, the earlier one has higher activity than latter.30FexW(100-x) Ac (x = 10–50) catalyst shows > 50% initial CH4 conversion, ∼50% initial H2-yield and ∼ 0.9 initial Y H 2 / C C H 4 ratio.30Fe25W75Ac catalyst has the highest concentration of reducible surface-active species (compared to the rest tungsten incorporated catalysts). It shows 66.04% initial CH4 conversion and 63.12% initial H2 yield and > 0.9 initial Y H 2 / C C H 4 .30Fe50W5pAc catalyst has a comparable concentration of reducible surface-active to 30Fe25W75Ac catalyst. Both catalysts have severe carbon deposits, higher carbon yield, higher initial CH4 conversion and higher initial H2 yield than other catalysts due to the potential dissociation of CH4 into carbon and H2.On longer time on stream, the activity of the 30Fe50W50Ac catalyst drops fast than 30Fe25W75Ac catalyst due to improper matching between the rate of carbon formation and the rate of diffusion over highly crystalline 30Fe50W50Ac catalyst (compared to 30Fe25W75Ac catalyst). Even after 160 min, CH4 conversion and H2 yield over 30Fe25W75Ac catalyst does not drop below 35%.Inferior catalytic activity over 30FexW(100-x)Ac (x = 75, 90, 95) is due to low surface area catalyst and few catalytic active sites.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors extend their appreciation to the Deanship for Research & Innovation, Ministry of Education in Saudi Arabia for funding this research work through the project no. (IFKSURG-2-055).Supplementary data to this article can be found online at https://doi.org/10.1016/j.arabjc.2023.104781.The following are the Supplementary data to this article: Supplementary data 1

Production of COx-free H2 from CH4 (a major global warming contributor) over cheap catalysts is a dominant task for the scientific community to accomplish environmental-friendly clean H2 energy sources. Herein, a tungsten oxide-activated carbon-supported Fe catalyst is prepared by impregnation method, characterized by X-ray diffraction, surface area-porosity measurement, temperature programmed reduction/oxidation and thermogravimetry analysis. 30wt.%Fe supported tungsten oxide incorporated activated carbon catalyst is found superior to 30 wt% Fe supported on activated carbon incorporated tungsten oxide due to higher surface area and high concentration of reducible catalytic active sites. 30wt.%Fe impregnated over 25 wt%WO3-75 wt%activated carbon support catalyst has the highest concentration of reducible surface-active species and it had excellent performance among other tungsten oxide incorporated catalysts. The catalyst showed 66.04% CH4 conversion, 63.12% H2 yield and Y H 2 / C C H 4  > 0.9 initially which didn’t fall below 35 % up to 160-minutes. Improper matching between the rate of carbon formation and the rate of diffusion over a highly crystalline 30Fe50W50Ac catalyst resulted in rapid deactivation.

No data was used for the research described in the article.The reduction of organic compounds is an important chemical technology used for the fabrication of valuable components for agrochemicals, pharmaceuticals, and cosmetics [1].Different Ni catalysts have been employed for hydrogenation of hydrocarbons and functionalized organic compounds [2–7]. The Raney Ni is a representative example of heterogeneous catalysts used in the laboratory scale as well as in the industry [8]. In most cases, heterogeneous catalysts contain a metal active phase bonded to the inorganic or organic support. Exceptionally, Raney Ni catalyst is used in the unsupported form. Although catalytic activity of Ni compounds is in many cases lower than precious metals-based catalysts, they are studied intensively as a cheaper and more environmentally friendly alternative [9–11]. Not only the low price but also the abundance of Ni motivate development of catalytic processes based on this metal [12–14].Hydrogenation of acetophenone (APh) is a promising method to produce 1-phenylethanol (PhE), an important intermediate for fabrication of ibuprofen. This process can be carried out using gaseous hydrogen or transfer hydrogenation methods [15–17].Yus et al. reported on transfer hydrogenation of APh in the presence of Ni NPs and 2-propanol/NaOH as a hydrogen donor. Modest to high yields of the corresponding alcohols were obtained, however, a serious drawback of this systems was the application of stoichiometric amounts of nickel [18,19]. In the applied conditions Raney Ni formed a mixture of products, while Ni/Al2O3 was not active. In 2014, Marchi described a catalyst based on NiO/SiO2 for APh hydrogenation at 10 bar of H2. They found that the kind of solvent strongly affects the catalytic activity, however, without any impact for the selectivity to PhE [20]. The highest reaction rates were noted for C2 – C3 alcohols.Ni@C catalyst, containing Ni supported on graphene, was used for hydrogenation of APh to PhE in a flow reactor under 1 MPa of H2 at 100 °C. After 48 h of continuous reaction selectivity to PhE was 97.77% [21]. Amorphous Ni-B-P materials supported on macroporous SiO2 were used in selective hydrogenation of APh to PhE under 1.5 MPa of H2 [22].The harmful pollutant, 4-nitrophenol (4-NP), can be transformed by hydrogenation to 4-aminophenol (4-AP) which is used for drugs synthesis [23]. Yang et al. have fabricated a new Ni catalyst supported on mesoporous carbon and employed it for reduction of 4-NP with NaBH4. Spherical Ni NPs showed the largest activity factor of 20.9 s−1 g−1 [24]. Dong et al. prepared an Ni catalyst supported on N-doped mesoporous carbon, Ni/m-CN, by pyrolysis of Ni-MOF at 700 °C. This nanocatalyst catalyzed hydrogenation of 4-NP to 4-AP with the activity factor of 9.1 s-1g−1 [25]. This is the only example of using a catalyst obtained by thermal decomposition of Ni-MOF material in hydrogenation of 4-NP.Ni(0) containing MCM-41-carbon materials, prepared by carbonization, efficiently catalyzed reduction of 4-NP with the rate constant up to 0.09 min−1 [26]. Kuo et al. presented hydrogenation of 4-NP to 4-AP with application of the NiO/NiS composites containing different amounts of sulfur [27].Our interest lies in using MOFs calcination for the synthesis of catalytically active composites [28,29]. Herein, we present calcination of Ni-BDP MOF (BDP = 1,4-Bis(pyrazol-4-yl)benzene) which resulted in the formation of two new Ni composites. Remarkably, calcination under N2/O2 atmosphere (4:1) led to the Ni/NiO composite, an active catalyst of transfer hydrogenation of APh and 4-NP, used in an amount of only 8 mol%.Nickel(II) acetate tetrahydrate (Ni(OCOCH3)2·4H2O, 98%), sodium borohydride (NaBH4, 98%), APh (99%) and 4-NP (99%) were purchased from Sigma-Aldrich. Water purified by an Ultrapure Water System was used in all experiments. 4,4′-Benzene-1,4-diylbis(1H-pyrazole) (BDP) was prepared according to the procedure reported previously [30].Inductively coupled plasma mass spectrometry analysis (ICP-OES) was carried out in an iCAP 7400 DUO icp (Thermo Fisher Scientific). Transmission electron microscopy (TEM) images were performed using a FEI Tecnai G2 20 X-TWIN microscope. X-ray diffraction (XRD) patterns were recorded with a powder X-ray diffractometer D8-ADVANCE Bruker (Cu-Kα, λ = 1.54056 Å). X-ray photoelectron spectra (XPS) were recorded by XPS/AES system EA10 (Leybold-Heraeus GmbH, Cologne, Germany). All acquired spectra were calibrated to adventitious carbon C1s at 285 eV. A microRaman apparatus (inVia™ Renishaw) was used to register both the Raman and the emission spectra with 830 and 514 nm excitation lines, respectively. GC-FID and GC–MS were performed using a Shimadzu QP 2010 SE. UV–vis spectra were recorded in Varian Cary 50 UV–Vis spectrophotometer. Magnetization measurements at 300 K were carried out on 0.00161 g sample of compound using a Quantum Design SQUID Magnetometer (type MPMS-XL5) with an applied field between −20000 Oe to 20000 Oe. NiBDP was prepared according to the procedure reported previously [30]. 3 mmol of ligand BDP (4,4′-benzene-1,4-diylbis(1H-pyrazole)) was dissolved in 160 mL of N,N′-dimethylformamide and 4 mmol of Ni(CH3COO)24H2O was dissolved in 40 mL of H2O. The two solutions were mixed and refluxed for 12 h under stirring. The obtained material was filtered off and washed with N,N’-dimethylformamide, ethanol and dried in air.0.31 g of NiBDP MOF was transferred to a quartz vessel and the solid was heated at 700 °C for 10 min under a flow of N2/O2 (4:1) with heating/cooling rates of 2.5 °C/min. Yield: 0.064 g. ICP: Ni 31.7%; Elemental anal.: C 1.10%, N 0.03%, H 1.22%.0.31 g of NiBDP MOF was transferred to a quartz vessel and the solid was heated at 700 °C for 10 min. under a flow of air with heating/cooling rates of 2.5 °C/min. Yield: 0.051 g. Elemental anal. C 0.5%; N 0.04%; H 0.37%.Aqueous solutions of 4-NP (75 mL, 0.15 mmol) and NaBH4 (75 mL, 15.0 mmol) were transferred into a three-neck roundbottom flask. After several minutes of stirring, NiOBDP or Ni/NiOBDP catalyst (8 mol%), were transferred into the flask. The mixture was continuously stirred at room temperature (25 °C). The progress of the reduction reaction was monitored by taking 0.5 mL aliquots of the supernatant solution, diluting them with cold deionised water (9.5 mL, 5 °C), and examining by UV–Vis spectroscopy.The transfer hydrogenation reaction was performed in a glass bottle. The catalyst (8 mol%), APh (1 mmol), NaOH (0.26 mmol), 2-propanol (3 mL), were introduced into a bottle, and then the bottle was sealed and placed in an aluminium block and heated to 60–80 °C for 2–5 h. After the reaction was finished, the catalyst was separated by centrifugation, the organic products were analyzed by means of GC-FID and GC–MS.Two new Ni-based catalysts, NiOBDP and Ni/NiOBDP, were prepared by thermal decomposition of the Ni-BDP MOF in different conditions (Scheme 1 ).The composites were characterized first by XRD (Fig. 1 A) and Raman spectroscopy (Fig. 1B). X-ray diffraction (XRD) patterns of NiOBDP and Ni/NiOBDP are shown in Fig. 1A. The major diffraction peaks of NiO at 37.2° for (111), 43.2° for (200), and 62.9° for (220) planes are observed in both samples, in agreement with the standard pattern peaks of cubic NiO structure (JCPDS No. 04–0835). Moreover, the major peaks of Ni(0) phase (JCPDSNo. 04–0850) were also observed at 44.5° for (111), and 51.8° for (200) planes for Ni/NiOBDP . The Raman spectra additionally revealed the presence of NiO in both composites. The small peaks centered at 200 cm−1 can be assigned to wide one-phonon (1P). The strong band at 500 cm−1 stemmed from wide one-phonon (1P) Ni-O Raman bands. Additionally, bands at 721, 893 and 1090 cm−1 can be identified as the two-phonon (2P) Ni-O Raman bands [31].It is important to note that despite of very similar synthetic procedures, in one of the samples, Ni/NiOBDP , Ni(0) was found besides of the main fraction of NiO. Thus, thermal decomposition of Ni-BDP MOF performed under N2/O2 (4:1) atmosphere produced some amount of Ni(0), whereas only NiO was formed in the air atmosphere.The TEM images (Fig. 2 ) showed the crystallites of the NiO particles with the diameter varying from 23 to 51 nm. The average size of nanoparticles calculated using the Scherrer equation was 29.7 nm, close to the average size obtained from the TEM images. Fig. 3 exhibits the Ni 2p3/2 spectra of NiOBDP and Ni/NiOBDP catalysts. Both spectra are similar and the major peak at 854.6 eV with the broad satellite peak centered at 861.8 eV that can be attributed to the surface Ni2+ species [32]. The shoulder peak at 856.7 eV originated from the surface Ni3+ species [33]. The peaks of Ni2+ and Ni3+ are slightly shifted compared to the commercial NiO (853.7 eV, 855.5 eV [34]. It is noteworthy that the signal of metallic Ni was not found in the spectrum of Ni/NiOBDP .O1s XPS spectrum (Fig. S1A) also confirmed the presence of NiO in both samples. The deconvoluted peak located at 529.3 eV stemmed from O bonded to Ni2+. In addition, the peak located at 530.3 eV was assigned to the O adjacent to the Ni vacancy [35]. Due to the different sample preparation routes (using the air or N2/O2 mixture), various amounts of water were present on the surface of NiO. According to the XPS results NiOBDP contained more water than Ni/NiOBDP (Fig. S1A). It should also be noted that for the Ni/NiOBDP sample more surface defects are visible in XPS and TEM pictures.Magnetic properties of Ni/NiOBDP were analyzed by a vibrating sample magnetometer. The magnetization curve reached a saturation value Ms about 15.09 emu g−1 (Fig. 4 ). The observed hysteresis loop characterized by the remanent magnetization Mr = 0.374 emu g−1 and coercivity Hc = 35 Oe is an evidence of ferromagnetic behavior of the material at room temperature.It is worth to note that magnetic nature of the catalyst facilitates its separation from the reaction mixture using a simple magnet. This enables to avoid additional operations such as centrifugation, filtration or other procedures before the catalyst recycling.The evaluation of catalytic activity of the new Ni catalysts was first performed for 4-NP transfer hydrogenation with NaBH4. The progress of the transformation of 4-NP to 4-AP was monitored by UV–Vis spectroscopy (Fig. 5 ).The absorption peak at 400 nm, originating from 4-nitrophenolate ion formed after addition of NaBH4, decreased in time in the presence of the Ni/NiOBDP or NiOBDP nanocatalysts confirming conversion of 4-NP. Concomitant increase in the intensity of the peak at 300 nm, corresponding to 4-AP, indicated the product formation. The reactions were carried out for 45 min for the Ni/NiOBDP catalyst and 55 min for NiOBDP . The rate constants, equal to 0.032 and 0.015 min−1 for Ni/NiOBDP and NiOBDP , respectively, were obtained assuming pseudo-first-order kinetics by fit the equation ln(At/A0) =  − kt where A0 and At represent the absorbance values of 4-NP at the start and at the time t. For comparison, the rate constant 0.043 min−1 was obtained for MCM-41–Ni [26]. Interestingly, in the same paper the potential activity of NiO in the hydrogenation of 4-NP was mentioned but not confirmed. Higher values of the rate constants, from 0.045 to 0.854 min−1 were reported for NiO/NiS catalysts of different composition [27].In our case, both catalysts contain mainly NiO but the presence of Ni(0) in Ni/NiOBDP enhanced the catalytic activity. Our results, confirming that NiO can catalyze the reduction of 4-NP, corroborate the studies of [27]. In contrast, other authors did not obtain 4-AP from 4-NP using NiO as the catalyst [36].The XRD (Fig. S2) and XPS (Fig. S1B) data of the catalysts recovered after 4-NP hydrogenation, did not confirm unequivocally the presence of Ni(0). However, analysis of the TEM images (Fig. S2) showed some changes which could suggest the presence of Ni(0) nanoparticles dispersed on the NiO surface. Therefore, we presume that a certain fraction of NiO might be reduced to Ni(0), which might be the true catalytically active species. In fact, the catalytic tests performed with reused catalysts clearly showed increase of the rate constants to 0.048 and 0.026 min−1 for Ni/NiOBDP and NiOBDP , respectively, which can be attributed to the increase of Ni(0) content (Fig. S4-S5).Transfer hydrogenation of APh occurred selectively with formation of PhE as the only product. The kinetic curves for the reaction catalyzed by Ni/NiOBDP at different temperatures are shown in Fig. 6 .As a result of the temperature optimization, it was found that the highest conversion of APh to PhE occurred at 80 °C with 63% conversion after 5 h reaction. The activation energy (E a) was calculated by the temperature dependence of the hydrogenation rate. (Fig. 7 ) It was assumed that hydrogenation of APh to 1-PhE is a first-order reaction (dCAP/dt = -kCAP, ln( C AP o / C AP )  = kt) and based on the Arrhenius relation, k = A·exp(-E a/RT), the activation energy was determined to be 88 kJ/mol (21 kcal/mol). This value of the activation energy is slightly higher than that reported for Ni–B–P/SiO2 catalyst (50.73 kJ/mol) [22].After optimization of the reaction conditions, NiOBDP was used as the catalyst. We found that in the first 3 h the reaction is slower than with Ni/NiOBDP and after 3 h the conversion was 39%. However, the final conversion of APh obtained after 5 h was similar for both catalysts (Fig. 8 ). After the reaction, the NiOBDP and Ni/NiOBDP nanocatalysts were separated and reused under the same conditions without any loss of their catalytic activity. The conversion of APh was 56% for NiOBDP -R and 65% for Ni/NiOBDP -R.Next, both the recovered catalysts, NiOBDP -R and Ni/NiOBDP -R, were examined using TEM, XRD and XPS methods. Figure S3 shows only minor changes in the TEM images. However, the analyses of FFT (Fast Fourier Transform) allowed to determine d-spacings values (Fig. S6). Notably, the fringes of 0.207 and 0.205 nm for Ni/NiOBDP-R and NiOBDP-R, assigned to Ni(111), evidenced the presence of Ni(0) [37].Furthermore, both samples of the used catalysts were examined by XPS with Ar+ ion etching. The spectra of the Ni 2p3/2 region (Fig. S7) exhibited a main peaks at 852.6 eV characteristic for Ni(0) [38]. Additionally, signals at 853.8 eV and a satellite at 861.03 eV were assigned to Ni2+ species. The peak at 856.72 eV is attributed to Ni3+.XRD analysis of the Ni/NiOBDP-R sample indicated some decrease in the intensity of the (111) reflection of Ni(0) (Fig. S3A).In summary, two new NiO-based catalysts were synthesized by a calcination of Ni-MOF. Calcination of MOF in a N2/O2 (4:1) gas stream led to a mixed composite Ni/NiOBDP while thermal decomposition in an air atmosphere provided only NiOBDP. The presence of Ni(0), formed in the absence of any reducing agent, affected the catalytic activity and Ni/NiOBDP provided better results in the transfer hydrogenation than NiOBDP. Reduction of the –NO2 group in 4-NP with NaBH4 proceeded for both catalysts, however, it was incomplete even after 55 min for NiOBD P indicating insufficient number of active sites for formation of 4-AP. The Ni/NiOBDP catalyst reacted faster due to the presence of active Ni(0) centers. During the catalytic process the structure of both catalysts was modified and their activity increased. Reduction of NiO to Ni(0) by NaBH4 was probably slow because of the presence of a strong Ni-O bond [39], however, the TEM images of the catalysts recovered after the reaction suggest the presence of ultra-small Ni(0) NPs. Consequently, the recycled catalysts exhibited higher activity in the second run than in the first one, as it was evidenced by the increase of the rate constants. Our kinetic results corroborate well with these reported for MCM-41-Ni containing mostly NiO phase [26]. The similar value of rate constant, 0.045 min−1 was also obtained for the NiO catalyst [27]. In general, catalysts containing higher amounts of Ni(0) reacted faster [24,25]. Similarly, Ni/NiOBDP was more efficient in the transfer hydrogenation of APh to PhE than NiOBDP . In this reaction 63% of PhE was obtained after 5 h using 8 mol % of catalyst. The TOF value for this reaction, equal 1.6 h−1, is comparable to the TOF values presented for transfer hydrogenation of furfural using different NiO catalysts (0.4 – 2.6 h−1) [34]. Adam W. Augustyniak: Conceptualization, Investigation, Writing – original draft. Andrzej Gniewek: Funding acquisition, Writing – review & editing. Rafał Szukiewicz: Investigation. Marcin Wiejak: Investigation. Maria Korabik: Formal analysis. Anna M. Trzeciak: Conceptualization, Writing – review & editing, Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Andrzej Gniewek is grateful for financial support from the National Science Centre (NCN, Poland) with grant MINIATURA 2021/05/X/ST4/00402.Supplementary data to this article can be found online at https://doi.org/10.1016/j.poly.2022.116029.The following are the Supplementary data to this article: Supplementary data 1

The NiOBDP and Ni/NiOBDP composites obtained by calcination of NiBDP MOF (MOF = metal–organic framework; BDP = 1,4-Bis(pyrazol-4-yl)benzene) in air at 700 °C were characterized by XRD (X-ray diffraction), TEM (Transmission electron microscopy), XPS (X-ray photoelectron spectroscopy) and Raman spectroscopy. Both nanocatalysts presented high activity in selective transfer hydrogenation of acetophenone (APh) and 4-nitrophenol (4-NP) in mild conditions, using 2-propanol/NaOH or NaBH4/H2O as reducing agents. In both reactions, the catalytic activity of Ni/NiOBDP was higher compared to NiOBDP suggesting the participation of Ni(0) in the hydrogenation process.

CO2 and CH4 are not only typical greenhouse gases, but also important carbon-containing resources. The CH4-CO2 reforming to syngas process can combine the comprehensive utilization of CH4 and the resource utilization of CO2, which provides a technological route for the comprehensive utilization of carbon and hydrogen sources and the conversion of greenhouse gases on a large scale [18,40,49]. This process is also known as dry reforming of methane (DRM) because no water vapor is involved in the reaction. The DRM process can be used to treat industrial off-gases such as coke oven gas and choke gas containing CH4 and CO2, and is also suitable for the conversion of CO2-rich field gas and offshore natural gas resources [9,43]. Compared to water vapor reforming and partial oxidation reforming, the DRM process utilizes both CH4 and CO2, the two major greenhouse gases, and has both environmental and economic benefits [27,29,42,47]. Compared to water vapor reforming, DRM eliminates the need for an energy-intensive evaporation process and can be used in water-scarce regions. DRM is a strong heat absorption reaction, storing energy in the form of syngas, which can be used to store and transport energy. However, there are both opportunities and challenges, as CH4 and CO2 are chemically stable resource small molecules whose conversion is thermodynamically unfavorable [6,14]. The efficient activation and directional reorganization of the bonds depends to a large extent on the catalyst used [45].The synthesis of phosphides dates back to the seventeenth century, when transition metal phosphides were initially studied by hydrodesulfurization, deoxidation, and other reactions. Among them, Ni2P has the best catalytic effect on the hydrodesulfurization reaction due to the key role of the (100) crystal plane of Ni2P material as the catalytic active plane [25]. In 2005, Zhang et al. [48] used DTF to calculate the energy changes of [Ni-Fe] hydrogenase and other compound materials of Ni during hydrogen precipitation. It was found that the atomic structure between Ni and P on the active crystal plane of [Ni-Fe] hydrogenase and Ni2P is very similar, thus inferring that Ni2P also has a relatively strong catalytic activity for hydrogen precipitation reactions [4,12]. In 2013, researchers tested for the first time Ni2P catalysts synthesized by the solvothermal method and concluded that Ni2P NPs have excellent catalytic properties [34]. The catalytic activity of Ni2P NPs is substantially higher compared to other previous transition metal compounds. Since then a large number of CoP materials with different preparation methods and morphological structures have been much studied for use as catalysts [13,31,41,46]. In the same year, nanostructured CoP with different morphologies such as nanowires, nanosheets and nanoparticles by low-temperature phosphorylation reaction without organic solvent were prepared under Ar conditions [36]. However, designing and modulating metal phosphides with specific morphological structures and stability is not easy. For this purpose, MOFs were used as precursors to obtain transition metal phosphides with different morphological structures and less collapse [17,35]. By modulating the electronic structure in this way, the specific surface area of the catalyst becomes larger, the conductivity is improved, and the catalytic activity is greatly enhanced.It has been shown that different synthesis methods can affect the particle size and distribution as well as the crystalline shape of the material, resulting in differences in the properties of MOF materials [19,22,30]. The traditional synthesis methods include hydrothermal and solvothermal methods, but there are complex processes, high energy consumption, long reaction times and high requirements for instrumentation that need to be improved. Compared with traditional methods, electrochemical methods have the advantages of mild reaction conditions (room temperature and pressure), simple process, easy operation, short reaction time, and safety and environmental protection. Therefore, electrochemical synthesis techniques are widely used in the field of novel nanoparticle synthesis. In 2005, Mueller et al. [24] research workers from BASF Corporation reported for the first time the synthesis of MOF by electrochemistry. The principle is that the metal ions are first obtained by in situ anodic oxidation and then dispersed into a solution containing organic ligands and electrolyte. The organic ligands are deprotonated and combined with the metal ions to produce the MOF material.So far, there have been many research results on the synthesis of microcrystalline MOF powder or thin film by electrochemical methods. Yang et al. [39] synthesized flower-like MOF-5 with a more perfect crystal structure in a tunable ionic environment by electrochemical methods. Gascon et al.[1] synthesized a variety of typical MOF materials, such as HKUST-1 (i.e. Cu3(BTC)2), ZIF-8, MIL-100(Al), etc., by electrochemical methods, and investigated the effects of different parameters on the crystalline structure and particle size during the reaction. Similarly, Denayer et al. [38] used an electrochemical method to synthesize HKUST-1 materials with controlled particle size in a short time at room temperature by adjusting the ratio of ethanol and water in the solvent. Therefore, the synthesis of MOF materials by electrochemical methods is well regulated in terms of controlling the crystalline structure and particle size. In addition, the rapid synthesis reaction allows for continuous production, and these characteristics show important advantages in industrial settings. In this work, Cu3(BTC)2 with high yield, good crystal structure and pure components were synthesized by electrochemical method. The effects of different phosphating temperatures, phosphating ratios and phosphating methods on the performance of the synthesized cuprous phosphide were explored to explore the optimal synthesis conditions. After a series of characterizations, different catalysts were applied to the DRM reaction.1,3,5-H3BTC was purchased from Aladdin Technology Co. Tetrabutylammonium tetrafluoroborate (TBAFB) was purchased from Beijing Bailingway Technology Co. Methanol, ethanol (EtOH) and N, N-dimethylformamide (DMF) was purchased from Tianjin Comio Chemical Reagent Co. Copper electrodes (99 %) of 2 mm × 5 mm × 0.8 mm were used for both the anode and cathode of the electrochemical synthesis. NaH2PO2·H2O was purchased from Sinopharm Chemical Reagent Co.0.1 M 1,3,5-H3BTC, 0.05 M TBAFB were dissolved in 50 mL of methanol. The electrodes were inserted and electrolyzed at an applied voltage of 10 V for 2.5 h. The Cu3(BTC)2 powder was formed on the surface of the Cu electrode and fall off into the solution. The samples were then filtered and washed repeatedly with ethanol until the supernatant was clarified. The sample was dried in a drying oven at 60 °C for 12 h to obtain a sky blue sample of Cu3(BTC)2. Cu3(BTC)2 was activated at 230 °C under N2 protection for 2 h. The crystalline water adsorbed on Cu2+ was removed to obtain a catalyst with unsaturated metal coordination. The color of the Cu3(BTC)2 catalyst also changed to dark blue.Cu3(BTC)2 and NaH2PO2·H2O were placed in different proportions in two different small porcelain boats and placed in a tube furnace. The porcelain boat containing NaH2PO2·H2O was located at the upstream of the furnace. Under the protection of argon atmosphere, the boats were heated at 300 ℃∼400 ℃ for 2 h at a heating rate of 2 ℃/min. After allowing the tube furnace to cool naturally to room temperature, argon was turned off and the black powder was collected and named Cu3P-X (X is a different heating temperature). The ratio of Cu3(BTC)2 and NaH2PO2·H2O and the phosphorylation temperature were varied to explore the optimal reaction conditions for the synthesis of Cu3P-X catalyst.The catalyst activity was evaluated using H2 reduction and CH4-CO2 reforming reactions simultaneously in a fixed-bed reactor. Weigh 0.2 g of catalyst into a quartz tube reactor with ϕ = 8 mm and fix it with quartz wool. N2 at a flow rate of 40 mL/min was used as a shield gas. The temperature was purged at a rate of 10 ℃/min for 70 min to reach 700 ℃. After 700 °C, it was switched to H2 for reduction. After 2 h of reduction, the reaction was switched to reaction gas (CH4/CO2/Ar = 44.0/47.2/8.8, flow rate of 40 mL/min) for 6 h. The reaction was carried out at atmospheric pressure. Samples were taken every 30 min during the reaction. The export products were analyzed online by GC-TCD, and the conversion of the reactants CH4 and CO2. The selectivity of the products H2 and CO and H2/CO were calculated.The amount of carbon accumulation increases when the phosphating temperature reaches 400 °C. We examined the differences in the synthesis of catalysts in different solvents. The dissolution rate of copper metal flakes in the solvent was observed to be MeOH> EtOH> DMF during the experiment. Because the nature of the solvent affects the solubility of the organic ligand and the conductivity of the solution, this also results in differences in the specific surface area, pore volume and yield of the catalyst [15]. Table 1 shows the specific surface area, pore volume and yield of Cu3(BTC)2 catalysts synthesized in different solvents. It can be seen from Table 1 that the specific surface area and pore volume of the catalysts synthesized in MeOH were the largest with 800.3 m2/g and 0.411 cm3/g, respectively.During the electrochemical reaction, the electrolyte can reduce the resistance brought by the solvent and promote the dissolution of metal ions in the anode. By controlling the current density, the concentration of metal ions near the electrode can be adjusted [37]. Current density and voltage are correlated via system geometry and solution conductivity. Therefore, the electrolyte concentration and voltage regulation play a key role in the electrochemical reaction process [28]. When the voltage is fixed (10 V) and the concentration of electrolyte increases (0.01–0.1 M), high conductivity leads to high yield ( Fig. 1A). At a certain concentration of electrolyte (0.2 M), the increase in voltage (8–16 V) also brought about an increase in yield (Fig. 1B).The length of electrolysis time also has an effect on the yield. It can be seen from Fig. 1C that the yield increases with the increase of electrolysis time [32]. When the electrolysis time was less than 2 h, the yield of catalyst was basically linear with the electrolysis time. The increase of reaction time brings obvious yield changes. When the electrolysis time exceeded 2 h, the yield tended to be stable, indicating that the electrolysis reaction was basically completed after the electrolysis time exceeded 2 h. This is also evidenced by the change in current. At the beginning of the reaction, the anode dissolution rate is high and the initial current is high. As the reaction proceeds and the reactants are consumed, the current drops sharply. When the reaction time exceeded 2 h, the current was lower than 0.1 A, and the reaction rate decreased significantly. Fig. 2 A shows the XRD spectrum of the Cu3(BTC)2 catalyst. Among them, 6.6°, 9.5°, 11.6°, 13.4°, 14.6°, 15.0°, 16.5°, 17.4°, 19.0°, 20.2° are the characteristic peaks of Cu3(BTC)2, which correspond to (111), (200), (220), (222), (400), (331), (422), (333), (440), (442) crystalline planes of Cu3(BTC)2 [44], respectively. The sharp peaks in XRD indicate that the synthesized catalyst has a good crystalline structure. No peak of Cu2O (2θ = 36.4°) was observed in the XRD spectrum of the catalysts prepared by the optimized electrochemical synthesis method, indicating that the electrochemically synthesized Cu3(BTC)2 materials are pure and structurally intact [16]. Fig. 2B shows the thermogram of the Cu3(BTC)2 catalyst. As can be seen from the figure, Cu3(BTC)2 has three weight loss intervals. The volatilization of solvent molecules ethanol and water introduced during sample washing at 50–130 °C resulted in a mass loss of about 17.4 %. There is a continuous weight reduction process at 130–310 °C, about 25.3 %, mainly for the removal of water molecules adsorbed on Cu2+ coordination bonds in Cu3(BTC)2, thus obtaining free active sites. This is also the reason for the catalytic activity of the catalyst [21]. After the temperature exceeds 325 °C, there is a clear weight loss peak and the sample loss decreases sharply. This indicates that the organic ligand decomposes and the Cu-MOF structure collapses [20]. The decomposition was complete by about 400 °C, with 36.4 % mass remaining. The decomposition products are oxides of metallic copper, which indicates that Cu-MOF has good thermal stability below 310 °C. Fig. 3 A shows the SEM image of Cu3(BTC)2 catalyst. From the figure, it can be seen that the Cu3(BTC)2 particles are octahedral in shape with homogeneous crystalline phase and the particle size is in the range of 0.2–0.5 µm.In order to further investigate the molecular structure of Cu3(BTC)2 catalyst, Raman characterization was carried out to investigate the valence bonding pattern. Fig. 3B shows the Raman diagram of the catalyst. The characteristic peaks in the low frequency region 170–600 cm−1 in the figure are attributed to the Cu2C4O8 metal cluster in the Cu3(BTC)2 catalyst structure. Among them, 177 cm−1 is attributed to the stretching vibration of the Cu-Cu bond, and the attribution of the characteristic peak at 502 cm−1 has not been clearly confirmed [26]. The characteristic peaks in the high frequency region 730–1800 cm−1 are attributed to the organic component of the Cu3(BTC)2 catalyst structure. The characteristic peaks of 1461 and 1544 cm−1 can be attributed to the symmetric stretching vibration and asymmetric stretching vibration peaks of -COO, indicating the complete deprotonation of 1,3,5-H3BTC carboxyl group [10]. The peak located at 1005 cm−1 and 1610 cm−1 corresponding for the stretching vibration peak of C = C on the benzene ring [2]. The peak located at 827 cm−1 and 745 cm−1 are the bending vibration peaks of the C-H bond on the benzene ring and the benzene ring vibration peaks [33]. The above results indicate that the Cu3(BTC)2 catalysts synthesized by electrochemical methods are structurally sound.We used the hydrogen precipitation properties of the catalysts to first verify their catalytic activity. Fig. 4 shows the polarization curves of Cu3P hydrogen precipitation performance at different phosphorization temperatures. At j = 1 mA/cm2, the starting overpotential η0 is 79 mV, 89 mV, 97 mV, 125 mV and 103 mV, and at j = 10 mA/cm2, the overpotential η10 is 136 mV, 145 mV, 155 mV, 181 mV and 164 mV, respectively. The hydrogen precipitation performance of the prepared catalysts was reduced by both high and low phosphorylation temperatures [23].Subsequently, we performed DRM reaction tests on different catalysts and the results are shown in Table 2 and Fig. 5. At first, the Cu3P catalysts prepared at elevated temperatures have more excellent activity and stability. However, the activity of the catalyst did not keep increasing with the increase of temperature. After the DRM reaction at 700 ℃ for 6 h, the conversion of Cu3P-350 did not change significantly, and the stability was significantly higher than that of Cu3P-300 and Cu3P-325. In addition, the amount of carbon accumulation was significantly reduced, thanks to the dual physical and chemical domain-limiting benefits of Cu3P-350, which inhibits copper particle migration at high temperatures and keeps the catalyst highly stable. The final conversion of CH4 was 77.44 % and 66.41 % for Cu3P-350 and Cu3P-300, respectively, after 6 h of reaction. The final conversion rates of CO2 were 80.15 % and 77.47 %, respectively. In addition, the inverse water gas conversion reaction results in a higher CO2 conversion than CH4 conversion, so the H2/CO in the product is less than 1. After 6 h of reaction, the H2/CO of Cu3P-300 catalyst (0.94) was greater than that of Cu3P-350 catalyst (0.92). In addition to the reforming reaction, there are other side reactions such as reverse water gas conversion, CH4 cracking and Boudouard reaction [7]. The Gibbs free energy can be used to determine the possible side reactions at different reaction temperatures. The reverse water gas conversion always exists below 820 ℃, resulting in the DRM process of CO2 conversion is usually greater than CH4 conversion, and H2/CO ratio is less than 1. CH4 cracking can occur above 557 ℃. The Boudouard reaction needs to be lower than 700 ℃, so more carbon deposition will occur between 557 ℃ and 700 ℃. Previous works conducted thermodynamic simulations of the effects of temperature, CH4/CO2 ratio, reaction pressure, and other oxidants on the formation of carbon deposits [3,8]. They suggest that operating at temperatures above 850 ℃, low pressure and high CO2/CH4 ratios can achieve higher conversion rates and less carbon deposition. Fig. 6 shows the carbon combustion curves of different catalysts after 6 h of DRM reaction at a reaction temperature of 700 ℃. All catalysts produced a weight loss plateau from 0 to 100 ℃, which was caused by the evaporation of residual water in the catalysts that were not completely dried after the reaction. A clear weight loss plateau was produced at 400–600 ℃, corresponding to the obvious exothermic peak of the DSC curve, which was attributed to the combustion of carbon. It was found that all the carbon accumulation of Cu3P was less when the phosphorylation temperature was below 375 °C. The first step of DRM reaction is the adsorption of CH4. At low temperature, CH4 molecules with low kinetic energy adsorbed on the metal surface to form an intermediate state, and then further desorption or dissociation occurred. At higher temperatures, CH4 molecules tend to dissociate directly after adsorption on the metal surface. The cleavage of CH4 on the metal surface is one of the slower steps in the DRM reaction because the dissociation energy of CH3—H bond is as high as 439.3 kJ/mol. The total energy required for CHx—H bond dissociation depends on the catalytic system, and the selection of appropriate catalyst will be helpful for CHx —H bond dissociation. adford and Vannice [5] summarized the behavior of CHx species on various metals. BCHx tends to be located at the active site that causes it to form a tevalent form. CH2 is a bridge adsorption, while CH and C need to be attached to a vacancy with three or four adjacent sites. This hypothesis does not take into account the changes in metal surface structure caused by the adsorption of CHx species and the effects of adsorbed species at adjacent sites [11]. The amount of carbon accumulation increased when the phosphorylation temperature reached 400 °C, which was consistent with the DRM reaction activity and stability pattern.In this work, Cu3P was synthesized by direct phosphorylation at different temperatures using Cu3(BTC)2 as the precursor, and their catalytic activity was first verified using the hydrogen precipitation performance of the catalysts. It was found that both high and low phosphorylation temperatures reduced the hydrogen precipitation performance of the prepared catalysts. The Cu3P catalysts prepared at higher temperatures showed better activity and stability. However, the activity of the catalysts did not always increase with the increase in temperature. The most effective catalyst was Cu3P-350.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the Research Foundation of Education Bureau of Hunan Province, China (Grant Nos. 20A060, 21C0869, 21C0900).

The CH4-CO2 reforming reaction can realize the utilization of both CH4 and CO2, which is important to control the greenhouse effect and protect the environment. The development of low-cost, high-activity and high-stability reforming catalysts is the focus of research. This work used electrochemical methods to synthesize Cu3(BTC)2 in high yields. Cu3(BTC)2 was subsequently used as a precursor and Cu3P was synthesized by direct phosphorylation at different temperatures. Both high and low phosphatization temperatures can reduce the hydrogen precipitation performance of the prepared catalyst. In the CH4-CO2 reforming reaction, the Cu3P catalyst prepared at elevated temperature has more excellent activity and stability. However, the activity of the catalyst did not keep increasing with the increase of temperature. After the CH4-CO2 reforming reaction at 700 ℃ for 6 h, the conversion of Cu3P-350 did not change significantly, and the stability was significantly higher than that of Cu3P-300 and Cu3P-325. At the same time, when the phosphating temperature is lower than 375 ℃ Cu3P carbon accumulation are less. The amount of carbon accumulation increases when the phosphating temperature reaches 400 °C.

The so-called Guefoams (guest-containing foams) are a new family of recently developed materials that are attracting increasing attention due to their promising behaviour in several applications. Guefoams consist of multifunctional porous materials hosting granular or fibrous phases with specific functionality (guests) in open-pore foam (host) cavities [1–4]. There is no bonding between the host matrix phase (the phase that forms the porous skeleton) and the guest phases other than mere physical contact. For this reason, the entire surface of the guest phases is functional, and a fluid can flow through the Guefoam with a relatively low pressure drop. Guefoams were conceived to provide broader or newer functionalities to conventional foams or those incorporating new phases by full or partial embedding in the matrix phase. The Guefoams manufacturing process is simple and economically feasible for large-scale production, as it is based on the conventional replication method commonly used to produce most foams on the market today. The production consists of infiltrating a foam matrix precursor under gas pressure into preforms containing guest phases in the form of particles or fibres coated with NaCl, which is dissolved after solidification of the matrix precursor [5,6]. As an example, Guefoams with an aluminium matrix phase containing both steel and activated carbon particles as guest phases have been reported for the pre-concentration and desorption of volatile organic compounds (VOCs) by rapid magnetic induction heating [3]. Recently, magneto-inductive carbon matrix Guefoams with embedded iron nanoparticles and activated carbon as guest phase have been shown to be effective VOC preconcentrators [4].Conventional foams are used as catalyst supports, whose intricate interconnected porous structure enables the development of a higher specific surface area per unit volume than honeycomb monoliths. The interaction of fluids with active phases loaded on foams is generally greater than in honeycombs, since fluids usually adopt laminar flow in honeycomb channels, while foams favour a turbulent regime, which improves mass and heat transfer [7]. In addition, foams also undergo a moderate pressure drop, which can be tailored by controlling the size of the interconnection windows between pores.However, given the complex pore space of foams, loading of active catalytic phases is often non-trivial for these materials [8]. Some examples of loading foams with active phases have been described by electrochemical or hydrothermal processes for Ru-Ce/Ni [8], Ni-SiO2/GO-Ni [9], BiFeO3/Ni [10], Cu(NP)s/Gf-Ni [11], Co-W-B/Ni [12], Ni/SiC [13], ZSM-5/SiC [14] and MOF/SiC [15]. Recently, open-pore graphite foams derived from mesophase pitch have been developed in which TiC nanoparticles are conveniently distributed in two positions (on the pore surface and in the foam struts) [16]. Depending on their location, the TiC nanoparticles fulfil two different roles: those at the struts catalyse the pitch graphitization process and allow high thermal conductivity to be achieved, while those anchored on the pore surface can serve as metal supports for catalytic purposes. These particles at the pore surface are partially embedded in the matrix phase. A major drawback of the embedded phases is that they lose part of their surface area, which significantly reduces their surface functionality.Despite the interesting features of foams as catalyst supports, the use of Guefoams in heterogeneous catalysis has not been developed yet. These materials have great design potential, which makes them even more interesting than conventional foams as catalytic supports. Its processing, besides being simple and economically feasible for large-scale production, avoids the difficult step of active phase loading. In addition, Guefoams can be prepared with a metal matrix and their high thermal conductivity promotes heat transfer from or into the catalyst phases. This contributes to the catalytic conversion rate of a reactor having a more homogeneous radial distribution than that of a particle bed, allowing larger reactors to be built with higher catalytic efficiency. Moreover, Guefoams have the advantage of being materials that can be customized by varying design parameters such as the fraction of pores occupied by the guest phase(s) or the volume fraction that they occupy in the porous cavity, which allows modification of properties that are important for fluid dynamic applications, such as permeability or relative pressure drop [3,4]. This design versatility enables the development of catalytic reactors with higher efficiency and consequently lowers operating costs, in line with near-term expectations for heterogeneous catalysis in the context of greener and more energy-efficient chemical processes. Guefoams have such high design potential for catalytic purposes that they can successfully meet challenges that numerous scientists have proposed as research topics in the field of heterogeneous catalysis for the coming years. By using Guefoams, scientists could make scientific progress in designing reactors for multi-catalysis or simultaneous tandem catalysis by combining guest phases with different catalytic functions and differentiated localization [17,18].The preparation and applicability of Guefoams as heterogeneous catalysts were investigated in this study. Alumina particles loaded with a Ni/CeO2 active phase were used as a guest phase hosted in a Al-Si foam. The resulting material was tested for methane production by CO2 hydrogenation. This reaction is of practical relevance to reduce CO2 emissions to the atmosphere and to produce a valuable fuel that can be easily distributed through the existing natural gas network [8,9,19–25]. This reaction will become particularly important in a new energy scenario, where H2 will be massively obtained from renewable energy sources, as CO2 methanation will be a chemical route for energy storage. The selected active phase is one of the most efficient noble metal-free CO2 methanation catalysts of practical interest [26]. The aim of this study is therefore not to investigate the methanation reaction or the active phase behaviour, as this has already been studied by several researchers [20,27–29], but to use this reaction and the active phase as a proof of concept to evaluate the potential applicability of the novel Guefoam catalyst. A multidisciplinary team with metallurgical and catalytic background was necessary to optimize the manufacturing process and avoid thermal and chemical degradation of the active phase.A Ni/CeO2/Al2O3 active phase was prepared with the following nominal composition: 5 wt% Ni + 47.5 wt% CeO2 + 47.5 wt% Al2O3. The composition was confirmed by ICP-AES. Commercial α-Al2O3 pellets were ground and sieved to produce Al2O3 particles of 0.75–1 mm, which were first impregnated with cerium (III) nitrate hexahydrate. After calcination at 500 °C for 2 h (heating at 5 °Cmin−1), nickel (II) nitrate hexahydrate was further impregnated, dried and calcined at 450 °C for 2 h (heating at 5 °Cmin−1). The above three raw materials were supplied by Alfa-Aesar (Kandel, Germany).The catalyst particles were coated with NaCl (99.5 wt%; Panreac Química S.L.U., Barcelona, Spain) by spray coating with a 20 wt% NaCl-water solution, as described elsewhere [1–4]. The coated particles were sieved and fractions with a diameter of 1.7–2.2 mm were used for the Guefoam catalyst preparation.Replication method was followed to prepare the Guefoam catalyst [30,31]. The coated guest phase particles were packed with the help of vibrations into a graphite crucible with an inner diameter of 26 mm and a height of 100 mm, which had been previously sprayed with a BN coating (ZYP Coatings Inc., Oak Ridge, USA) to facilitate demoulding (see more details in [32] for the packing procedure). To prevent the particles from moving or floating during metal infiltration, a 2 mm thick graphite disk with holes of about 0.5 mm was properly fixed to the top of the compacted preform. A eutectic aluminium–silicon alloy (Al-12 wt% Si), prepared with commercially pure aluminium (99.95 wt%) and silicon powder (99.9 wt%) both purchased from Alfa Aesar (GmbH & Co KG-Karlsruhe, Germany), was used as the metal matrix precursor. A solid piece of aluminium alloy was placed on top of the graphite disk and the crucible was then inserted into a gas pressure infiltration chamber [33,34]. A vacuum up to 0.2 mbar was applied, with a heating rate of 4.5 °Cmin−1 up to 665 °C. After 10 min at constant temperature, the vacuum was closed and the chamber was pressurized with 0.8 bar argon to infiltrate the packed preform with liquid metal.After infiltration, the chamber was rapidly cooled at 50 °Cmin−1 to solidify the metal. The solid was extracted by removing the surrounding excess metal. This yielded a piece 25 mm in diameter and 45 mm in length containing 350 mg of active phase particles. The sodium chloride coating was removed by dissolution with a pressurized water flow, as described in [35]. The result was an interconnected Al-Si alloy pore structure in which the guest phase particles are located inside the porous cavities without chemical or physical bonding. The Guefoam catalyst was finally calcined at 500 °C for 4 h, and catalytic tests were performed with the Guefoam catalyst before and after this heat treatment.The guest phase particles and the surface morphology of their NaCl coating were characterized using a SEM-Hitachi S3000N scanning electron microscope operating at variable voltage. The spatial distribution of the Ni active phase and the composition of the guest phase particles were analysed using the same microscope equipped with a Bruker XFlash 3001 X-ray detector for point and map analysis (EDX).Geometric parameters (circularity and aspect ratio) were determined from image analysis. Circularity is defined as 4⋅π⋅area/perimeter2, where 1.0 represents perfect circularity. The aspect ratio is the ratio between the average major and minor axes of the particles. These last two parameters were determined from measurements of over 300 particles.The density of the active phase particles was measured by densitometry using dichloromethane (density = 1.330 gcm−3 at 25 °C) according to the ASTMD854 standard. The use of dichloromethane avoids the dissolution of the NaCl used as coating of the active phase particles.Thermal conductivity was experimentally determined by a set-up assembled at the University of Alicante laboratories in compliance with the international standard ASTM E-1225-04, based on a relative steady-state (equal-flow) technique [35–37]. Each sample, with cylindrical geometry, was placed between two blocks. The bottom of the sample remained in contact with a cooled cylindrical block (refrigerated by a room temperature water flow) and the top was in contact with a brass reference block connected to a 70 °C water bath. Two sets of thermocouples were connected to the sample and three more to the brass reference so that the temperature gradients required to estimate thermal conductivity could be measured with an uncertainty of less than ±5%.XPS characterization was performed in a K-ALPHA Thermo Scientific device using Al-Kα radiation (1486.6 eV) and a twin crystal monochromator that yields a focused X-ray spot with a diameter of 400 μm at 3 mA × 12 kV. The binding energy scale was adjusted by setting the C1s transition to 284.6 eV.CO2 methanation experiments were carried out in a cylindrical reactor with a 64% H2 + 16% CO2 gas mixture balanced with N2 (100 mlmin−1 total flow and atmospheric pressure). The experiments were performed with a packed bed of the active phase particles between quartz wood plugs and with the Guefoam catalyst. In both cases, the amount of catalyst particles was 350 mg. Gas composition was monitored using specific AwiteFLEX COOL gas analysers, with NDIR, electrochemical and TCD detectors for CO, CO2, CH4, O2 and H2. The catalysts were pretreated with 50% H2/N2 at 500 °C for 1 h and cooled to room temperature under inert gas. Then the reaction mixture was fed into the reactor and the gas composition was measured under steady-state conditions at selected temperatures from room temperature to 500 °C.The gas flow pressure drops generated by the novel Guefoam catalyst and a packed bed of active phase particles were determined experimentally using the setup described in Fig. 1 .The permeability (k) can be derived from the Darcy-Forchheimer equation, which relates the fluid velocity (v) and the pressure drop (ΔP/ΔL). This equation contains the viscous term of Darcy’s law and the inertial effects generated by the flow in the porous medium [5,37]: (1) Δ P Δ L = μ k v + ρ C i v 2 where µ and ρ are the dynamic viscosity and density of the fluid (taken as 1.85 × 10-5 kgm-1s−1 and 1.184 kgm−3 at 25 °C, respectively). Ci refers to the inertial coefficient. The viscous loss (v·µ/k) is linear with velocity and includes a viscous resistance coefficient of 1/k, which is the inverse permeability. The inertia term ( ρ C i v 2 ) accounts for the nonlinear pressure behaviour as a function of fluid flow by including an inertial resistance coefficient Ci.Temperature gradients within the Guefoam and the packed bed allow understanding how quickly a radial section of the reactor approaches the minimum and maximum temperatures of catalytic conversion, which mainly depends on the permeability and thermal conductivity of the material as well as fluid velocity. Radial temperature gradient calculations were performed with the ANSYS Fluent software package using a computational fluid dynamic (CFD) approach. The software was employed to simulate the above materials using simplified porous media configuration under local thermal non-equilibrium (LTNE) conditions, which assumes the difference between fluid and solid temperatures in two energy equations. Real-dimensioned reactors of 30 cm length (L) and 15 cm diameter (d) with the composition and pore volume fraction of Guefoam and packed bed were modelled following the computational domain schematic diagram and boundary conditions shown in Fig. 2 . The system, considered to be at a constant temperature of 400 °C, was subjected to a fluid flow of 6 × 103 lmin−1. The fluid was deemed incompressible with analogous air physical properties and inlet temperature of 180 °C. Heat losses due to convection or radiations were assumed to be negligible. The governing energy equations are as follows [38,39]: Fluid energy equation: (2) ρ f C f v · ∇ T f = ε K f ∇ 2 T f + h sf a v T s - T f Solid energy equation: (3) 1 - ε K s ∇ 2 T s + h sf a v T f - T s = 0 where ε is the pore volume fraction of the porous material, C is the specific heat, K is the thermal conductivity, T is the temperature, hsf is the interfacial heat transfer coefficient and av is the interfacial area density. The subscripts f and s refer to the fluid and solid phases, respectively.CO2 methanation is an exothermic reaction that typically proceeds between 200 °C and 500 °C, depending on the catalyst and experimental conditions [19]. Its enthalpy is −165 KJmol−1 at 25 °C, but it decreases rapidly with temperature, becoming virtually nil at high temperatures close to its common operating limit [40]. In this context, it can be assumed that the heat released by the methanation reaction does not substantially alter the local temperature conditions of the catalytic monoliths, since the conversion rate at low temperatures is low and therefore the heat released can be considered negligible. At high temperatures, where the conversion rate is high, the heat released is also negligible due to the near zero enthalpy. Therefore, the effect of the heat released by the methanation reaction was not included in the modelling calculations of the temperature profiles. Fig. 3 shows several images of active phase particles before and after NaCl coating. Fig. 3a and b show that the uncoated active phase particles have an angular geometry, whereas they become more spherical after coating with NaCl (Fig. 3c). This is confirmed by the circularity values listed in Table 1 , which increase from 0.68 to 0.86 after NaCl coating, and by the aspect ratio values, which also increase from 0.68 to 0.89. Table 1 also compiles the densities determined by densitometry with dichloromethane. Fig. 3b displays an EDX-Ni mapping on the surface of an active phase particle, which confirms the homogeneous spatial distribution of the active phase on the surface of the alumina particles. The thickness of the NaCl coating can be seen in Fig. 3d, which shows a profile after a controlled fracture. It should be noted that, since the active phase particles have an angular morphology, the thickness of the NaCl coating around each particle is not homogeneous. Using image analysis, the average diameters of both the uncoated and coated active phase particles were measured (see supplementary material for more information).For the characterization of Guefoams, the guest loading (GL) and guest occupation (GO) parameters are essential: (4) GL = n g N p (5) GO = v g V p where ng is the number of pores hosting a guest phase, Np is the total number of pores, vg is the average volume of guest phases and Vp is the average volume of hosting pores.GL represents the fraction of pores hosting a certain type of guest phase. This parameter is determined by the relative ratio between the amount of NaCl-coated active phase and the amount of massive NaCl spheres that do not contain active phase. In the present study, all the pores of the foam material were intended to host an active guest phase particle, i.e., the GL parameter should be as close as possible to 1 (or, as a percentage, 100%). The experimental results showed that the GL = 97% (supplementary material).The GO parameter is the ratio between the average volumes of active phase particles and the cavities containing them. For fully spherical active phase particles and cavity geometries, GO=(r/R)3, where r and R are the average radii of the uncoated active phase particles and NaCl-coated active phase particles, respectively. The calculation of the GO parameter in non-spherical geometries such as the current angular geometry of the active phase is not so straightforward. This parameter was determined in two independent ways and resulted in a percentage value of 42–43%, as explained in the supplementary material.Although constant GO and GL values are employed in this work, given the proof-of-concept nature of the present research, both parameters can be modified to significantly alter the fluid dynamic behaviour of the fluid passing through the material. The effect of GO and GL on critical parameters such as permeability and relative pressure drop was demonstrated in [3].The general structure of a Guefoam is depicted in Fig. 4 a. Guefoams consist of an interconnected (or open-pore) foam material hosting freely moving guest phases in their cavities, since there is no chemical or physical matrix-guest bond. The dimensions and geometry of the Guefoam herein fabricated can be seen in the photograph in Fig. 4b. Fig. 4c provides an optical micrograph of the cavities containing the guest phases (Ni/CeO2/Al2O3 active phase particles).Experiments on CO2 methanation were performed with a packed bed of active phase particles and with the Guefoam catalyst. The results of CO2 conversion and CH4 selectivity are shown in Fig. 5 .The as-prepared Guefoam catalyst obtained after NaCl removal with water showed no activity (Fig. 5a; green triangles), and our hypothesis was that the active phase was poisoned by chlorine, consistent with what other authors have found about the inhibition of various active phases by chlorine species [43–45]. This hypothesis was confirmed by XPS characterization, and a detailed analysis is described in the next section. In order to reverse this chlorine poisoning, the monolith catalyst was calcined at 500 °C for 4 h, and then successful catalytic activity was achieved in CO2 methanation. The onset reaction temperature was 225 °C, and thermodynamic equilibrium was reached at 500 °C. CH4 selectivity was 100% up to 400 °C, and few CO was detected above this temperature, with selectivity dropping to 80% at 500 °C. This catalytic behaviour was compared with powder of Ni/CeO2 active phases reported in the literature, as well as with our own previous catalytic results [26], and the onset CO2 methanation temperature obtained with the novel Guefoam catalyst prepared in this study is similar to that previously measured for Ni/CeO2 powders. However, thermodynamic equilibrium was reached with Ni/CeO2 powder at 350 °C, while this temperature was shifted to 500 °C in the current study. This indicates that the catalytic activity of the Guefoam catalyst is in some way lower than that of the active phase powder, and this is the penalty to be paid for the novel supported catalyst. Nevertheless, the catalytic activity of the novel monolith is high enough to be properly used.A catalytic experiment was performed with the active phase particles subjected to the same treatments used for the preparation of the monolith catalyst (NaCl coating, washing and calcination) but without the AlSi foam support. The catalytic behaviour is similar for the particles and for the Guefoam catalyst, proving that the shape of the catalytic bed (particle or monolith) does not affect the catalytic behaviour under the experimental conditions of these tests. That is, the Guefoam catalyst is able to perform the same as the packed bed made of the same active phase particles, but with the benefits of a structured piece (easy manipulation, improved thermal conductivity due to the metal matrix, etc.).To analyse the effect of chlorine on the catalytic performance of the active phase, XPS analysis was performed on the active phase particles in the as-prepared state, after NaCl coating and washing, and after further calcination at 500 °C for 4 h. Fig. 6 shows the spectra obtained in different energy regions corresponding to Cl2p, Ni2p, Ce3d and O1s.The presence of chlorine on the active phase was confirmed after NaCl coating and water washing (Fig. 6a), which is consistent with our hypothesis of chlorine poisoning. Calcination at 500 °C for 4 h removed part of the chlorine, and successfully activated the active phase, resulting in suitable catalytic activity.In addition to the chlorine changes, additional changes were observed on the active phase surface after NaCl coating, washing and calcination. Fig. 6b shows the Ni2p region, and differences in the position of the main peak after the different treatments are distinguished. The detailed interpretation of Ni2p spectra is still a matter of debate, but information about the electronic environment of the nickel species can be obtained from the position of the main peak [46–49]. It has been reported that the main peak of metallic nickel appears at around 853 eV, while cationic species of Ni2+, such as NiO or partially hydrated oxides, usually appear at 854 eV and higher energies. The values measured in our spectra confirm the presence of cationic nickel in all cases, but there is a shift in the position of the peak from 855.2 eV in the as-prepared particles to 855.9 eV after NaCl coating and washing. This 0.7 eV shift evidences a lower negative charge density on Ni2+ cations after NaCl coating and washing, consistent with a Ni2+-Cl- interaction. The Ni2+ main peak is further shifted to 856.4 eV after calcination treatment, indicating a significant change in the electronic environment of the nickel cations, which could be consistent with the partial substitution of chlorine by oxide anions.The electronic density of the cerium cations (Fig. 6c) also changes before and after the different treatments. A mixture of Ce3+ and Ce4+ cations is usually found on ceria, and the Ce3d spectra combine the contributions of both types of cations, as shown in the graph. The percentage of Ce3+ cations slightly decreases from 21 % to 18 % after NaCl coating and washing. This oxidation is consistent with the chlorine-poisoning hypothesis and proves that chlorine affects not only nickel cations but also cerium cations. Stabilization of the oxidised state of cerium cations by the presence of chlorine probably inhibits the reversible Ce3+/Ce4+ redox cycle required for proper catalytic activity. The percentage of Ce3+ cations increases significantly from 18% to 27% after calcination treatment. In accordance with the nickel behaviour, calcination treatment seems to replace chlorine by oxide also on ceria, allowing the partial reduction of Ce4+ to Ce3+, which would explain the recovery of catalytic activity.The increase in Ce3+ content upon calcination is consistent with the O1s spectra (Fig. 6d). The O1s spectra can be deconvoluted into different contributions. The peak at the lowest energy is attributed to lattice oxygen, while other peaks at higher energies are attributed to surface species such as carbonates, hydroxyl groups and chemisorbed oxygen on ceria vacancies associated with Ce3+ cations. The position of the oxygen lattice peak shifts from 428.7 eV to 429.3 eV after NaCl coating and washing, which may be a consequence of the presence of a highly electronegative element such as Cl. After thermal treatment, the position of this lattice oxygen peak shifts by 0.8 eV (from 429.3 eV to 530.1 eV), which is also consistent with the partial removal of chlorine. Special attention must be paid in this case to the surface oxygen species, whose contribution to the O1s spectra increases dramatically after calcination. This is consistent with the appearance of oxygen vacancies due to the increase in the proportion of Ce3+ cations during calcination, which are filled by chemisorbed oxygen.In summary, XPS analysis confirms the presence of chlorine on the active phase after NaCl coating and washing, which is expected to inhibit the redox processes of the nickel and cerium cations required to achieve adequate catalytic activity. The negative effect of chlorine can be partially reversed by calcination at 500 °C for 4 h, and XPS analysis provides evidence for the substitution of chlorine by oxide with the expected restoration of redox behaviour.To further analyse the potential advantages of the novel Guefoam materials in catalytic applications, the thermal conductivity measurements are discussed in this section.The measured thermal conductivities of the materials herein evaluated are summarized in Table 2 . The thermal conductivity of the Guefoam catalyst is 43 Wm−1°C−1, which is significantly higher than that of the packed bed obtained by compacting the active phase particles, with a value of 1.7 Wm−1°C−1. The Guefoam value of 43 Wm−1°C−1 is consistent with the thermal conductivity measured for an analogous guest-free foam (i.e. GL = 0%). This confirms that the absence of chemical bonding between the matrix and the active phase is responsible for the nil contribution of the guest phases to the overall thermal conductivity of the material. Therefore, this conductivity value is consistent with the estimates that can be made with analytical models for metal foams [35–37], which state that. (6) K = K s 1 - ε n where Ks is the thermal conductivity of the solid (matrix) and n is a parameter dependent on the pore geometry (n = 1.5 for spherical pores).Considering that ε = 0.59 (value of the pore volume fraction for a guest-free foam) and that the thermal conductivity of the Al-12 wt% Si alloy (Ks) was measured to be 179 Wm−1°C−1, the value estimated by Eq. (6) is 47 Wm−1°C−1 when n is assumed to be 1.5, which is in perfect agreement with the measured value, considering the experimental measurement error.The pressure drop across the Guefoam and packed bed was determined using the setup described in Fig. 1. Fig. 7 a shows the resulting pressure drop curves.As expected, the pressure drop caused by the packed bed of active phase particles is substantially higher than the pressure drop generated by the Guefoam. Unlike other foam materials, Guefoams can experience different pressure drops depending on the orientation (Fig. 7a). This is because guest phases can block the interconnecting windows and force the fluid to take less direct paths, resulting in higher pressure drops (Fig. 7b). The pressure drop of the Guefoam was measured in the horizontal orientation and in the vertical position with gas flow from top to bottom and from bottom to top. Fig. 7a confirms, as expected in view of Fig. 7b, that the lowest pressure drop is achieved in the horizontal configuration. In this horizontal arrangement, gravity acts perpendicular to the fluid passage so that the guest phase particles lean on the lower cavity walls, away from the most direct path defined by the fluid as it passes through the material, thus offering less resistance to the fluid passage. In contrast, gravity and fluid passage are parallel in vertical arrangements and the guest phase particles lean on fluid-trajectory aligned cavity regions, sometimes even blocking the interconnecting windows that define the easiest fluid path and offering greater resistance to fluid passage.The permeability values (k) can be obtained from the curves in Fig. 7a by quadratic fits according to Eq. (1). The results are summarized in Table 2, which shows that the Guefoam catalyst is 1.7–2.7 times more permeable than the packed bed.The design of catalytic materials in heterogeneous catalysis is a complex process that must be considered from a holistic perspective. The success of a catalytic material in industry is not constrained to its catalytic activity, but encompasses other considerations, such as its pressure drop, which represents a large portion of the energy required to function as a catalyst, or its thermal conductivity, which allows for considerations of its scalability to useful dimensions. In this work, these considerations have been further explored in order to show a performance comparison between the studied systems, i.e. Guefoam and particle bed.In [50], conversion efficiency and pressure drop were related in a parameter defined as catalytic performance index (Ip) to compare different reactors using the following equation (slightly rewritten to replace ΔP with ΔP/ΔL to normalize the pressure drop with sample length): (7) I p = - ln ( 1 - η ) Δ P / Δ L being η the conversion efficiency. Fig. 8a shows the Ip values obtained from the experimental characterization as a function of the conversion temperature. The graph reveals that the Guefoam (washed and calcined) has the best performance as a catalytic reactor at temperatures above 350 °C, since the conversion efficiency/pressure drop ratio results in the highest Ip index.The thermal conductivity and permeability of a catalytic material affect the heat transfer into or from the material, thus influencing the temperature gradients, which can ultimately have a significant impact on the catalytic performance of materials with large dimensions. In order to analyse the temperature gradients inside the Guefoam catalyst and the packed bed, a CFD modelling of the temperature profiles was performed. Fig. 8b and c shows radial temperature profiles on three y-z planes at 1, 15 and 29 cm inlet length of the Guefoam and the particulate bed (the dimensions and fluxes taken, which were explained in Section 2.7, correspond to those of a possible industrial application for the methanation reaction considered).Significant differences are observed in the temperature gradients of the two systems considered. As expected, the temperature for both catalyst beds is higher near the outer walls than in the middle of the reactor, since the model assumes that heating is provided by an external heat source. The lowest temperature gradients were obtained for the Guefoam catalyst due to its metallic nature (Fig. 8b), while the highest gradients were obtained for the packed bed of active phase particles (Fig. 8c), which consists of metal oxides (mainly alumina and ceria) with poor thermal conductivity.Thus, the temperature at the centre of the y-z plane, located 15 cm from the inlet, is 250 °C for the Guefoam catalyst, while it is significantly lower for the packed bed. Considering, for instance, that the onset temperature for CO2 methanation is about 250 °C (see Fig. 5), the effective heating of the Guefoam catalyst allows a larger volume of the catalyst bed to have temperatures above the threshold required for the reaction.The preparation and use of Guefoams as heterogeneous catalysts were investigated in this study. The general structure of the prepared Guefoam catalyst consists of an interconnected (or open-pore) Al-Si foam hosting freely moving guest phases in their cavities, since there is no chemical or physical matrix-guest bond. Alumina particles loaded with the Ni/CeO2 active phase were used as the guest phase. A eutectic Al-12Si alloy was chosen for the foam body to lower the melting temperature and prevent thermal sintering of the active phase during liquid metal infiltration.CO2 methanation experiments were performed using the novel Guefoam catalyst as a reaction test. The obtained activity and CH4 selectivity (close to 100%) were similar to the values obtained with a packed bed of the same active phase particles, but with the benefits of a structured reactor. Guefoam manufacture requires the coating of the active phase particles with a NaCl shell. The salt is dissolved once the Al-Si alloy is infiltrated to obtain the foam. As shown by XPS characterization, the presence of chlorine anions poisons the active phase and inhibits the catalytic activity. A critical step in the Guefoam synthesis is the final calcination (500 °C; 4 h) to replace the chlorine with oxide anions, which only then ensure the catalytic activity.The thermal conductivity of the Guefoam catalyst is significantly improved with regard to the packed bed of active phase particles. This reduces the temperature gradients in the catalytic reactor, as demonstrated by computational fluid dynamic modelling.Pressure drop measurements showed that the permeability of the Guefoam catalyst is up to 2.7 times higher than that of the packed bed, resulting in a better catalytic performance index (Ip), especially at temperatures above 350 °C.Beyond the specific conclusions drawn in the present study and to comment on the future perspectives that can be achieved with Guefoams, the authors foresee a great potential of these materials in the context of new challenges in heterogeneous catalysis, such as the design of multi-catalytic reactors or for tandem reactions by combining different guest phases as differentiated catalytic centres. The versatility in varying the GL and GO parameters in the different cavities of a material is a tool with enormous potential for the design of future catalytic reactors adapted to specific working conditions and in which the catalytic performance can be optimised to values adapted to particular needs.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Financial support from the Spanish Agencia Estatal de Investigación (AEI), the Spanish Ministry of Science and Innovation and the European Union (FEDER and NextGenerationEU funds) [projects MAT2016-77742-C2-2-P, PDC2021-121617-C21 and CTQ2015-67597-C2-2-R] and the Conselleria d'Innovació, Universitats, Ciència, i Societat Digital of theGeneralitat Valenciana[projects GVA-COVID19/2021/097 and PROMETEO/2018/076, PhD grant of C.Y. Chaparro GRISOLIAP/2017/177 and contract of E. Bailón APOSTD/2019/030]. L.P. Maiorano also acknowledges the financial support from the University of Alicante through grant “ Programa Propio para el fomento de la I + D + I del Vicerrectorado de Investigación y Transferencia de Conocimiento ” (UAFPU2019-33).The raw data required to reproduce these findings are available to download from [https://zenodo.org/deposit/5719300]. The processed data required to reproduce these findings are available to download from [https://zenodo.org/deposit/5719300].Supplementary data to this article can be found online at https://doi.org/10.1016/j.matdes.2022.110619.The following are the Supplementary data to this article: Supplementary Data 1

The preparation and use of Guefoams as heterogeneous catalyst is reported. The Guefoam catalyst consists of an open-pore Al-Si foam that accommodates a freely mobile guest phase (Ni/CeO2/Al2O3 particles) in its cavities, with neither a physical nor a chemical matrix-guest bond. A eutectic Al-12Si alloy was used as a low-melting matrix precursor to prevent thermal sintering of the active phase during liquid metal infiltration. CO2 methanation was chosen as the reaction test. The activity and CH4 selectivity (close to 100%) achieved with the Guefoam catalyst were similar to those obtained with a packed bed of the same active phase particles, but with the advantages of a structured reactor such as robustness and ease of handling. The thermal conductivity of the Guefoam catalyst is significantly improved with regard to the packed bed of active phase particles, which reduces the temperature gradients in the catalytic reactor, as demonstrated by computational fluid dynamic modelling. Since the permeability of the Guefoam catalyst is 2.7 times that of the packed bed, the pressure drop caused by the passage of a fluid through the novel material is reduced, resulting in a significantly higher catalytic performance index than the packed bed.

The global demand for energy fossil fuels has increased tremendously due to the increase in economic growth and pollution; notably, nonrenewable fossil fuels are continuously depleting. Several research emphases have focused on developing alternative routes for bioenergy production, such as the use of agricultural residue after harvesting as an abundant renewable natural resource [1,2]; this has led to the development of a variety of techniques by which these products can be recovered. These techniques have immense potential in regard to the production of biofuels while minimizing negative effects on the environment. Pyrolysis is an increasingly promising thermochemical technology for biomass conversion and can produce biofuels (char, bio-oil, gases) in the absence of oxygen [3–6]. The pyrolysis of lignocellulosic biomass is based on the thermal decomposition of its components. In particular, woody biomass is divided into hardwood and softwood, which substantially differ in the concentrations of cellulose, hemicellulose, lignin and various compounds; for example, hardwood typically has less lignin (∼23–30%) than softwood and non-wood (∼26–34%) [7,8]. In particular, lignocellulosic components show different thermal degradation results, decomposing at low temperatures. Cellulose demonstrates thermal resistance over a peak decomposition range, while lignin degradation also occurs over a broad range after reduced volatilization when compared with holocellulose. In contrast, the initial degradation of softwood begins at lower temperatures, and both the decomposition of hemicellulose and the cellulose range are broader than that of woody biomass, causing different thermal decomposition results [9–11].Among the bio-oils produced from pyrolyzed woody biomass, although they can be used as a burner fuels or precursor chemicals [12], their physicochemical properties should be further modified before being used due to their acidity, viscosity and high oxygen content, which make them have very low thermal stability. Several studies have suggested enhancing the quality of bio-oil, one of which is the catalytic pyrolysis of biomass [13–15], deoxygenation, decarboxylation, decarbonylation reactions and the secondary reaction of tar cracking take place, resulting in an increasing hydrocarbon content and decreasing oxygen content which contributes to stability, high heating values and low acidity in a bio-oil yield. HZSM-5 shows enhanced catalytic activity to remove oxygen via decarbonylation and decarboxylation and the active acid sites of the catalysts continuously perform at high temperatures; however, some disadvantages are its rapid catalytic deactivation due to the active sites being blocked by carbonaceous coke along with dehydration reactions promoting an increase in the water component of the produced bio-oil [16,17].Another catalyst candidate for deoxygenation has been alternatively emphasized for use due to its inexpensive preparation from natural elements and has been widely used as a potential upgrade for catalytic pyrolysis to produce a bio-oil. In addition, compared to the pore size of the basic catalyst, mesoporous catalysts are important to have enough space for reactions; this also reduces the diffusion limitations of large hydrocarbon molecules with minimal steric hindrance when compared to that of a microporous zeolite [18,19]. Furthermore, the occurrence of coke causing catalytic deactivation is prevented [19–21]. MgO has been described as a strong Lewis base affecting the effectiveness of CO2 adsorption and exhibits resistance to carbon formation on the surface of the catalyst structure during devolatilization. Further pyrolysis reactions result in the enhanced conversion of high molecular weight hydrocarbons into smaller hydrocarbons [21,22]. Several studies have indicated that the use of CaO in the pyrolysis reaction, CaO acts as CO2 absorbent and reactant in biomass pyrolysis and exhibits the ability to absorb the moisture released [23–26]. CaO reacts with moisture and CO2 released to generate CaCO3 at low pyrolysis temperature and further decomposed at high temperature [27], enhancing decarbonylation during the thermal degradation of lignocellulosic components causing the formation of CO [28], while dolomite exhibits a role in absorbing, reacting, and catalyzing the thermal degradation conditions of the employed lignocellulosic components. Importantly, dolomite is an abundant natural carbonate mineral that is used as an inexpensive reactant, and mesoporous heterogeneous catalysts for gasification and pyrolysis reactions enhance the mass transfer of reactants and pyrolysis products [13,18,21,22,29,30]. Natural dolomite can be converted into a highly active basic oxide that consists of CaO, MgO, SiO3, Al2O3, and others after calcination at 700–900 °C affecting the thermal decomposition of MgCa(CO3)2 to an active metal oxide and modifying the chemical including both surface area and morphology [31]. Moreover, the calcined dolomite exhibits a significantly change during thermal decomposition at high temperatures, had many orderly pores on the surface, which established a pore structure. Both the surface area and pore volume values were higher than those of natural dolomite affecting the catalytic reaction, making an inexpensive and practical basic oxide catalyst for the catalytic cracking and reforming of tar in biomass pyrolysis gas [22,31]. The modification of calcined dolomite with both noble metals and several catalyst preparation techniques usually enhances its effectiveness as an active catalyst [32,33]. In particular, nonnoble metal oxides are important materials that are widely applied in various reactions to improve catalytic activity, enhancing the formation of unsaturated hydrocarbons and the effectiveness of converting unsaturated hydrocarbon compounds and aromatics into saturated hydrocarbon compounds. Therefore, the modification of small nonnoble metals, such as Ni particles, over a Lewis base catalyst results in an alternative catalyst with high resistance to Ni sintering and carbonaceous materials; moreover, it also been of interest due to its effectiveness toward C–O cleavage rather than C–C cleavage [29,31].Ni-modified dolomite catalysts are used to improve both the quality and quantity of bio-oil through deoxygenation, decarbonylation, and decarboxylation reactions. However, its use for the catalytic pyrolysis of softwood and non-wood biomass has not yet been reported. In this study, the effects of the catalytic pyrolysis operating parameters on lignocellulosic biomass were also determined. Rubberwood sawdust is a softwood candidate obtained from the production of wood pellets, which contain approximately 30% waste sawdust that has little or no economic value; however, this sawdust still has the potential to be used as an energy feedstock. Additionally, the cassava rhizome is a residual biomass candidate after agricultural harvesting. Farmers usually tend to burn it before planting the next crop cycle, which results in the release of small particles, smoke and air pollution to the environment. The catalytic pyrolysis of these two different lignocellulosic components using Ni-dolomite was also determined to investigate the chemical components, physiochemical properties of the crude pyrolyzed oil and production of valuable chemicals; additionally, catalytic performance was thoroughly investigated and discussed.Rubberwood sawdust (RWS) was provided from a wood pellet manufacturer in Rayong Province, Thailand, and cassava rhizome (CRZ) samples were collected after harvesting before the next crop in Uthaithani Province, Thailand. Both samples were subjected to moisture reduction, milled in a SW-2 high-speed rotary cutting mill (Hsiangtai, the People's Republic of China), sieved into a size distribution of 0.250–2.000 mm, dried at 105 °C overnight and stored before further pyrolysis. The biomass components were determined according to the TAPPI standard. The volatiles, ash and fixed carbon were obtained by proximate analysis using the ASTM D3172-73(84) standard. Elemental analyses were performed using a 628-CHN analyzer (LECO Corporation, USA.), while the oxygen content was calculated by the difference. Table 1 represents the characterization of the lignocellulosic components, and elemental analyses were employed.Calcined dolomite (DM) was produced by calcining at 850 °C for 4 h in a muffle furnace. After that, the DM was allowed to slowly cool until reaching approximately 100 °C before being quickly stored in a desiccator [30]. Next, the DM was dried at 105 °C for 2 h and ground to a particle size of less than 40 μm prior to use. In this study, x% Ni loading over DM was prepared using the wet impregnation method, where x wt.% Ni(NO3)2·6H2O (Sigma-Aldrich, Singapore) was placed in 10 mL of deionized water and 14 g of DM was placed in 90 mL of deionized water at a temperature of 60 °C and with constant stirring at 600 rpm for 6 h. The Ni-modified DM mixtures were filtered using vacuum-assisted filtration, rinsed with deionized water and placed into a hot air oven at 100 °C for 12 h. Then, the Ni-modified DM catalyst was crushed and filtered with a 20-mesh sieve before being calcined at 550 °C for 4 h. Ni-DM was characterized by an ASAP 2020 instrument (Micromeritics Corporate, USA) to determine the surface area, pore size, and pore volume. Prior to the test, a 0.15 g catalyst sample was placed into a sample tube and degassed under vacuum at 300 °C for 3 h. The nitrogen adsorption-desorption isotherms of the catalysts were obtained at −196 °C, and the surface area was also calculated by defining the adsorption isotherm. The crystallinity of Ni-DM and DM was characterized by a D8-Advance X-ray diffractometer (Bruker Corporation, Germany) that was operated with the following parameters: 30 mA, 40 kV, a Cu Kα radiation source at 0.154439 nm, a scanning rate of 4 °min−1, and a 2θ scanning range of 5°–90°. The crystal morphology and mesopores of the catalysts were characterized by using a JSM-6400 scanning electron microscope (JEOL, Japan).Catalytic pyrolysis tests were performed in a fixed bed pyrolyzer (3.81 cm i.d. and 120 cm length; SS316), as shown in Fig. 1 .Prior to the tests, 5 g of feedstock was placed into the hopper at the top of the pyrolyzer and dropped with the catalyst placed between the quartz wool in the middle of the pyrolyzer; the complete volatilization of the feedstock to vapor due to thermal decomposition occurred before being introduced to the catalyst layer, thereby preventing mixing between the pyrolyzed solid and catalyst. The Ni-modified DM was reduced with 5% H2 at a flow rate of 20 mL min−1 at 450 °C for 1 h to reduce NiO-DM to metallic Ni-DM. Then, N2 was circulated through the pyrolyzer to the top of the hopper until reaching the desired reaction temperature (450–600 °C) and a constant N2 flow rate (60–240 mL min−1) was kept for approximately 20 min before dropping the biomass into the pyrolyzer. The tests were performed for reaction times of 45–90 min. The volatile vapor due to the thermal decomposition of the biomass feedstock, which was dropped into the middle of the pyrolyzer, diffused through the glass wool and reacted at the Ni active sites and in the pores. The pyrolyzed vapors were condensed in an ice bath at the bottom of the pyrolyzer, while the noncondensable gas passed through the gas drier before being collected in a gas bag to further determine the gas composition. The obtained bio-oil from catalytic pyrolysis, which was collected in a cool trap, was diluted using acetone and separated using a funnel to determine the organic phase and aqueous phase. Then, the organic phase was evaporated at 60 °C to remove the solvent. After that, the bio-oil was obtained and weighed to determine the yield. The physicochemical properties were determined according to ASTM standards, such as the density (ASTM D1298), kinematic viscosity (ASTM D445), and higher heating value (HHV, ASTM D3286-91a); moreover, an 840-Trinoplus automated titration instrument (Metrohm AG, UK) was used to determine the modification acid number (ASTM D664). In addition, the ultimate analyses of the bio-oil from pyrolysis were performed using a CHN-628 instrument (LECO Corporation, USA.). The pyrolyzer was cooled to ambient temperature, and the solid char was collected and weighed; furthermore, the catalyst was separated from the quartz wool layer for weighing, the coke deposition was determined using the weight difference of the catalysts before and after the tests. Both solid char weight and weight of carbonaceous over the catalysts are classified as solid yield in percent by weight. The noncondensable gas yield was determined using the yield difference between the fed sample and the total yield of condensed bio-oil and solid char.1 μL of bio-oil was diluted in acetone to 1 mL before analysis using gas chromatography-mass spectrometry. A GC-7890A coupled with MS-5975C instrument (Agilent Technologies, USA.) that was equipped with an HP5MS nonpolar capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness) was used, and the gas chromatography (GC) oven was heated to 40 °C for 4 min and then to 280 °C at a heating rate of 4 °C min−1 for 10 min. The injector and detector temperatures were set to 250 and 280 °C, respectively. Helium (99.999%) was used as the carrier gas at 1 mL min−1, and a split ratio of 1:10 was used when injecting the a 1 μL sample. Electron ionization was performed at 70 eV for the molecular mass range of m/z = 50–550. The National Institute of Standards and Technology (NIST) mass spectral library was used to identify the main peaks in the chromatogram to determine the chemical compounds. The noncondensable gas was collected in a sampling bag from the gas drier unit, and then 60 μL was injected into the Agilent gas chromatograph coupled with a thermal conductivity detector using the following GC program parameters: argon flow of 30 mL min−1 pressure of 75 psi, oven temperature of 50 °C, injection temperature of 80 °C, and detector temperature of 90 °C. Table 1 lists the proximate and ultimate analyses and the lignocellulosic components in the rubberwood sawdust (RWS) compared with cassava rhizome (CRZ). The different components of both types of lignocellulosic biomass make it difficult to determine the differences in the thermal behavior [32], product yield and characteristics of the pyrolysis products.The results of the ultimate and proximate analyses (on a dry basis) of RWS and CRZ are presented in Table 1. As a result, both RWS and CRZ show no significant difference in the ultimate analyses. RWS and CRZ contain oxygen contents of 44.81 ± 0.23 wt% and 60.77 ± 0.31 wt%, respectively, which are primarily attributed to aldehydes, phenolics, carboxylic acids and oxygenated hydrocarbon compounds typically obtained in bio-oil. RWS contains 47.42 ± 0.35 wt%, which is higher than CRZ owing to its higher carbon content and lower oxygen content; moreover, the HHV of RWS (17.38 MJ/kg) is also higher than that of CRZ. The proximate analyses show fixed contents in RWS and CRZ, with values of 18.19 ± 0.42 wt% and 14.49 ± 0.88 wt%; this result accounts for the high char yield obtained from the catalytic pyrolysis reaction. It is notable that CRZ has an ash content of approximately 28.06 ± 1.41 wt%, while RWS has a lower ash content of only 2.75 ± 1.44 wt%. This affects the volatile content in the feedstock and favors the production of volatile vapor during catalytic pyrolysis, which results in the difference in the liquid and gas yield distribution. In particular, the high volatile matter contents indicate high volatility and reactivity, which enhance bio-oil production from pyrolysis, while a high ash content inhibits the production of bio-oil and leads to increased char and decreased gas contents [32,33]. Fig. 2 (A) shows the X-ray diffraction (XRD) pattern of natural dolomite, which consists of CaCO3, Ca(OH)2 and Mg(OH)2. Notably, the intensity of Ca(OH)2 and the crystalline structures of natural dolomite and calcined dolomite are not very different. Moreover, the XRD pattern of DM does not show Mg(OH)2 peaks, as illustrated in Fig. 2(B). In accordance with the thermal decomposition at a high calcination temperature above 700 °C, the intensities of the diffractograms represent the crystallinity of CaO and MgO that also appear [30].Ni-modified dolomite samples with different Ni loadings was prepared to investigate their catalytic activity during the pyrolysis of lignocellulosic biomass. As shown in Fig. 2(B)–(E), the calcined dolomite has the same intensity peak as the Ni-modified dolomite embedded in the calcined dolomite template. It is notable that the XRD pattern of several Ni-modified dolomite samples has a peak intensity that does not significantly change from the parent dolomite when increasing the Ni-ion concentration; additionally, the intensities of the diffraction peaks of Ni-DM are similar to the peaks of the parent calcined dolomite. The diffractograms show that the weight percentage of modified Ni in the calcined dolomite template does not change the original framework, according to the Ni concentrations, thus similar results are obtained in regard to the decreasing crystallinity of the Ni-modified DM samples when compared to the original dolomite template; this result can be attributed to the uniform metal dispersion throughout the Ni-calcined dolomite surface, which should be due to the complete crystal framework structure and lower Ni loading amount on the calcined dolomite [31]. Furthermore, the peak intensity gradually decreases with increasing Ni loading because the incorporation of Ni in the calcined dolomite template decreases the crystallinity of the calcined dolomite or causes the accumulation of NiO on the surface of the parent dolomite during calcination [30,31]. Fig. 3 shows the surface morphologies of natural dolomite, DM and Ni-DM. DM has many orderly pores on the surface, which results in a porous structure, while Ni-DM shows the same morphology as that of calcined dolomite. This can be confirmed by the fact that calcined dolomite retains its original textural form after the metal is loaded during the wet impregnation procedure. The addition of Ni slightly decreases the SBET of the parent calcined dolomite because some Ni metal is adsorbed on the surface of the pores, leading to a reduction in the pore volume and surface area. Nevertheless, it is also quite difficult to observe the difference in the pore volume of the Ni-doped calcined dolomite to that of the calcined dolomite because Ni cations are small and can diffuse through the mesopores of calcined dolomite; thus, they may fill the pores of the calcined dolomite. Furthermore, the SEM images illustrate the relative BET surface area listed in Table 2 , showing a total pore volume of 0.02 cm3/g and a surface area of SBET = 10.02 m2/g. The surface area and pore volume of both DM and Ni-DM are lower than those of natural dolomite.In particular, the average particle size distribution of samples is significantly affected by the product distribution of the biomass pyrolysis reaction due to the mass and heat transfer of the lignocellulosic components. Moreover, the components in the softwood and non-wood show different thermal decomposition behavior that enhances the decomposition of lignocellulosic components and volatilization to vapors coupled with deoxygenation, decarbonylation, and decarboxylation due to the catalytic activity. Thus, the influence of the average biomass size was determined to investigate the product yield and the characteristics of the produced bio-oil. As shown in Fig. 4 (A), the average size distribution (0.250–2.000 mm) of both RWS and CRZ is determined to investigate the pyrolysis product distribution when other operating parameters are kept constant: reaction temperature of 450 °C, reaction time of 45 min, inert nitrogen flow rate of 180 mLmin-1, 10%Ni-doped calcined dolomite and a catalyst loading of 10 wt%. The highest bio-oil yield (32.56 wt%) is obtained from RWS when using sawdust with a size distribution of 0.355–0.710 mm. The bio-oil yield slightly decreases, and the generation of noncondensable gas also increases when increasing the average size distribution of the feedstock from 0.710 mm to 2.000 mm. In contrast, the bio-oil yield obtained from the catalytic pyrolysis of CRZ tends to increase and have a maximum liquid yield (30.41 wt%) when the average size distribution of CRZ is increased to 2.000 mm. It is worth noting that the yield of solid and noncondensable gases tends to decrease slightly by increasing the average size of CRZ during catalytic pyrolysis. The results indicate that an increase in the average particle size of RWS leads to a decrease in bio-oil yield and a slight change in the solid yield; in contrast the noncondensable gas yield shows an increasing trend. This explains why sufficient heat transfer to the RWS particles during pyrolysis enhances the thermal decomposition of hemicellulose and cellulose into vapors, accomplishing the production of more noncondensable gas and decreasing the biochar yield. In particular, the heat transfer resistance more strongly affects bio-oil production than an increase in the average particle size because an insufficient temperature may occur during the thermal decomposition of lignocellulosic components [30–34]. Therefore, the catalytic pyrolysis of RWS, which is mainly composed of hemicellulose and cellulose, is easy to thermally decompose into volatile vapor and then further decompose via deoxygenation and decarboxylation; moreover, the secondary reaction of tar into bio-oil can be induced by heat transfer with an RWS size distribution of 0.250–0.355 mm. The catalytic pyrolysis of CRZ shows that when the particle size is increased, the bio-oil yield tends to slightly increase and reach a maximum of 30.41 wt% at an average size distribution of 0.850–2.000 mm. Notably, an increase in the average particle size of CRZ does not significantly affect the bio-oil yield but causes the thermal decomposition of the lignocellulosic components in CRZ. The influence of temperature and heat transfer from the wall of the reactor to the surface of the particle also mainly affects the decomposition of cellulose components to volatile vapor and high molecular weight hydrocarbon compounds [33,34]. Next, deoxygenation, including the catalytic activities of the dolomite and Ni-modified calcined dolomite, has a sufficient influence on the bio-oil and slightly changes the gas yield. It is worth noting that the catalytic pyrolysis of CRZ with an average particle size of 0.355–0.710 mm results in a bio-oil yield of only 27.93 wt%. However, when increasing the average particle size to 0.850–2.000 mm, the yield of biooil increases to 30.41 wt%, which is lower than the bio-oil yield from RWS when using an average particle size of 0.355–0.710 mm. This result shows that large particles have an effect on temperature, making it insufficient to complete thermal degradation during the primary thermal degradation step and resulting in a higher solid yield and lower bio-oil yield.Typically, thermal decomposition mainly influences the temperature and greatly affects the bio-oil yield. Fig. 4(B) illustrates the influence of temperature on the product distribution when varying the temperature from 450 to 650 °C while the other parameters are kept constant: average size distribution of 0.250–0.355 mm (RWS) and 0.710–2.000 mm (CRZ), inert nitrogen flow rate of 180 mL min−1, reaction time of 45 min, 10%Ni-DM, and a catalyst loading of 10 wt%. As the temperature is increased from 400 to 550 °C, it is found that the pyrolysis bio-oil yields of RWS and CRZ tend to increase from 32.36 wt% to 39.48 wt% and 30.41 wt% to 40.39 wt%, respectively. As seen from the results, the bio-oil produced from CRZ is higher than that produced from RWS because it consists of approximately 93.52 wt% hemicellulose and cellulose, which affects the thermal decomposition of volatile vapor and the subsequent deoxygenation reaction to a medium hydrocarbon vapor. Under the operating conditions, the secondary reaction and the tar carking reaction are not sufficiently accomplished; thus, it is found that the production of noncondensable gases has no significant tendencies. In addition, it is worth noting that when the average particle size of CRZ is larger, it receives sufficient heat transfer, causing thermal decomposition of the lignocellulosic component to a large amount of volatile vapor. Then, deoxygenation and decarbonylation are also enhanced to obtain medium-weight hydrocarbon compounds before being influenced by the Ni-DM catalyst, allowing the hydrocarbon chain to break into small hydrocarbon compounds at the active Ni sites on the surface of the dolomite structure to obtain the highest bio-oil yields of 39.48 wt% and 40.39 wt% when using RWS and CRZ, respectively. As the temperature is increased to 600 °C, the bio-oil yield decreases, the gas yield also significantly increases due to the secondary reaction, and the cracking of tar also occurs, causing an increase in noncondensable carbon dioxide and carbon monoxide gases [35–37]. In addition, the tendency of solid char is also likely to increase the process temperatures from 400 to 550 °C. The pyrolyzed solid from CRZ is found to be higher due to the influence of this temperature increase, which causes uniform heat transfer into the lignocellulosic particles in CRZ; thus, the uniform heat transfer into the particles enhances thermal degradation, devolatilization and the secondary reaction of tar cracking [30,33,35,37]. Moreover, catalytic pyrolysis of small volatile vapors over the active Ni sites results in large bio-oil and gas yields. In contrast, with an increasing process temperature up to 600 °C, the content of pyrolyzed solid decrease from both RWS and CRZ, which is due to the influence of the temperature in accomplishing the devolatilization of the lignocellulosic components. Notably, the fixed carbon remains stable. Furthermore, a high pyrolysis temperature might decrease the catalytic activity, leading to a lower yield of desirable products, such as bio-oil, which is affected by the influence of temperature on deoxygenation and is slowed with an increase in operating temperature; in contrast, coke decomposition on the surface of the Ni-DM structure causes catalytic deactivation [31,35,38]. Thus, a desirable product distribution at high temperature shows a significant decrease. Fig. 4(C) shows the influence of the reaction time on the catalytic pyrolyzed product distribution from RWS and CRZ using the following operating conditions: average size distribution of 0.250–0.355 mm RWS and 0.710–2.000 mm CRZ, the temperature of 550 °C, inert nitrogen flow rate of 180 mL min−1, 10%Ni-DM, a catalyst loading of 10 wt%, and a varying reaction time of 45–90 min. There is a significant decrease in the bio-oil yield of both RWS and CRZ by continuously increasing the reaction time. The results can be explained by the continuously increasing reaction time affecting the thermal decomposition of the lignocellulosic volatile vapor at high temperature. Then, deoxygenation, decarboxylation and decarbonylation reactions enhance its conversion into smaller volatile vapors and further decomposition by secondary reactions; additionally, the cracking of tar results in a large quantity of noncondensable gas. Then, the small vapor gas easily diffuses into the mesopores of the catalyst [23,32]. Therefore, an increase in reaction time with a temperature over 45 °C enhances the devolatilization of lignocellulosic components to small volatile vapors that can be further decomposed into small noncondensable hydrocarbon vapors. This reaction is still influenced by a temperature of 550 °C, causing the secondary degradation reaction and tar cracking reaction to result in a bio-oil yield that decreases to 37.64 wt% RWS and 33.56 wt% CRZ when catalytic pyrolysis is continuously performed for 90 min. These results show that longer reaction times lead to an increased noncondensable gas yield which is affected by the influence of high temperature [37,39]; this is contrary to the yield of bio-oil, while the solid char yield shows an insignificant difference due to the devolatilization of the lignocellulosic components to form stable fixed carbon [22,40,41].The influence of the inert nitrogen flow rate was investigated with the following constant catalytic pyrolysis parameters: average size distribution of RWS (0.250–0.355 mm), CRZ (0.710–2.000 mm), temperature of 550 °C, reaction time of 45 min, and 10 wt% loading of 10%Ni-DM. The inert N2 flow rate was varied from 60, 120, 180, and 240 mL min−1. As shown in Fig. 4(D), the control of volatilization from pyrolysis of both RWS and CRZ at a N2 flow rate of 180 mL min−1 obtains the highest bio-oil yields of 39.48 wt% and 40.39 wt% when using RWS and CRZ, respectively. Both the catalytic pyrolysis of RWS and CRZ obtain the same trend of bio-oil production when increasing the flow rate of inert N2 gas. The reason for these results is that the lower flow rate of N2 affects the longer reaction time of the devolatilization of lignocellulosic components into volatile vapor, which is influenced by the reaction temperature of 550 °C, leading to the formulation of smaller volatiles due to the deoxygenation, decarbonylation, and decarboxylation reactions. Notably, the long residence time results in a large amount of volatile vapor and further continuous cracking by the secondary reaction. The cracking of tar is affected by continuing to convert small volatile vapor into noncondensable gas that then diffuse through the surface of the Ni-DM structure and throughout the pyrolysis reactor [18,21,31,33]. In contrast, the high N2 flow rate causes a short residence time of devolatilization. All vapor gases receive insufficient heat transfer to decompose the lignocellulosic components, while the N2 carrier gas leads to the larger molecular weight volatile vapor passing throughout the pyrolysis reactor before the secondary reaction. Tar cracking and catalytic pyrolysis are also accomplished. The decreasing trend in the bio-oil yield when increasing the inert N2 flow rate is caused by a short residence time and insufficient heat transfer into holocellulose is not complete despite flowing throughout the reactor with the N2 carrier gas [30,37].The advantage of dolomite is that it is an inexpensive mesoporous strong Lewis base catalyst, providing effective mass transfer of larger hydrocarbon molecules into its pore structure with minimal steric hindrance when compared to that of a well-known commercial catalyst. DM can might absorb moisture at low temperature of pyrolysis reaction and will then be released by increasing the pyrolysis temperature, promote devolatilization and deoxygenation of holocellulose and the adsorption of both CO and CO2 on the pore structure [13,26–28]. The stronger basicity and more developed pore structure of Ni-DM demonstrate its strong potential in improving the deoxygenation reaction and further catalytic pyrolysis enhances C–O bond cleavage rather than C–C cleavage [40], which then passes through the mesoporous dolomite and undergoes a secondary reaction, namely, tar cracking to form low molecular weight hydrocarbon compounds [41–43]. Thus, the bio-oil yield increases with the Ni concentration on the modified DM. As seen in Table 2, the increase in impregnated Ni from 5 to 20 wt% in the DM template shows a slight change in surface area and pore volume but the concentration of Ni-modified on calcined dolomite from 5 wt% to 20 wt% has an effect on the pyrolysis yield as seen in Fig. 4(E). In particular, when using a Ni concentration of 20 wt% on DM, the highest yields of noncondensable gas for both RWS (38.27 wt%) and CRZ (33.17 wt%) are obtained which can be explained by the DM having additional catalytic performance with an increased concentration of Ni. This enhances the devolatilization of holocellulose, deoxygenation of volatile vapor due to C–O cleavage, and formation of CO2 due to the Lewis basicity [13]. When the Ni concentration is enhanced to provide higher catalytic activity, the secondary reaction and further vapor cracking to small volatile vapors [31,36,42,43] is also observed. When increasing the Ni impregnated from 5 wt% to 10 wt% into calcine dolomite, a slight change in the bio-oil yield produced from RWS and CRZ occurs, with values of approximately 37.91 wt% to 39.48 wt% and 38.61 wt% to 40.39 wt%, respectively. This suggests that increasing the Ni modification on DM further promotes deoxygenation, decarbonylation, and decarboxylation; then, secondary reactions and tar cracking also occur [31,32,37], resulting in a decrease in bio-oil and leading mainly to a dramatic increase in noncondensable gas. With 20 wt% Ni-DM, a strong acid site is obtained, but the dispersion of a high Ni modification might decrease both the surface area and pore volume, which affects the deactivation of the Lewis base catalyst, including the influence of the high temperature enhancement on coke formation. The highest yield of noncondensable gas is likely the product distribution from noncatalytic pyrolysis. Furthermore, hydrogen transfer during catalytic pyrolysis is promoted by the active Ni modification, which might have inhibited carbonaceous material on the surface and in the pores of the dolomite support and slightly enhanced the solid char yield.The influence of catalyst loading (0%, 5%, 10%, 20% by weight of the feedstock) of RWS and CRZ pyrolysis on the product distribution were investigated under a constant operating condition of 550 °C, reaction time of 45 min, and inert N2 flow rate of 120 mL min−1 at the appropriate 10 wt%Ni modified-DM. As shown in Fig. 4(F), the use of 10 wt% loading of 10%Ni-DM catalyst increased both the pyrolyzed -oil yield and noncondensable gas yield. This increased use of catalyst could produce a higher pyrolyzed oil yield (39.48 wt% from RWS, 40.39 wt% from CRZ) compared with the noncatalytic pyrolysis of both lignocellulosic biomasses, which were only 28.63 wt% from RWS and 27.87 wt% from CRZ because calcined dolomite can enhance the devolatilization of large-molecule compounds to small-molecule compounds, making it easier to enter the pore structure and react with 10 wt%Ni-modified at the catalyst surface due to enhanced catalytic activity, C–C cleavage, and decarboxylation. Then, the secondary reaction also occurred and promoted low molecular volatile vapor to obtain pyrolyzed oil [44,45]. With an increase catalyst loading in the 10 wt%Ni-DM from 5 wt% to 20 wt% into the pyrolysis reaction, both pyrolyzed oils produced from RWS and CRZ dramatically decreased, but in contrast, it increased the amount of gaseous RWS and CRZ in both catalytic pyrolysis processes and reached the maximum yield of a noncondensable gas of 35.43 wt% from RWS and 29.12 wt% CRZ at a catalyst loading of 20 wt% and decreased the pyrolyzed oil yield. Notably, catalytic pyrolysis using 20 wt% catalyst loading enhanced the devolatilization, deoxygenation, decarboxylation, decarboxylation and secondary reaction due to the influence of high temperature, which led to the decomposition of volatile vapor to a small noncondensable gas similar to noncatalytic reactions. These results might explain why an increase in the catalyst loading enhances the large production of noncondensable gas, while the noncatalytic reaction and thermal degradation have a significant effect on the devolatilization of lignocellulosic components [30,31]. In particular, the pyrolysis of RWS, which contains hemicellulose that decomposes more easily at low temperature than the thermal decomposition and devolatilization during the pyrolysis of CRZ. The solid char yield showed a fluctuating trend when increasing the catalyst loading. Compared with the noncatalytic experiment, the production distribution of solid char (30.43 wt% from RWS, 37.59 wt% from CRZ), pyrolyzed oil yield (28.63 wt% from RWS, 27.87 wt% from CRZ) and maximized noncondensable gas (40.94 wt% from RWS, 34.54 wt% from CRZ) are obtained.The pyrolyzed oil produced from the catalytic pyrolysis of RWS and CRZ over Ni-DM in the fixed bed reactor was investigated using GC-MS. The use of the catalyst improved the activation energy of catalytic pyrolysis and promoted thermal catalytic cracking during the pyrolysis of lignocellulosic components in accordance with the enhanced devolatilization of lignocellulosic components to improve the pyrolyzed oil yield. Table 3 illustrates the relative intensity peak and identified chemical compounds of catalytic pyrolysis oil by varying the percentage of Ni modified on calcined dolomite at operating condition of 550 °C, an inert N2 flow rate of 180 mL min−1, the reaction time of 45 min using catalyst loading at 10% by weight.A large number of peaks were also observed, indicating many types of organic compounds, e.g., aliphatic, monoaromatic, polyaromatic, and oxygenated organic compounds. In particular, calcined dolomite represents a stronger Lewis basic catalyst that obtains more basic sites, and higher specific surface area and mesoporous properties affect the active catalytic performance in C–O cleavage and improve the deoxygenation reaction [31,40,41,43], resulting in volatile vapor and small molecular weight oxygenated hydrocarbon compounds, e.g., phenol furan, ketone, some sugar and carbohydrate derivatives. Phenols are greatly produced from the catalytic pyrolysis of lignin components. The degradation of lignin via catalytic pyrolysis mainly obtained the main source of active free radicals, while furan was obtained by thermal decomposition, dehydration of cellulose components at low temperature and further deoxygenation, decarboxylation, and decarboxylation to produce aliphatics [46]. Thus, the catalytic pyrolysis of RWS, which contains a lignin component of approximately 29.97 wt%, might yield phenols and phenol derivatives. Then, phenolic oligomers abstract hydrogen transfer during thermal degradation at low temperature. Then, smaller alkyl phenolic and oxygenated compounds undergo secondary reactions to smaller phenols [47–49], while the pyrolysis of CRZ, which consists of a higher proportion of cellulose, might be converted to furan via deoxygenation and further decarbonylation and decarboxylation and is influenced by Ni-modified catalytic performance on calcined dolomite. However, the use of Ni-DM significantly reduced furan and phenol production and eliminated carboxylic compounds and cyclopentanone formation during the catalytic pyrolysis of both RWS and CRZ. Notably, long chain aliphatic hydrocarbons were formed and further converted to shorter chain hydrocarbons due to the active catalytic performance of Ni modification. Moreover, Ni-dolomite catalysts play a catalytic role with active sites where increasing Ni modification on the dolomite support reduces the activation energy of the catalytic pyrolysis and promotes thermal catalytic cracking during pyrolysis of the lignocellulosic component, which is in accordance with the enhanced devolatilization of holocellulose. In particular, the thermal decomposition of hemicellulose and cellulose into volatile vapor and further deoxygenation, decarbonylation, and decarboxylation caused by the role of Ni on the stronger Lewis basic catalyst enhanced C–O cleavage rather than C–C cleavage at Ni-acidic sites on the dolomite surface led to smaller amounts of oxygenated compounds. The most abundant aliphatic product of both alkanes and alkenes was observed from the use of Ni-DM in the catalytic pyrolysis of both RWS and CRZ. This result might explain why alkyl radicals readily transfer H radicals to volatile vapor during the thermal degradation of hemicellulose at low temperatures, resulting in the production of alkanes and alkenes from alkyl radicals. Moreover, the presence of a sufficient concentration of Ni-modified dolomite catalyst might convert furan to monoaromatics by enhanced deoxygenation, decarbonylation, decarboxylation, and secondary reactions, including tar cracking [21,30,31,41,42], contrary to the decrease in aldehydes and ketones and the disappearance of long chain carboxylic acids due to deoxygenation, decarbonylation, and decarboxylation at the Ni active site. Compared with the pyrolysis reaction with the use of catalyst, all the Ni-modified calcined dolomite slightly reduced the solid char yield while increasing the liquid and noncondensable gas yield at the conversion of large molecular weight volatile vapor. These phenomena are also due to various hydrocarbon conversion reactions, e.g., thermal degradation at low temperature, catalytic activity in C–C cracking, deoxygenation, and oligomerization, which are catalyzed by Ni active sites on the Lewis basic support [41,42]. Table 4 presents a comparison of the pyrolysis oil from the pyrolysis without Ni-DM and catalytic pyrolysis by varying the percentage by weight of Ni modified on calcined dolomite at the optimal condition of 550 °C, an inert N2 flow rate of 180 mL min−1, the reaction time of 45 min using catalyst loading at 10% by weight, which represented a lower oxygen content and higher H/C content than the noncatalytic pyrolysis due to the thermal decomposition of holocellulose at low temperature. It was found that RWS contained 22.71 wt% hemicellulose and 47.32 wt% cellulose, which were affected by thermal decomposition and converted to volatile vapor in the form of alkyl radicals. Furthermore, RWS easier C–O cleavage and enhanced deoxygenation, decarboxylation, and decarboxylation to low oxygen volatile vapor and further reacted with the Ni-modified active site on calcined dolomite to produce bio-oil, which contained low amounts of oxygenation components. The thermal decomposition of cellulose, which mainly consisted of CRZ, was also more difficult; thus, the pyrolyzed oil from CRZ may have a higher O/C ratio affecting the HHV of approximately 27 MJ/kg. The HHV in noncatalytic pyrolysis was 26.86 MJ/kg and 24.89 MJ/kg, indicating that the presence of Ni-DM promoted thermal catalytic pyrolysis of lignocellulosic components in accordance with enhanced devolatilization into volatile vapor and further deoxygenation reaction [18,21,22] caused by the Ni active site on the stronger Lewis basic catalyst and that its pore structure led to oxygen removal during the whole pyrolysis reaction [29]. As seen from Table 4, the solid char obtained from catalyzed with Ni-modified calcined dolomite showed quite similar elemental analyses in biochar composition for both RWS and CRZ due to thermal devolatilization and further deoxygenation to pyrolyzed oil, meanwhile, the noncatalytic pyrolysis of biomass was rapidly thermally decomposed. When the pyrolysis reaction undergoes at high temperatures, some volatile vapor may be decomposed into the hydrocarbon radicals and depolymerization to form a large hydrocarbon compound as a residual component in the solid char, it was found that the elemental analyses of noncatalytic biochar exhibit a quite difference in carbon and oxygen content, has a higher O/C ratio is considered to be less candidate solid fuels due to its lower heating value. These results indicated that the practical application of catalytic pyrolysis from both softwood and non-wood could produce useful fuel-like compounds and chemicals with high heating values, and the physicochemical properties of pyrolyzed oil and solid char were higher than those of raw materials. Table 5 presents the compositions of the noncondensable gases produced from the catalytic pyrolysis of RWS and CRZ using the difference percentage by weight of Ni modified on calcined dolomite at the optimal condition of 550 °C, an inert N2 flow rate of 180 mL min−1, the reaction time of 45 min using catalyst loading at 10% by weight, compared with the noncatalytic reaction, are mainly composed of CO2, H2, CO and CH4, while small hydrocarbon (C2–C3) gases also occurred. Both softwood and non-wood lignocellulosic biomass obtained similar components of pyrolysis of noncondensable gas due to the catalytic activity enhanced the thermal decomposition of lignocellulosic components into volatile vapor, then both C–C cleavage and C–O cleavage including deoxygenation, decarbonylation, decarboxylation and further the secondary reaction high temperature and longer residence time enhanced to cracking of volatile vapor into noncondensable gas [22,31,38,40,47]. As seen from Table 5, CO2 is mainly a gas component from the pyrolysis of lignocellulosic biomass due to the deoxygenation reaction of volatile vapor from the thermal degradation of holocellulose. The increased Ni-modified concentration enhanced the appearance of the CO component in the presence of Ni-DM and further enhanced the deoxygenation ability of DM. These results might explain why the deoxygenation of volatile vapors enhanced the production of phenols and monoaromatics, including some aliphatic hydrocarbon compounds. Notably, an increase in Ni-DM also leads mainly to increased production of H2, contrary to the reduction in CO2 concentration in a noncondensable gas component, which may be affected by the adsorption of CO2 the presence of Ni-DM, is able to react with CO2 then converting the CaO component in Ni-DM to CaCO3 [13,26], including the water gas-shift reaction enhancing the occurrence of the H2 component [31,34,47–49]. CH4 and olefins change a little, was formed when methoxy organic derivatives were formed during oxygenation and then dealkylation reactions, while the appearance of C2H4, C2H6, C3H8 hydrocarbon gases might be attributed to the degradation of alkyl groups in the oxygenated compounds during carboxylation and decarboxylation of volatile vapor. Then, secondary reactions, including the secondary reaction of alkyl groups attached to alcohol or phenol compounds, also converted lower molecular weight hydrocarbons to C2–C3 gases.This study compared the catalytic pyrolysis of soft wood and non-wood biomass in a fixed bed reactor. Ni-modified calcined dolomite acted as an active catalyst in the pyrolysis reaction. The lignocellulosic components of both biomasses showed a difference in product yield due to thermal decomposition depending on their components. The highest bio-oil yield of ∼34 wt% was obtained from the catalytic pyrolysis of rubberwood sawdust at a temperature of 550 °C, an inert nitrogen gas flow rate of 180 mL min−1,a reaction time of 45 min, catalyst loading of 10 wt% with 10%Ni-modified calcined dolomite. Both pyrolyzed bio-oils obtained from softwood and non-wood biomass consisted of aliphatic (paraffin, olefin), monoaromatic, polyaromatic hydrocarbon compounds and derivative oxygenated compounds, while the physicochemical analyses indicated that oxygen compounds were removed via thermal decomposition, deoxygenation, decarbonylation and decarboxylation. Ni-DM acted as an acid site on the mesoporous dolomite parent, affecting the C–C cleavage of large volatile vapors into smaller vapors, and then the secondary reaction and pore structure enhanced the production of small hydrocarbon compounds. The solid char yield from the pyrolysis of non-wood was higher than from softwood because the amount of fixed carbon and ash. The noncondensable gas product mainly consisted of CO2 and H2, with increasing Ni-DM loading reducing the CO concentration, whereas the yield of CO2 also increased due to decarboxylation rather than decarbonylation. Notably, the yield of H2 increased with an increase in the Ni-DM loading due to the dehydrogenation reaction that occurred in the secondary reaction. Kittidech Praserttaweeporn, Methodology, Data collection, Visualization, Data curation. Tharapong Vitidsant: Conceptualization, Supervision. Witchakorn Charusiri: Conceptualization, Methodology, Visualization, Data curation, Validation, Writing – review & editing, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by the Center of Excellence on Petrochemical and Materials Technology (PETROMAT), Center of Fuels and Energy from Biomass, Chulalongkorn University in the form of facilities support, and research projects through the Program Management Unit Competitiveness (PMUC), National Higher Education Science Research and Innovation Policy Council (No. C10F630094) and Srinakharinwirot University also partially supported this study.

This study investigated the effect of process conditions during the catalytic pyrolysis of softwood and non-wood on the pyrolysis product. Ni modified on calcined dolomite (Ni-DM) was used to determine the effectiveness of the catalytic activity and the effect of doping non-noble metals on the product distribution of both pyrolyzed biomasses. The lignocellulosic biomass components showed different effects on the decomposition characteristics of the pyrolysis vapor and its devolatilization, while Ni-DM showed a catalytic effect to enhance the decarbonylation, decarboxylation, and secondary reaction of tar cracking affecting CO2 and CO removal; additionally, the catalytic activity also promoted the formation of aliphatic and olefin hydrocarbon compounds and facilitated C–C cleavage and scission to smaller hydrocarbon compounds. With increased Ni loading, the yield of noncondensable gas increased. Furthermore, gas chromatography analysis indicated that the composition of gas mainly consisted of hydrogen gas, which increased significantly due to a water-gas shift during the catalytic pyrolysis of biomass at high temperature with 10% Ni-dispersed calcined dolomite acting as the catalyst. Meanwhile, the pore structure and the modified-Ni on calcined dolomite enhance decrease acids and sugars in bio-oil yield and favor the formation of alkane gases of liquid hydrocarbon fuels including the dramatically increased in alkane gases.

No data was used for the research described in the article.At present, >300 different grades of polyolefins are commercially available for different applications. Particularly, polyethylenes (PE) and polypropylenes (PP) are the most important materials produced by the petrochemical industry [1]. Together, they account for the largest segment of commercial polymeric materials [2]. Thanks to the dramatic developments in catalyzed polymerizations, the annual production of polyethylenes exceed 170 million tons, which is roughly 50 % of the global plastics production [3]. Although faced with environmental concerns, their superior mechanical performance still provides substantial advantages over their bio-based alternatives [4]. Coordination-insertion polymerizations catalyzed by transition-metal complexes are the predominant solutions for synthesizing polyethylenes on the industrial scale [5]. As already mentioned above, the major advantage of coordination-insertion chemistry is the outstanding control over polymeric microstructure, which also determines the macromolecular characteristics of polymers, such as the mechanical, thermal, and optical properties, and thus their final commercial values [6,7]. Consequently, the improvement of the catalytic performance of organometallic catalysts in ethylene (co)polymerization is a key driving force in catalyst research.Over fifty years ago, Ziegler and Natta won the Nobel Prize in Chemistry for their discovery of heterogeneous catalysts applied to olefin polymerizations [8,9]. Since then, the research field of transition-metal catalyzed olefin polymerization has seen great progress [10]. Compared to their early-transition metal rivals, the late-transition metal organometallics have been a topic of interest, because of their low oxophilicity, as well as their great potential in producing polymers containing various types of branches and polar functional groups [11–13]. Currently, a variety of structurally related late-transition metal organometallics have been applied in the ethylene (co)polymerization which includes α-diimine (N^N), phosphine (P^O), phenoxyimine (N^O) and pyridinylimine (N^N) types of catalysts (Fig. 1 )[14–20]. Disparate catalytic performance and varying polymeric properties can be obtained via the utilization of different metal complexes. For example, phosphine-containing palladium catalysts have emerged as powerful alternatives for copolymerization of ethylene with polar monomers, leading to linear polymers [21–24]. However, the α-diimine Ni and Pd complexes still have a lot of advantages over the rest, including the ease of synthesis and structural modification of ligands. In addition, late-transition metal complexes have attracted more considerable attention with the report of the highly active α-diimine Ni and Pd complexes by Brookhart (Fig. 1, A), for application in ethylene polymerization [25–28]. The branching density and topology of polyethylenes could be predictably adjusted by modifying the ligand structure and polymerization conditions. Thus elastic polymers could be synthesized using ethylene as a single monomer [29].According to the previous studies, it was observed that small changes in the ligand structure entail significant variations in the macroscopic properties of the resulting polyethylenes [30]. N-aryl steric effects (Fig. 1. B, C, D, and E), backbone effects (Fig. 1. F and G), and remote substituent effects (Fig. 1. B and G) have been considered as main variables, influencing the catalytic behavior of α-diimine Ni(II) and Pd(II) complexes and consequently modifying polymer properties [31–39]. It is worth noting that the incorporation of phenyl groups on the ortho-moieties of N-aryl has helped achieve the control over the synthesis of either linear or branched polyethylenes, exhibiting ultrahigh molecular weight (up to 4 × 106 g mol−1). These complexes were stable in the presence of hydrogen during ethylene polymerization (Fig. 1. B) [31]. Recent research activities in α-diimine Ni and Pd catalysts are mainly devoted to the modification of complex structures to achieve more efficient and higher degree of catalytic properties in homogenous (co)polymerization. Along these lines, unique and functional polymers, such as ultra-high-molecular-weight polyethylenes, elastomeric polymers, and highly branched and high-molecular-weight polyethylenes, have been reported through the elaborated designs of α-diimine Ni and Pd catalysts in catalyzed ethylene (co)polymerization [5].Over the last few years, several high-quality reviews on Ni and Pd complexes have been published, which partially includes the α-diimine Ni and Pd complexes for ethylene or α-olefin (co)polymerizations [5,11,30,40–47]. However, we still see the obvious gap between the published literatures and the specifically state-of-art discussion (especially fastened in the last decade). The lack of a comprehensive overview of α-diimine Ni and Pd complexes still remains, which includes the catalytic mechanisms, catalytic behaviors, and heterogeneous catalysis applications. Hence, this review emphasizes the structural variation of α-diimine Ni and Pd complexes, their catalytic performance during ethylene (co)polymerization, catalytic mechanisms in the ethylene polymerization and copolymerization with polar monomers, and the resulting polymer properties. Although researchers have invested extensive efforts in the modifications of ligands and complexes structures, the discovery of promising catalytic systems is inevitably accompanied by trial and error. The relationship between structural variations (steric and electronic effects) of α-diimine Ni and Pd complexes and their catalytic performance is of great importance and thus need to be addressed in detail. More importantly for future application, the use of heterogeneous catalysts in the gas- and slurry-phase polymerization currently represents the predominant route in the industry [48]. Studies related to the immobilization of α-diimine Ni and Pd complexes onto solid substrates are a crucial step for their successful commercialization. This review has four main sections: i.e. i) Coordination-insertion chemistry; ii) Relationship between structural details and catalytic performance, iii) heterogeneous polymerization using supported α-diimine Ni and Pd complexes, iv) the mechanical properties and hydrophilicity of the produced polymers.Thermoplastic elastomers (TPEs) are rubber-like materials, which offer the ease of processing, recyclability, and enhanced mechanical characteristics. Their annual production is driven by industrial and consumer demands [49]. For TPEs synthesis, Dow's constrained geometry catalysts (CGC) were widely applied to produce the long-chain branches, which involve the incorporation of ethylene and higher a-olefin monomers. Short-chain branches in TPEs are achieved via copolymerization between ethylene and a-olefin monomer, catalyzed by transition-metal-based complexes. In contrast, α-diimine Ni and Pd complexes exhibit unique catalytic behaviors, producing high molecular weight polyethylenes with highly branched structures through the so-called chain-walking mechanism. This facile synthetic approach enables control of polyethylenes microstructures (either linear or branched polymers), solely using ethylene as the monomeric feed. The ethylene (co)polymerization catalyzed by the α-diimine Ni and Pd complexes provides a cost-effective alternative to more complicated and multi-step approaches to synthesize elastomeric materials, as it requires only a single step for their manufacturing [44]. A notable report describing the chain-walking mechanism was originally put forward by Fink et al. [50,51]. Soon afterward, Brookhart et al. refined this mechanistic method, which was then validated by both experimental and theoretical research [25,26,52,53]. Scheme 1 illustrates the concept of the chain-walking mechanism for ethylene polymerization. Initiated by a cocatalyst, the chain-walking mechanism could be divided into several parts, namely chain propagation, chain transfer, and chain walking or isomerization. Cationic alkyl-metal active species provide superior activities. Hydrogen elimination and chain transfer lead to the formation of highly branched polyethylenes (Scheme 1). The catalytic metal center migrates along the polyethylene backbone via rapid β-H elimination and reinsertion as a chain-walking process [46]. The molecular weight of the obtained polyethylenes depends on the competition between the monomer insertion (chain growth) and the β-H elimination (chain transfer), where the latter process, followed by the reinsertion, leads to the formation of branched structures. More recently, Pei et al. have proposeda unique and new mechanistic model for the formation of long-chain-branches (LCBs), based on the classic chain-walking process catalyzed by α-diimine Ni complexes. The methyl branch was produced by a one-step chain walking followed by ethylene insertion. Consequently, the long-chain branching (LCBs) was directly obtained by ethylene insertion into the primary Ni-alkyl species, which is originally formed from the migration of the catalytic Ni center to the methyl terminal. Steric interactions from the ortho-aryl substituted anilines play a central role to limit the ethylene insertion to selectively generate the primary Ni-alkyl species and/or the secondary Ni-alkyl species with the α-methyl substituent. The proposed mechanism could explain the existence of methyl and LCBs without the formation of other short branches during the ethylene polymerization [54]. The bulky substituentson the ortho-N-aryl offer a steric crowding at the axial sites of the catalytic metal centers, which are perpendicular to the metal-diimine plane. This steric hindrance suppresses the associative chain transfer process. Such metal complexes bearing more bulky groups can lead to higher molecular weight polymers. Additionally, the diimine backbone influences the polymerization activity as well as the polymer molecular weight. The backbones with alkyl substituents are reported to yield higher molecular weight of polymers with narrower PDI than the planar acenaphthyl backbones [30]. Meanwhile, the steric enhancement on the diimine backbones significantly improves the thermal stability of the catalyst during ethylene polymerization. Side-arm effects (electronic effects and weak interactions) from the remote substituents of the ligands similarly influence the catalytic activity and ensuing polymer properties.Generally, the polymerization temperature and ethylene pressure have a significant impact on the catalytic performance of the α-diimine Ni and Pd complexes. The high temperature increases the rotation of the C-Naryl bond, reducing steric hindrance at axial sites. Furthermore, a high rate of chain transfer is expected at high polymerization temperature, which leads to an increased reinsertion rate and the formation of branching. As the ethylene pressure is increased, chain propagation is preferred over chain transfer, therefore more linear and less branched polymers are formed. In addition, the α-diimine Pd complexes tend to undergo the chain-walking process, as compared to the corresponding Ni complexes.Hyperbranched and amorphous polyethylenes can be produced via the α-diimine Pd catalyzed ethylene polymerization, while the α-diimine Ni complexes can produce mainly linear (few short-branches) polyethylenes with a well-defined melting point [25].Due to this unique catalytic behavior of α-diimine Ni and Pd complexes, alkyl chains, such as methyl, ethyl, propyl, butyl, and even longer branches could be generated in the polymer backbones. Various polymers with elastic, semi-crystalline, and amorphous properties could be modulated with controlled chain-walking polymerization, using different reaction conditions and /or specially-synthesized Ni and Pd catalysts [42,55].Polyethylenes possess essential characteristic like excellent chemical resistance, ease of manufacturing, and low production costs [8]. However, its nonpolar backbone also has a lack of added-value functionalities, which are important applications in many fields [56]. Functionalized polyethylenes exhibit improved surface and mechanical performance due to the incorporation of functional polar groups [19,57]. The synthesis of functionalized polymers is primarily performed by post-polymerization functionalization, free radical copolymerization, or, by use of special methods like ring-opening metathesis polymerization (ROMP). These polymerization approaches have some drawbacks, such as the use of harsh conditions or poor controls of polymer microstructure [40,58]. In contrast, flexible synthesis of functionalized polyethylene with well-characterized structures and properties is currently an important field of research in coordination-insertion polymerization [59]. Early-transition metal complexes, like the Ziegler-Nata catalysts, were certainly applied to copolymerize ethylene with polar monomers. However, researchers could not achieve any success in synthesizing copolymers, due to the poisoning effect from the polar monomers. Thanks to the low oxophilicity, the late-transition metal complexes (especially for Pd-based complexes) exhibit remarkable capacity to copolymerize ethylene with polar monomers (Scheme 2 ) [11]. Following the insertion of polar monomers, the metal center coordinated with the CC bond to form the intermediate I. There were two types of the insertion process, namely the 1,2 insertion (intermediate II) and 2,1 insertion (intermediate III). The polar group, X (Lewis-basic groups) and catalytic metal center, M (Lewis-acidic groups) in Scheme 2 potentially generated the stable metalation of X-M chelates. These chelated metalations deactivate the cationic alkyl-metal species and terminate the copolymerization process [5,60].The initially reported α-diimine Ni and Pd catalysts could surprisingly accomplish the ethylene copolymerization with methyl acrylate as a comonomer [26]. This indicated that the late transition metal catalysts could provide an effective solution for the challenging ethylene copolymerization with polar monomers, because of the low oxophilicity of the metal centers [11,40,61]. This discovery could address the deactivation problems associated with the polar groups near the metal center. However, the field of copolymerization of ethylene and polar monomers has been mainly dominated by the use of α-diimine palladium catalysts. α-Diimine nickel catalysts are generally less tolerant toward polar groups than the palladium catalysts. The nickel catalysts could only catalyze the ethylene copolymerization with a limited number of polar monomers, such as polar derivatives of norbornene, silane-based and long-chain α-olefins[40,62]. With the extensive exploration of new α-diimine Ni and Pd complexes, a variety of polar monomers have been broadly investigated in ethylene copolymerization studies. Fig. 2 displays the collection of polar monomers applied in the ethylene copolymerization catalyzed by α-diimine Ni and Pd complexes. These polar monomers could be classified into two types of vinyl monomers: alkene-connected and long-chain polar monomers [60]. Alkene-connected polar monomers refer to the monomers, where the polar groups were directly connected to the CC bond. As reported, these types of monomers are the most challenging monomers, potentially poisoning the ethylene copolymerization [47]. Copolymer A in Fig. 2 is synthesized from the ethylene copolymerization with alkene-connected polar monomers. These copolymers led to a straightforward attachment of the polar groups to the polymer backbones. The ethylene copolymerization with long-chain polar monomers generates the copolymer B, where there is an alkyl spacer between the CC bond and polar groups. However, copolymer B can be generated from the ethylene copolymerization with alkene-connected polar monomers due to the chain-walking process. The functionalized copolymers can even be synthesized with both, in-chain and terminal polar groups, catalyzed by the α-diimine Ni and Pd catalysts. These unique structures of the long-chain polar monomers allowed the easier copolymerization with ethylene, reducing the possibility to poison the catalytic center by means of the coordination with polar sites. This strategy facilitates the polymer products with polar groups away from the polymer backbone. Both types of copolymers (A and B) provide very interesting and useful properties for future industrial application; and thus, it is currently the driving force in such research field [11,60]. Additionally, polar functionalized norbornenes are also an interesting class of polar substrates (Fig. 2). Ethylene-norbornene (E-NB) copolymers are an important class of polyolefins with the high refractive index and high transparency. They are suitable for optical applications including lenses, blister packs and medical equipment. As strong π-donors, norbornene-type monomers can efficiently coordinate with the metal centers compared to other polar vinyl monomers. The β-hydride elimination is relatively prevented by the cyclic structure. The presence of long spacers between the CC bond and the polar groups offers less likeliness for deactivation of the catalytic metal centers [63–66].As mentioned earlier, the α-diimine Ni and Pd complexes are unique in their capacity to control the microstructure of the resulting polymer, while allowing ethylene copolymerization with polar monomers. An additional advantage of the α-diimine Ni and Pd complexes over their early-transition-metal competitors is the ease of synthesis and higher air stability [42,45]. The synthesis largely involves two main reactions (Fig. 3 ); namely, 1) the α-diimine formation from the reaction of modified anilines and diketones in the presence of the catalytic amount of the acid; 2) the coordination reactions of α-diimine ligands and metal (Ni and Pd) salts or complexes to synthesize α-diimine Ni and Pd complexes. The resulting complexes are very stable to moisture and oxygen. This inertness of α-diimine Ni and Pd catalysts suits the industrial application, where storage stability is a well-known issue. The structural modifications of anilines and diketones initially based on ligands bring about the versatile synthesis of α-diimine Ni and Pd complexes and different coordination environments to the catalytic metal centers.Generally, the catalytic performance of a metal center is influenced by the electronic and steric effects of the ligand structures [67]. Similarly, previous researchers have reported that high pressures, low polymerization temperatures, bulky backbone substituents, and/or N-aryl groups enable polymerization of high-molecular weights, low branching densities, and thus higher melting point [30]. On the other hand, low pressures, high temperatures, and the use of less bulky ligands result in polymers with lower molecular weight, a higher degree of branching, and low melting points. The electronic effects of the ligand are also crucial. Studies on ligands with electron-withdrawing substituents induce the electrophilicity at the catalytic metal center, which indirectly promoted chain propagation, thus enabling higher molecular weight of polyethylene [43,45]. In summary, the catalytic performance can be readily tuned with variations of ligand structures and reaction conditions. In the following section α-diimine complexes are grouped in four classes based on their structural features and reaction conditions, i.e: N-aryl modifications (Table 1 . and Figs. 4-18), backbone modifications (Table 2 . and Figs. 19-23), binuclear complexes (Table 3 . and Figs. 24-28), functional group modification (Table 4 . and Figs. 29-36).In recent years, one of the most popular modifications on α-diimine Ni complexes has been the incorporation of the 2, 6-dibenzhydryl group on the ortho-N-aryl positions reported by Rhinehart et al. in 2013 (Fig. 4 )[68]. These α-diimine Ni complexes exhibited remarkable thermal stability in ethylene polymerization. Activated by methylaluminoxane (MAO), the catalytic activity of Ni precatalysts (C1b in Fig. 4) as high as 2.81 × 106 g of PE (mol of Ni) -1h−1 at 100 ℃ (10 min) (Table 1) was estimated. The resulting polymer showed a well-defined and narrow molecular weight distribution (M w/M n ≤ 1.31), moderate degree of branching (63 to 75B/1000C), and high molecular weight (M n > 600 000 g/mol). The robust nature and thermal stability of C2 achieved even higher activity (1.4 times in TOFs) than C1 in ethylene polymerization [69]. An increase in melting point by ∼20 °C and fewer branching content was also observed for polyethylene. In addition, these Ni complexes exhibited the capacity for living polymerization at 75 ℃ [70].Inspired by the C1 and C2 structures, Dai et al. have further developed the 2, 6-dibenzhydryl-substituted α-diimine Pd complexes (C3 in Fig. 5 ) containing either electron-donating (-OMe, - Me) or electron-withdrawing groups (Cl, -CF3) [71]. In this work, new synthetic strategies led to the improvement in the yield of sterically demanding ligands using the more efficient fashion (yield above 90 %). In comparison to the classical Pd complex A ( Fig. 1), these Pd complexes exhibited remarkable thermal stability and catalytic properties in ethylene polymerization. Higher catalytic activity up to 3.2 × 106 g of PE (mol of Pd) -1h−1 (60 °C, 15 min) was achieved, accompanied by a higher molecular weight (Mn up to 538 000 g/mol) and lower branching density (23–29B/1000 C) of polyethylenes (Table 1). The melting point (T m) of the synthesized polyethylene was as high as 99 °C. A semi-crystalline E-MA copolymer was synthesized through Pd-catalyzed copolymerization. It was notable that the slow-chain-walking behaviors of these Pd catalysts resulted in unique polymer microstructures (higher molecular weight and lower branching).Some new α-diimine Ni complexes (C4 and C4′ in Fig. 5) bearing similar coordinating structures as C3 were synthesized by Guo et al. [72]. Namely, C4 and C4′ displayed a very high catalytic activity of 6.18 × 106 g of PE (mol of Pd) -1h−1 (100 °C, 30 min) (Table 1). The synthesized polyethylene exhibited a molecular weight of more than one million with a rather narrow PDI. All Ni complexes exhibited a robust catalytic behavior combined with high activity, producing a high molecular weight of polyethylenes. Ni complexes were able to polymerize the ethylene even at 100 °C. In the ethylene polymerization, the dibromonickel-catalyzed polymerization was relatively insensitive to the electronic perturbation introduced by structure C4′. In contrast, electronic effects of the Ni(acac) systems (C4) were clearly observed. Trifluoromethyl-substituted C4d catalyst exhibited exceptionally high activity and thermal stability at elevated temperatures.Simultaneous tuning of both electronic and steric effects was rarely investigated in previous studies. Muhammad et al. advanced a series of symmetrical α-diimine Ni and Pd catalysts bearing both benzhydryl N-aryl with various substituents of methoxy/fluoro groups (C5 in Fig. 5) for ethylene polymerization and copolymerization with AA and MA [73]. The six methoxy/fluoro substituents located both at para and meta-positions were able to significantly enhance and alter the electronic effects of ligand and the coordination environment around catalytic metal center. It was hypothesized that the meta-methoxy groups on C5-Pd-OMe interacted with the benzhydryl groups, increasing the steric constraints around the metal center. This palladium (C5-Pd-OMe) and nickel (C5-Ni-OMe) catalysts exhibited an increased catalytic activity [up to 5 × 106 g of PE (mol of Ni) -1h−1] 1 (100 °C, 30 min). The resulting polymers had a high molecular weight (2.54 × 106 g/mol) and thus a high melting point (T m = 112.2 °C), along with the reduced polymer branching densities (Table 1). Accordingly, improved mechanical properties were observed. For C5-Pd-OMe catalyzed copolymerization, the incorporation of monomer (MA and AA) in the polymer chains were reduced because of the ligand's bulkiness.In 2018, Guo et al. reported a series of sterically hindered and acenaphthene-based α-diimine nickel complexes with the remote R' (-OMe, -Me, -tBu, -Ph) groups in para-positions of diarylmethyl moiety (C6 in Fig. 6 ) [74]. Activated by the Et2AlCl, these nickel catalysts exhibited high activities [up to 5.1 × 106 g of PE (mol of Ni) -1h−1] (20 °C, 30 min) and high thermal stability (stable at 100 °C) in ethylene polymerization. The synthesized polyethylenes were characterized as ultra-high molecular weight (UHMWPE) (M w up to 4.5 × 106 g/mol) with a moderate branching in the range of 26–71B/1000C (Table 1). It was interesting that the presence of remote substituents (-OMe, -Me,-tBu, and -Ph) in the para-position had a strong influence on the catalytic properties of these corresponding α-diimine nickel complexes and the UHMWPE mechanical properties. These branched UHMWPE materials displayed the typical properties of thermoplastic elastomers with well-defined stress–strain curves and elastic recovery.More recently in 2020, Xia et al. introduced the concerted double-layer steric strategy concept in designing a new series of α-diimine nickel catalysts. This method involved the incorporation of bulky diphenylaniline into the ligand structures (C7 in Fig. 7 ) [75]. These newly designed α-diimine Ni and Pd catalysts exhibited both significant thermal stability (stable at 150 °C) as well as very high activity [on the level of 1.03 × 109 g of PE (mol of Ni) -1h−1] (30 °C, 1 min) in ethylene polymerization. The resulting polyethylenes exhibited ultrahigh molecular weight (M w = 4.2 × 106 g/mol) with a controlled degree of branching from quasi-linear (2B/1000C) to lightly branched (32B/1000C) structures (Table 1). Ethylene copolymerization with a good incorporation of methyl 10-undecenoate was also observed. The key structural innovations introduced in C7 explained its typical catalytic performance. The first steric layer offered by the inner phenyl rings provided enough space for ethylene coordination and insertion. The second steric layer from outer phenyl rings gave rise to restraining chain transfer. This strategy of catalyst design led to simultaneously high catalytic activity and high molecular weight of polyethylene, normally better than the reported conventional modifications. Kanai et al. initially reported the Nicomplexes with symmetric bowl-shaped α-diimine ligands, consisting of two pentiptycenyl-substituents in [(a-diimine)NiBr2] (C8a in Fig. 8 ). The Ni complexes displayed good catalytic activity in ethylene (co)polymerization [76]. The Et2AlCl /C8a system exhibited a moderate catalytic activity of 3.4 × 104g of PE (mol of Ni) -1h−1 (25 °C, 30 min) at 7 atm. The molecular weight of the resulting polyethylene reached up to 1.5 × 105 g/mol with low branching densities (12B/1000C) and a high melting point of 133 °C. The catalytic performance of these complexes structures was attributed to the coordination environment of nickel, which was located in a highly shielded, hemispherical, and crowded space of the two pentiptycene units. Polar monomers (UAME, UA, UCl, and UO in Fig. 2) and ethylene could be efficiently copolymerized, leading to the copolymers with 4.2 mol% incorporations of polar monomers. The copolymerization highly depended on the amount of activator and was relative to the amount of polar monomer. By lowering the molar ratio of Et2AlCl/polar monomer to 0.5, the decreased activity for ethylene copolymerization with oxygen-containing monomers was observed, while 11-chloro-1-undecene could still be efficiently copolymerized with ethylene under the same conditions.Similar to the above work, Liao et al. then proposed a novel series of sterically demanding pentiptycenyl N-aryl substituted α-diimine Ni and Pd catalysts for ethylene (co)polymerization (C8b to f) [77]. These newly synthesized complexes were characterized as highly bulky substituents and backbone. For ethylene polymerization (20–80 °C) catalyzed by Ni complexes, the catalytic activities were in the range of 0.64 to 3.74 × 106 g of PE (mol of Ni) -1h−1 (60 °C, 30 min). The generated polyethylenes exhibited a moderate molecular weight of 3.77 × 105 g/mol, tunable branching densities from 6B/1000C to 55B/1000C, and high Tm (135 °C) (Table 1). These Pd catalysts ensured respectable MA incorporation up to 4.1 mol% in the copolymerization with ethylene. Compared with the free rotation of dibenzhydryl substituents, the restricted rotation of pentiptycenyl substituents offered superior activity and a slower chain-walking process for α-diimine Ni(II) species. These special bulky groups also enhanced comonomer incorporation for α-diimine Pd(II) species. It was notable that less steric blockage of substituents at the axial positions on the catalytic metal center led to a decrease in the molecular weight of the resulting polymer.In 2016, Dai et al. reported the synthesis of novel naphthalene and benzothiophene substituted N-aryl groups on α-diimine Pd complexes (C9a, C9b in Fig. 9 ) [78]. In ethylene polymerization, these Pd complexes displayed the moderate catalytic activity around 4.1 × 105 g of PE (mol of Pd) -1h−1 (60 °C, 15 min) and good thermal stability (Table 1). The produced polyethylenes achieved extremely high molecular weights (8.02 × 106 g/mol), low branching densities (as low as 6B/1000C), and comparatively high melting points (Tm up to 127.2 °C). In ethylene-MA copolymerization, the catalytic activity was in the order of 3.03 × 104 g of PE (mol of Pd) -1h−1. The E-MA copolymer possessed rather high molecular weights (M n up to 4.42 × 105 g/mol). The initial α-diimine catalyst A (Fig. 1) was deactivated in the presence of a long-chain polar monomer, which was speculated to the fast chain-walking process of catalyst A. Compared to catalyst A, C9a and C9b were more efficient in the copolymerization of long-chain monomers achieving high activity. The molecular weight of the copolymers was close to 1 × 106 g/mol. The surface wetting property of the resulting polymer was indeed improved via this incorporation of the polar functional groups into the polymer chains.In 2018, Na et al. demonstrated the specially designed α-diimine Pd complex containing steric thienyl-phenyl substitution (C10 in Fig. 9) [79]. The properties of hence generated polyethylene were similar to low-density polyethylene (LDPE). Tunable branching densities (16 to 37B/1000C), high melting points (Tm 101 to113 °C), and low polymer densities (0.89–0.92 g/cm3) were observed. Polar-functionalized low-density polyethylene (P-LDPE) was synthesized via ethylene copolymerization with polar monomers. The catalytic activities during copolymerization were up to 105 g of PE (mol of Pd) -1h−1. Copolymers with high incorporation (6.8 %), high molecular weights (M n up to 1.24 × 106 g/mol), high melting points (118 °C), and tunable branching densities (14 to 46B/1000C) were achieved. The incorporation of polar groups significantly influenced the mechanical as well as the surface wetting properties of the resulting copolymers. Rishina et al. investigated the catalytic effects from fluoro (C11a) and trifluoromethyl (C11b) substituted N-aryl modifications on α-diimine Ni catalyzed ethylene and propylene oligomerization (Fig. 10) [80]. Oligomerization of ethylene with C11 activated by a mixture of Et2AlCl /EASC and PPh3 at 30 °C resulted in only oligomers (i.e. oligomerization degree: 6 to 9) as mixtures of wax and liquid. A microstructure study indicated that the oligomers contained 14 to 20 mol% of methyl branches, 4 to 6 mol% of ethyl branches, and a small number of long-chain branches. In propylene oligomerization, these catalysts produced mixtures of very short oligomers (mostly dimers) at elevated temperatures from 30 to 70 °C. C11bNi generated the active species that exhibited no regioselectivity. Compared with C11b, the preference for primary insertion was observed in the oligomerization catalyzed by the C11a Ni complex. Mundil et al. designed the series of α-diimine Ni and Pd complexes bearing fluorinated alkyl substituents at the para-N-aryl groups C12 (Fig. 10 ) [81]. These complexes were used to carry out catalyzed polymerization of ethylene, propene, and 1-hexene. Remarkably, there were no significant effects on the catalytic properties due to fluoroalkyl groups. The branching densities of the generated polyolefins were rather tunable by the ligand's backbones and ortho-substituents of N-aryl groups.In 2017, Lian and Wang et al. reported the novel synthesis of PTPE-type polyethylenes through ethylene polymerization, which was catalyzed by the Xanthene substituted N-aryl of α-diimine Ni and Pd (C13 in Fig. 11 ) [82,83]. These α-diimine Ni complexes revealed rather high activities [up to 6.94 × 106 g of PE (mol of Pd) -1h−1] (20 °C, 30 min) and thermal stability at 80 °C for ethylene polymerization. The generated polyethylenes exhibited high molecular weight (M n up to 1.53 × 106 g/mol) and notably narrow molecular weight distributions (Table 1). The remote substituents (-Ph, -CF3, –NO2, and -OMe) had again a dramatic influence on the catalytic properties of ethylene polymerization. Specifically, the nickel complexes bearing the -Ph substituent (C13Ni-Ph) led to the formation of polyethylenes with exceptional elastic properties due to the branched structure of polyethylene (elastic strain recovery value of 70 % via C13Ni-Ph at 40 °C). The catalytic properties of Pd complexes were investigated in ethylene polymerization and ethylene/MA, ethylene/NB, ethylene/5-norbornene-2-yl acetate copolymerization. High molecular-weight E-MA and E-NB copolymers were produced by Pd-catalyzed copolymerization. C13Pd-Ph exhibited much higher activity (up to 2.5 × 104 g of PE (mol of Pd) -1h−1) than other complexes, and generated polymers and copolymers with much higher molecular weight (M n up to 1.21 × 105 g/mol).In 2017, Li et al. investigated a series of α-diimine Ni and Pd complexes bearing nitrogen-containing cyclic groups (C14 in Fig. 12 ) [84]. In ethylene polymerization, the nitrogen atoms situated on N-aryl groups of C14a and C14b interacted with catalytic metal centers in a remarkable manner. The catalytic metal center was then highly tolerant to polar functional groups, while the polymer branching densities were significantly reduced. The generated polyethylenes were characterized as nearly perfect linear structure (branching < 1B/1000C and high Tm (>130 °C) (Table 1). For the unsymmetrical PdC14c, a moderately branched polymer (around 70B/1000C) was produced. The presence of nitrogen also improved the thermal stability of the catalysts. Continuously high activity was achieved in polymerization even after 2 h at 60 °C, while the high molecular weight of polymers was still achieved at 80 °C. Additionally, these newly designed α-diimine Ni and Pd complexes were able to copolymerize ethylene with a series of polar comonomers such as UAME and the challenging AAc. Linear E-MA copolymers with high incorporation (up to 7.5 mol%) were still achieved in copolymerization. Furthermore, the functional-group tolerance of the catalytic metal center was greatly enhanced by the presence of additional side-arm heteroatoms. The high tolerance and unique catalytic performance were attributed to a “second-coordination sphere” strategy. It was hypothesized by the author that a second coordination sphere of the ligands was stronger than β-H or β-X (X being a polar group). Nevertheless, the activity was still weaker than with ethylene insertion. It was noteworthy that the experimental data and the hypothesis was backed up by a computational study.Compared to rigid steric modifications, Guo and Dai et al. worked with a series of flexible alkyl (C15) and cycloalkyl (C16) substituted N-aryl units of α-diimine Ni and Pd complexes (Fig. 13 ) [85,86]. Compared with fixed phenyl substitutions on α-diimine Ni complexes (B in Fig. 1), the flexible cyclohexyl complexes exhibited distinctive catalytic behavior. The polymers were generated with remarkably high branching densities, low Tm, and low molecular weight. PTPE-type polymers with these highly branched polyethylenes were obtained. The flexible cyclohexyl Pd catalyst exhibited a remarkably higher catalytic activity than the conventional α-diimine complexes (A in Fig. 1), providing an increased amount of molecular weight and branching density in ethylene (co)polymerization. The flexible modification of cyclohexyl α-diimine Ni and Pd complexes offered a much faster chain-walking process and higher catalytic activity. The synthesized polymers possessed higher molecular weight with appreciable comonomer incorporation. In terms of long-chain alkyl α-diimine Ni and Pd complexes, the Ni complexes presented high activities [up to 2.14 × 106 g of PE (mol of Ni) -1h−1](20 °C, 30 min) and generated highly branched polyethylene with ultra-high molecular weight (M n up to 1.2 × 106 g/mol) (Table 1 ). The synthesized polyethylene also displayed exceptional capacity in mechanical elasticity like TPE-type polymers (SR value up to 88 %). Brookhart et al. have described two “sandwich-like” arrangements of naphthyl substituted N-aryl groups in α-diimine Ni-based precatalysts (C17aNi in Fig. 14 ) [87]. The two 8-p-tolylnaphthylimino moieties were implanted on the α-diimine ligands to shield the Ni-axial direction and thus control the monomer insertion. The tolyl substituents are arranged perpendicular to the naphthyl rings, which were nearly coplanar with the square coordination plane. Activated by the modified methylalumoxane, these distinctive Ni complexes produced hyper-branched (up to 152B/1000C) polyethylenes with high molecular weights. It was believed that the increased axial bulk efficiently led to lower rates of chain transfer, relative to increase the chain propagation rates. Meanwhile, it resulted in high molecular weights and narrow PDIs of polyethylene. Subsequently, Vaidya et al. have developed additional similar derivatives of “Sandwich”-type nickel complexes (C17bNi) [88]. They were applied to catalyze higher α-olefin polymerizations with precise control of the chain-walking process, which favors the ω, 1-enchainment. The “sandwich” type catalysts also provided the chance to synthesize the low-branched polyethylene with a “chain-straightened” semi-crystalline property (high melting point). With the activation of MAO, O'Connor et al then used the C17bNi complex to generate polyolefin-based PTPE-type block copolymers (Table 1). The 1-decene monomer was responsible for high crystallinity and hard blocks, while low crystallinity soft blocks were synthesized from ethylene monomer [89]. Various block structures were characterized as copolymers in the range of block size from diblock to heptablock. All resulting polymers behaved as elastic PTPE-type materials. More recently, Allen et al. have reported newly synthesized “sandwich” types of α-diimine palladium catalysts (C17aPd in Fig. 14 ) for ethylene polymerization [90]. The Pd complexes were used in the ethylene polymerization with typical signs for living polymerization at 25 °C. The resulting polyethylene was hyper-branched and exhibited narrow molecular weight distribution (around 1.1). Ethylene copolymerization with MA using the Pd catalysts presented a high percentage of incorporation, which was up to 14 %. Zhai et al. reported a new ortho-menthyl substituted N-aryl on the α-diimine Ni complexes as the syn- and anti-conformers for ethylene and 1-hexene polymerization (C18 in Fig. 15 ) [91]. Both anti- and syn-conformers of C18b could be activated by the Et2AlCl for polymerization. The catalytic activity could be achieved from 2.5 to 6.6 × 106 g of PE (mol of Ni) -1h−1 (15 psi of ethylene pressure at room temperature and 15 min) in ethylene polymerization (Table 1). The syn- and anti-conformers of C18 exhibited a different catalytic performance. The polyethylene produced by syn-conformers tends to entail a higher molecular weight and branching density than the one obtained from the anti-conformer catalyzed polymerization. Compared to polyhexene produced from anti-conformer, the polymer produced by syn-conformer possessed a higher level of chain straightening and a higher percentage of methyl branches rather than butyl branches. This result also indicated a greater preference for the 2,1-insertion and chain-walking process for the syn-conformer C18syn. In recent years, Sun et al. contributed considerable research towards the α-diimine Ni-catalyzed ethylene polymerization [45]. All finely tuned α-diimine nickel complexes in Fig. 16 are selected examples of such modifications on the N-aryl groups [92–112]. The Ni complexes exhibited outstanding catalytic activity and generated polyethylene of high molecular weight and highly branched microstructures. For instance, C19Ni were unsymmetrically synthesized with various and modified benzhydryl substitutions on one of the two N-aryl groups. Activated by MAO, MMAO, Et2AlCl, or EASC, a remarkable catalytic activity of 1.48 × 107 g of PE (mol of Ni) -1h−1 could be reached. Super highly branched polyethylene with branching densities as high as 337B/1000C were obtained. The polymer produced solely by ethylene monomer exhibited typical PTPE properties. The molecular weight of the resulting polyethylene was in the range of 105 to 3 × 106 g/mol, which is characteristic of UHMWPE and it exhibited a narrow molecular weight distribution (Table 1). Complexes C20 were synthesized with the incorporation of a 2,4- or 2,4,6-substitution pattern using the steric benzhydryl groups. As a remarkable feature, high activity was retained at the high thermal stability of these catalysts. Activated by MMAO cocatalyst, C20aNi (ortho-R group as -Me) exhibited high activities of up to 8.9 × 106 g of PE (mol of Ni) -1h−1 and resulted in highly branched polyethylenes (166B/1000C) at 80 °C. In presence of relatively low amounts of EASC, C20bNi exhibited higher activities compared to C20aNi [up to 10.9 × 106 g of PE (mol of Ni) -1h−1], retained thermal stability by maintaining high activity (3.76 × 106 g of PE (mol of Ni) -1h−1) at 80 °C. C20c with the 2,4,6-substituted benzhydryl presented remarkable activity even at 90 °C [2.97 × 106 g of PE (mol of Ni) -1h−1]. The synthesized polyethylene contained hyperbranched microstructures as high as 135B/1000C, which were analyzed as methyl (84.4 %), ethyl (5.6 %), and longer chain branches (10 %). C21 containing N-naphthyl ligands displayed moderate activity. Activated by either MAO or Et2AlCl, C21 yielded polyethylene with typically low branching densities and a high melting point (131 °C). The catalytic performance of unsymmetrical N-naphthyl complexes C22 was investigated to determine the influence of the bulky difluorobenzhydryl substitution. With the activation of Et2AlCl, C22 exhibited high activity, exhibiting a narrow molecular weight of polyethylene (1.22–1.99). Activated by MAO and Et2AlCl, C23aNi showed high activities in ethylene polymerization with the activity of up to 107 g of PE (mol of Ni) -1h−1, illustrating the feature of the single-site active species and observing narrow molecular weight distributions of the resultant polyethylene. C23bNi exhibited higher activity over C23aNi, which was ascribed to the electron withdrawing nature of the para-fluorides and their influences on the active mental centers. C23cNi/EASC system generated both high activity and thermal stability, catalyzing the ethylene polymerization effectively even at 80 °C [6.01 × 106 g of PE (mol of Ni)−1h−1 (30 min)], while yielding high molecular weight (as high as 10.6 × 105 g mol−1) polymers for the same reaction time. Notably, the polyethylene produced by C24Ni was characterized as ultra-high molecular weight (>1 × 106 g mol−1) with a relatively high degree of branching (115 branches per 1000 carbons). C25Ni and C26Ni displayed a moderate catalytic activities up to 106 g of PE (mol of Ni)−1h−1 (30 min) upon the activation with either MAO and EASC. Particularly, C26Ni exhibited good thermal stability in the temperatures range from 60 to 80 °C, generating the polyethylenes with high molecular weight. Activated by either MAO or Et2AlCl, C27Ni exhibited outstanding catalytic activity in ethylene polymerization [1.02 × 107 g of PE (mol of Ni)−1h−1 (30 min) ]. C27Ni bearing the equivalent difluorobenzhydryl-substituted N-aryl groups was observed to have exceptional thermal stability. C27Ni/ Et2AlCl system exhibited high activity [ 1.02 × 107 g of PE (mol of Ni)−1h−1 (30 min) ] at 100 °C, while generating the polyethylene with the high molecular weight. Significantly, the polyethylenes possessed exceptional elastomeric recovery and high elongation at break determined by DMA and stress–strain testing. In summary, these research offered a promising route to alternative materials for the conventional thermoplastic elastomers (TPEs). Zhai et al. reported a series of novel α-diimine Pd complexes containing the secondary amide (−CONHMe) or tertiary amide (-CONMe2) substituents on the N-aryl groups (C28 in Figure 17 ) [113]. These Pd complexes were investigated in the catalytic performance of ethylene polymerization, ethylene/MA, and ethylene/AA copolymerization. With the replacement of two i Pr units using -CHPh2 groups, C28 led to a significant improvement in catalytic performance. The generated (α-diimine)PdMe+ species (C28) were activated by NaBArF 4 to produce polyethylenes with a molecular weight of around 5.9 × 104 g/mol. Compared to C29, the structure C28 exhibited lower catalytic activity [3.9 × 104 g of PE (mol of Pd) -1h−1] (20 °C, 120 min) (Table 1). This was due to the enhanced steric effects of -CHPh2 groups, which counteracted the negative effect of electron-withdrawing amide units. The resulting polymers exhibited moderate branching content (77 to 81B/1000C). In addition, the Pd complex of C28 incorporates higher levels of MA and AA in the copolymerization with ethylene than C29. Hu et al. recently reported the newly developed α-diimine Ni and Pd complexes with the steric enhancement of unsymmetrically pentiptycenyl-dibenzhydryl substituted N-aryl modifications (C30 in Fig. 18 ) [114,115]. It demonstrated some new catalytic features of C30 compared to previous studies [45]. Within a relatively long-lived reaction time, the increased bulk of α-diimine Ni and Pd complexes indeed raised the molecular weight of polyethylene, while this trend differed in the short polymerization time. With increased steric, the branching density initially increases followed by a decrease. In Ni-catalyzed ethylene polymerization, the activity can be as high as 6 × 106 g of PE (mol of Ni) -1h−1 (20 °C, 5 min), producing the polyethylene with ultrahigh molecular weight (1.58 × 106 g/mol) (Table 1). The C30Ni enabled the ethylene copolymerization with UAME, generating copolymers with high molecular weight (2.13 × 105 g/mol) and branching density (138B/1000C). Compared to C30Ni, the catalytic activity of C30Pd is comparatively lower.A substantial class of structural variations relates to modifications at the reactive center backbone in the catalyst. The table below summarizes structures C31-C38 and each of these structural variations will then be described in detail below. Song et al. explored the synthesis and characterization of α-diimine Ni dihalides (Cl and Br), bearing the 4,5-bis (arylimino)pyrenylidene (C31 in Fig. 19 ) [116]. After activation with a very low amount of cocatalysts including MAO, EASC, and Et2AlCl, the structure C31 exhibited high activity up to 4.41 × 106 g of PE (mol of Ni) -1h−1 (40 °C, 30 min) (Table 2). The microstructure of the synthesized polyethylene was analyzed using high temperature NMR, which revealed a high degree of branching density (up to 130B/1000) and narrow molecular weight distributions (around 2.5). This work also indicated that reaction parameters like the Al/Ni molar ratios, the reaction temperature, and polymerization time had a significant influence on the catalytic activity and the properties of the generated polyethylenes. Liu et al. proposed two chiral α-diimine Ni complexes containing (1R)- and (1S)- camphyl substituted backbone for ethylene and a-olefin polymerization (C32in Fig. 20 ) [117]. In this catalytic system, the chiral tunes on the ligand structure exhibited no influences on the catalytic behavior and region-selectivity for the Ni catalyzed polymerization. Activated by Et2AlCl, the catalytic activity of C32 revealed characteristics of a living polymerization for ethylene, propylene, 1-hexene, and 4-methyl-1-pentene under the optimized conditions (Table 2). Rather narrow molecular weight distributions (PDI < 1.2) were observed in the produced polypropylenes and poly(1-hexene)s with a wide range of polymerization temperatures. A high 1,3-enchainment fraction of 45 % was also observed in C32Ni-catalyzed propylene with polymerization at −60 °C, which was attributed to the 2,1-insertion of propylene and a chain-walking process. Zou et al. reported a series of α-diimine Ni(II) and Pd(II) complexes with different substituents on the acenaphthyl backbones (C33a-e in Fig. 21 ) [118]. The corresponding complexes were synthesized， characterized, and applied to the ethylene polymerization and E-MA copolymerization. In terms of ethylene polymerization, NiC33 a-d complexes exhibited high activities of up to 1.6 × 107 g of PE (mol of Ni) -1h−1 (20 °C, 10 min) (Table 2). The synthesized polyethylene displayed high molecular weight (M n) (up to 4.2 × 105 g/mol) with a molecular weight distribution of around 2.5. The structural variations C33 a-e had similar catalytic performance in ethylene polymerization. However, the polymer obtained from C33e-catalyzed polymerization was confirmed to exhibit much higher molecular weight and lower branching density than polymers from other catalysts. Substituents on the ligand backbones of PdC33 a-e had a significant influence on their catalytic performance during ethylene polymerization and E-MA copolymerization. The polyethylene and E-MA copolymers produced by the PdC33e complex exhibited higher molecular weights than polymers from PdC33 a-d.Along these lines, Zhu et al. reported the modification of acenaphthy backbones on the α-diimine Ni complexes (C33f in Fig. 21 ), which has a similar structure as C33a-e complexes [119]. Experimental and computational studies were carried out to reveal and analyze the thermal stability of the proposed α-diimine Ni complex (C33f). Compared to the initially reported α-diimine Ni complexes (A in Fig. 1), the complex C33f presented higher activity and thermal stability at elevated temperatures (Table 2). It was found that the presence of ethylene evidently affected the conformation of the C1–N1–Ni–N2–C2 five-membered ring (where the nickel center is located) of C33f. According to calculations, differences in the decomposition energy between C33f and A in Fig. 1 .were observed. Zhang et al. explored the synthesis and characterization of new nickel bromide complexes containing the rigid bidentate bis(arylimino)camphane ligands with different N-aryl substituents (C34 in Fig. 22 )[120]. Upon activation with either MMAO or Me2AlCl, the newly synthesized NiC34 complexes exhibited high catalytic activities and thermal stabilities during ethylene polymerization (as high as up to 11.2 × 106 g PE (mol Ni)−1 h−1 (80 °C and 30 min)), producing PEs of high molecular weights (23.7 × 105 g mol−1) and low PDI (1.6–2.6) (Table 2). In this catalytic system, the introduction of di(p-fluorophenyl)methyl on the ortho-position of N-aryl groups resulted in increasing of both the catalytic activity and the thermal stability of the corresponding Ni complexes. The synthesized PEs were moderate to highly branched nature with the tunable branch contents governed by various ligand structures. The bulkiness of substituents in the ligand structure led to high molecular weight polymers with a low branching -degree and limited -types. The polymers produced with ortho-hydrogen NiC34b/Me2AlCl possessed the highest branching density with the unique terminal vinyl (–CH═CH2) and internal vinylene (–CH═CH–) structures. Long et al. proposed the newly synthesized α-diimine Ni complex containing dibenzobarrelene-bridged backbone (C35 in Fig. 23 ) [121]. C35 exhibited an exceptional catalytic behavior in ethylene polymerization. The DBB-bridged C35 resulted in a steric hindrance around the cationic Ni center, decreasing catalytic deactivation and slowing the chain-walking process. Therefore, C35 produced linear polyethylene in ethylene homopolymerization. C35 Ni afforded a high molecular weight of polyethylenes (7.12 × 105 g/mol) with narrow distribution (1.18) and low branching densities (<1B/1000C) (Table 2). It also revealeda living polymerization behavior at room temperature, producing linear polyethylene with a high melting point (T m = 135 °C) (Table 2). It facilitated the copolymerization of ethylene with the methyl 10-undecenoate to yield highly linear ester-functionalized polyethylene (T m values at128 °C and 1 mol % comonomer incorporation). In the presence of ester functional groups, the catalytic activity during copolymerization dropped by an order of magnitude. This was hypothetically attributed to the reversible coordination between the ester-functionalized co-monomer and the cationic Ni center. Zhong et al. reported a series of α-diimine Ni and Pd complexes with the modifications of dibenzobarrelene-bridged backbone. (C36a in Fig. 23 ) [122]. These complexes also revealed high thermal stability and clear signs of living polymerization. PdC36a exhibited the ability for precision synthesis of functionalized copolymers by living ethylene copolymerization with various acrylate monomers (Table 2). The bulky enhancement of dibenzobarrelene backbone improved the insertion selectivity of methyl acrylate (MA) in a 2,1-insertion. This catalytic behavior prevented polar groups from poisoning the active PdC36a species. In this living chain-walking system, it was demonstrated that the composition, molecular weight, and branching topology of the copolymer could be controlled by the variation of the ethylene pressure. Based on backbone modifications of C36a, a series of novel dibenzobarrelene-derived α-diimine nickel complexes were also synthesized and applied in ethylene polymerization (C36b-d in Fig. 23 ). The increased steric effects on the ligand backbone and the repulsive interactions inhibited the N-aryl rotation of the α-diimine ligands and enhanced the thermal stability of the complexes. The living polymerization could be achieved at 80 °C. Bulky enhancement of dibenzobarrelene backbone also improved tolerance of Ni complexes towards the ethylene copolymerization with monomers containing polar groups. The living ethylene copolymerization with methyl 10-undecenoate was also carried out by the bulky NiC36d. PdC36e was applied in a precision synthesis of functionalized polymers by the living ethylene copolymerization with the variety of acrylate monomers [123]. The incorporation of the steric dibenzobarrelene backbone significantly improved the migratory-insertion selectivity of methyl acrylate (MA) in a 2,1-insertion manner. This bulk-enhanced strategy prevented the polar groups from poisoning palladium centers of the catalysts by a formation of the five-membered palladacycle intermediates PdC36e exhibited a living polymerization and good thermal stability (55 °C) during ethylene polymerization. Living ethylene copolymerization with MA monomer were also successfully achieved, which was in contrast to general knowledge that polar monomers poison the transition metal catalysts. PdC36 was reported to perform the (co)polymerization of petroleum-based ethylene and bio-based furfuryl acrylate by Du et al.[124].The cationic palladium catalyst exhibited higher thermal stability than the neutral chloromethyl palladium complex, while the later complex was more active at low temperature. The incorporation of tert-butyl on the dibenzobarrelene backbone improved the tolerance of the PdC36 toward polar groups such as the incorporation of furyl groups into the polymer chain. Ethylene living (co)polymerization with the furfuryl acrylate (FA) catalyzed by PdC36 was successfully carried out, which afforded copolymers with a uniform incorporation of FA. The mechanistic study indicated that FA was selectively inserted into the Pd–Me bond in a 2,1-insertion mode. There was no interactions observed between the palladium center and the furyl ring. Then, the dinaphthobarrelene-based backbone of α-diimine Ni and Pd complexes was synthesized, providing three-dimensional confinement for ethylene (co)polymerization (C37 in Fig. 23) [125]. These increased steric effects generated a 3D-confined space around the catalytic Ni and Pd centers, which strictly shielded the back and axial direction of the α-diimine Ni and Pd complexes. This confinement was assigned to the enhanced catalytic activity, thermal stability, and living fashion for ethylene polymerization (Table 2). The synthesized polyethylene possessed narrow molecular weight distribution (1.04–1,45). The steric accumulation effectively favored the ethylene copolymerization with polar monomers. Zhang et al. proposed two types of dibenzobarrenlen (backbone) and pentiptycenyl (N-aryl) substituted α-diimine Pd complexes (C38 in Fig. 23) [126]. Remarkably,the Pd complexeswere active in a very broad range of temperatures (from 0 °C up to 130 °C). The Pd complexes allowed the synthesis of polyethylene with high molecular weight (around 106 g/mol), see also Table 2. The microstructure analysis of the resulting polymers displayed highly methyl (220 Me/1000 C) branched features, which were close to the typical properties of commercial ethylene-propylene elastomers. In ethylene copolymerization with MA, elevating the reaction temperature from 30 °C to 90 °C gave rise to the increase in catalytic activity, molecular weight, and MA incorporation. This steric framework of C38 exhibited an abundant chain-walking process, while the formation of branches in the polymer structure was only limited to methyl branches. The region-selectivity of acrylate insertion was characterized as 1,2-insertion due to the steric constrains.A notable development of novel polymerization catalyst structures includes the integration of two active metal centers in asymmetric or symmetric binuclear complexes. It can pave the way to, for example, bimodal polymerization or synergistic activity enhancement. Table 3 summarizes these structures with the nomenclature C39-C44. Zhu et al. reported a series of binuclear α-diimine Ni and Pd complexes containing conjugated backbones (C39 in Fig. 24 ) [127]. Activated by MMAO, the catalytic activity of such complexes could reach up to 1.05 × 106 g of PE (mol of Ni) -1h−1 (RT, 30 min) (Table 3). The activity of NiC39b is almost twice as high as the remaining complexes of this family, provided identical polymerization conditions. The binuclear PdC39 produced polyethylene with bimodal features in GPC. This confirmed the simultaneous formation of two active species were in a binuclear catalyst system. The E/MA copolymerization was also investigated using the Pd complexes. The MA incorporation can be up to 2.36 mol% in the copolymerization catalyzed by PdC39 b. Xing et al. reported a series of binuclear α-diimine Ni(II) complexes consisting of 4,5,9,10-tetra(arylimino)pyrenylidene-bridged ligands (C40 in Fig. 24) [128]. To mediate ethylene polymerization, different parameters like type of cocatalysts, cocatalyst ratio, polymerization time, and temperature were varied to optimize the catalytic capacity of the binuclear Ni complexes. Both NiC40 (a and b)Ni complexes exhibited high activities [1.5 × 106 g of PE (mol of Ni) -1h−1] (30 °C, 30 min) in the presence of either MAO or Me2AlCl (Table 3). The Ni complexes exhibited a long time (60 min) of catalytic life when activated with MAO. The polyethylene obtained from the catalyzed polymerization revealed a minor amount of branches (7B/1000C). Compared to the analogous mononuclear complexes, these binuclear nickel complexes revealed no significant improvements in catalytic activity in ethylene polymerization. Wang and Na et al. demonstrated the xanthene-, naphthalene- and biphenylene-bridged α-diimine binuclear Ni and Pd complexes C41 (Fig. 25 ) [129,130]. These binuclear nickel complexes exhibit good thermal stability (stay active at 80 °C) during ethylene polymerization. They exhibited the catalytic activity up to 106 g of PE (mol of Ni) -1 (20 °C, 30 min) (Table 3). It was interesting to note that these binuclear Ni complexes exhibited higher activity and resulted in polymers with higher molecular weights than their mononuclear analogues. Polymers with a high molecular weight (M n), narrow PDI and low branching density were obtained from these catalysts. These results indicated that the Ni-Ni cooperativity slowed the β-hydride elimination and related chain-walking process. In terms of Pd-catalyzed ethylene (co)polymerization, the Pd-Pd cooperation had a significant impact on catalytic behavior, especially for E-MA copolymerization. No MA incorporation was observed in the case of polymerization with the binuclear Pd complexes, while the mononuclear analogues enabled MA incorporation. Kong et al. reported a series of methylene-bridged binuclear α-diimine Ni and Pd complexes C42 (Fig. 26 ) [131]. Uponactivationwith Et2AlCl or MAO, all the nickel complexes exhibited high activity towardethylenepolymerization [catalytic activity up to 7.86 × 106 g of PE (mol of Ni) -1h−1] (20 °C, 30 min). The resulting polymer displayed high melting points (Tm up to 130.9 °C) as well as high branching densities (151B/1000C) (Table 3). The binuclear Ni complexes exhibited a synergistic catalytic activity as compared to their mononuclear analogs. Furthermore, the synthesized polyethylene possessed a higher molecular weight and broader PDI (up to 4.8). Khoshsefat et al. reported the aryl-bridged binuclear α-diimine Ni and Pd complexes C43 (Fig. 27 ) [132]. Under the optimized conditions for ethylene polymerization ([Al]/[Ni] = 2000/1, 42 °C, 20 min), C43d reached its highest catalytic activity of 1.07 × 106 g of PE (mol of Ni) -1h−1 (42 °C, 20 min) (Table 3 ). Compared to other Ni complexes, the polyethylene produced from C43d also processed the highest molecular weight with the broad molecular weight distribution (PDI = 17.8). The bulky ortho-substituents on N-aryl groups presented positive influences on the catalytic activity, molecular weights, and degree of branching, which was confirmed by the theoretical study as well. Takano et al. have developed a unique binuclear double-decker structure of α-diimine Pd complexes containing the macrocyclic ligands (C44 in Fig. 28 ) [133–135]. In ethylene polymerization at high temperatures (at 60 °C and 100 °C), C44 exhibited more stability and higher activity than the mononuclear complexes (Table 3). The polyethylene formed by the binuclear catalyst C44 possessed less branched density (33B/1000) than the mononuclear catalysts (110B/1000C). A longer catalytic lifetime for C44 was also observed as it was still active after 18 h. In E-MA copolymerization, the use of binuclear C44 resulted in higher acrylate incorporation (5.2 mol %) in the copolymers than that formed by the mononuclear catalysts (1.1 mol%). NMR analysis confirmed branched structure for the E-MA copolymer. In E-AA (Acrylic Anhydride) copolymerization, C44 afforded copolymer containing a repeating unit of acrylic anhydride. High incorporation (up to 5.7 mol %) of cyclic and acyclic anhydride groups was evidenced in the main polymer chain, which was again much higher than the respective mononuclear complex (0.8 mol %). Zhong et al. developed α-diimine Pd complexes with two ferrocenyl units and applied them in the ethylene (co)polymerization (C45 in Fig. 29 ) [136]. The two ferrocenyl groups were sequentially and stepwise oxidized, which increased the electron-withdrawing capacity of the α-diimine ligand. This stepwise redox control was applied to modify the catalytic properties of α-diimine Pd complexes during the ethylene homopolymerization and copolymerization with polar monomers (norbornene, methyl acrylate, and 5-norbornene-2-yl acetate). The catalytic activity decreased as the two ferrocenyl units become oxidized. It seemed that the rates of chain propagation and chain transfer were greatly affected by this stepwise redox-control strategy. The branching density of polyethylene was only slightly increased along with the oxidation, while the polymer microstructure and PDI was significantly controlled during these stepwise oxidation processes. The same ligands containing two ferrocene units was also applied to the synthesis of the α-diimine Ni complex. The oxidation process of the ferrocene groups did not alter the catalytic behavior of the corresponding Ni complex in ethylene polymerization. It was proposed that the reducing nature of the aluminum species (MAO) was too strong to carry out such stepwise redox-control strategy for Ni-catalyzed ethylene polymerization.Peng et al. reported a new α-diimine Ni complex bearing azobenzene groups with photoresponsive properties toward ethylene (co)polymerization (C46 in Fig. 30 ) [137]. The axial steric environment of the metal center was directly influenced by the light-induced trans–cis isomerization. UV light can tune the properties of the ligand structure and therefore the catalytic behavior of the α-diimine Ni complexes for ethylene (co)polymerization. This light-induced control increased the polymer molecular weight and decreased the catalytic activity and the polymer branching density (Table 4). The authors suggested the incorporation of the photo-responsive functional units could induce even more dramatic changes in the electronic and steric coordination environments around the catalytic metal center.Metal-metal cooperation and synergistic effects have been extensively explored in ethylene polymerization, like the synthesis and application of the binuclear transition-metal catalysts. Contrary to this strategy, Wang et al. proposed a supramolecular chemistry strategy which was carried out to construct multinuclear catalysts for ethylene (co)polymerization. A new series of α-diimine Pd complexes was designed and synthesized with the urea-functional groups (H and N-methylated counterparts) (C47 in Fig. 31 )[138]. The experimental results indicate that self-assembly of Pd complexes took place via the urea-based hydrogen bonding interactions, which was evidenced from the Fourier transform infrared (FTIR) spectroscopy. During ethylene polymerization and copolymerization with MA, the catalytic activity and polymer properties such as molecular weight, PDI, branching density, comonomer incorporation of the (co)polymer were modified by the various catalyst concentration, ligand structures, and reaction conditions. In order to further explore the supramolecular-induced self-assembly effects, the photo-sensitive azobenzene group was incorporated in the urea-functionalized α-diimine Pd complexes. This work indeed showed the presence of photosensitive functional group in the nickel complex influenced the microstructure of the polyethylene.The aliphatic hydrocarbon solvents (hexane and heptane) were widely applied in the industrial research for ethylene polymerization, while the academic researchers predominantly worked on properties of the catalyst in aromatic solvents (toluene). In order to bridge this gap between academic studies to practical applications, Chen et al. designed the new α-diimine Ni complex with the diaryl-methyl aniline bearing eight tert-butyl groups (C48 in Fig. 32 ) [139]. The incorporation of the multiple tert-butyl substituents in the diaryl-methyl moiety increased both the ligand's steric and electronic-donating ability. It resulted in the enhancement of the catalyst stability and polymer molecular weight in ethylene polymerization. The presence of multiple tert-butyl groups enabled solubility of the metal complexes in aliphatic hydrocarbon solvents, leading to similar polymerization properties compared to the aromatic solvents. The incorporation of tert-butyl substituent in the ligands improves the solubility of the Ni complexes in typical polymerization solvents, which improves its application potential. Gong et al. developed a new series of acenaphthene-based sterically hindered α-diimine Pd complexes bearing bulky diarylmethyl moiety for ethylene (co)polymerization (C49 in Fig. 33 ) [140]. The π-π interaction between the acenaphthene moiety and the phenyl of diarylmethyl moiety was suggest to occur. The π-π interaction was considered as the capacity to freeze the N-aryl-bond rotation at room temperature, resulting the enhancement of the axial steric bulk and thus resulting in a low branching densities of the polyethylene. The highest catalytic activity after 60 min [6.73 × 104 g of PE (mol of Pd) -1h−1] was achieved at 60 °C. At high temperature, the effects of the π-π interaction was decreased while the branching densities of polyethylene was significantly increased. Consequently, the branching density and the microstructure of the polyethylene was modified by the various reaction conditions. In terms of the ethylene copolymerization with polar monomers, moderate catalytic activities (up to 6.8 × 104 g of copolymer (mol of Pd) -1h−1), high molecular weight copolymers (M n up to 4.8 × 105 g mol−1) and low incorporation ratios of polar monomers (up to 2.12 %) was observed. This work also indicated that the π-π interaction effect played a critical role in copolymerization as well as ethylene homopolymerization. Zhong et al. proposed a series of α-diimine Ni and Pd complexes with electron-donating/withdrawing groups on the dibenzobarrelene backbone for ethylene polymerization (C50 in Fig. 34 ) [39,141]. The electronic effects from the remote substitutions on the backbones influenced the catalytic performance of the corresponding Ni complexes. The electron-withdrawing halogens enhanced catalytic activity and polymer molecular weight, while electron-donating methoxy groups led to a decrease. C50 displayed the highest activity [5.6 × 105 g of PE (mol of Ni) -1h−1 (50 °C, 30 min)] and produced the highest molecular weight polyethylene (3.3 × 105 g/mol) (Table 4). Intra-ligand hydrogen bonding interactions (CH···OMe) were observed in the C50a. The weak and noncovalent interactions enhanced the catalyst thermal stability and brought about a living ethylene polymerization at high temperatures (80 °C) via inhibiting rotation of the N-aryl bonds. The dibenzobarrelene-based α-diimine Pd complexes (PdC50) exhibited the thermally robust characteristics for ethylene polymerization, due to the cooperative effect of hydrogen bonding interactions, electronic modification, and steric modification. The chloro-substituted Pd precatalyst presented the best thermal robustness for ethylene polymerization. In terms of the copolymerization with polar monomers, methoxy-substituted Pd precatalyst showed the most excellent tolerance toward both high temperature and polar groups. High MA incorporation (up to 9.5 mol%) can be achieved via the ethylene copolymerization with methyl acrylate (MA) at 80 °C. Compared to the previous strategies on weak noncovalent interactions in catalyzed ethylene polymerization, these hydrogen bonding interactions provided a fundamentally new approach in enhancing thermal stability of the α-diimine Ni and Pd complexes. Zheng et al. investigated the α-diimine Ni complexes with bulky 8-p-tolylnaphthylamine and dibenzo-/dinaphthobarrelene backbones for ethylene polymerization (C51 in Fig. 35 ) [142]. The weak Ni–phenyl interactions were considered as confining elements of the α-diimine Ni complexes. The interactions thus promoted the acceleration of the chain-growth process. 51Nia and 51Nib exhibited enhanced thermal stabilities and activities [1.39 × 106 g of PE (mol of Ni) -1h−1 (80 °C, 30 min)]. The synthesized PE was characterized as a linear semi-crystalline polymer. The combined experimental and theoretical study demonstrated the Ni-phenyl interactions decreased PE branching density in catalyzed ethylene polymerization. This work addressed the effects of Ni-phenyl interaction induced confinement, which provided an alternative strategy to prepare linear PE. Ni-phenyl interactions were also observed to promote E-MA copolymerization, where the MA incorporation was confirmed as 2.1 mol %. Wang et al. reported a new family of α-diimine Ni and Pd complexes bearing axially bulky terphenyl and equatorial bulky dibenzobarrelene groups (C52 in Fig. 36 ) [143,144]. Due to the presence of bulky groups, chain transfer was limited in the ethylene polymerization. These novel nickel complexes yielded polyethylenes of ultrahigh molecular weights (M w as high as 1.74 × 106 g/mol). Meanwhile, the use of the unsymmetrical skeleton with both bulky terphenyl group and less bulky aniline group resulted in a high catalytic activitiy as 1.52 × 107 g of PE (mol of Ni) -1h−1 (20 °C, 10 min) (Table 4). The incorporation of the electrowithdrawing trifluromethyl group into the terphenyl moiety cleary gave rise to H-F interactions between the N-terphenyl/anilinyl group and the dibenzobarrenlene backbone. This interaction efficiently suppressed the rotation of the N-terphenyl/anilinyl moiety, leading to relatively higher thermostability. The α-olefin polymerization studies revealed quite a large amount of ω,1-enchainments. Due to the axial and equatorial bulkiness and the weak H-F interactions, the deactivation of the active species at high temperatures was relatively reduced. At 80–140 °C, the Pd complexes efficiently catalyzed norbornene polymerizations with high catalytic activity (up to 5.65 × 107 g of PNB (mol of Ni) -1h−1) and yielding high molecular weights of PNB (up to 37.2 × 104 g/mol). Moreover, the Pd complexes could successfully promote the norbornene copolymerization with NB-MA, achieving moderate catalytic activities [104 g of copolymer (mol of Pd) -1h-1at 80 °C (60 min)].The immobilization of the catalytic metal center onto inorganic supports is a crucial step for gas and slurry-phase polymerization in the industry [145–147]. Compared to the common supported heterogeneous catalyst, α-diimine Ni and Pd complexes provided much higher catalytic activity up to 107 g of PE (mol of Ni) -1h−1 (or even higher) for ethylene polymerization. Molecular weight and molecular weight distribution of polyethylene could reasonably be controlled and modified using α-diimine Ni and Pd complexes, which gave us the chance to produce high-value polymers rather than the lower-quality polyolefins (broad dispersity and low molecular weight) from Ziegler-Natta systems [148,149]. Therefore, heterogenization of α-diimine Ni and Pd complexes provided a promising opportunity for a convenient “drop-in” approach for novel catalytic solutions in industrial-scale production processes. However, previous academic studies on α-diimine Ni and Pd complexes were mainly focused on homogeneous systems [44 150]. In industry, however, the heterogeneous polymerization is still the predominant polyethylene synthesis method. Consequently, it is rare to find commercial applications of α-diimine Ni and Pd complexes for olefin polymerization [151]. This section summarizes the various recently reported immobilizing methods (either physisorption or chemisorption) for α-diimine Ni and Pd complexes. Successful immobilization of the well-defined, single-site catalysts such as α-diimine Ni and Pd complexes on inorganic supports would definitely be the best solutions for their industrial applications. Heterogeneous catalysis has the advantage over the homogenous catalysis in terms of separating the catalyst from the polymer during recycling, which is considered as the key drawback of homogenous catalysis [44]. This strategy has been the most efficient tool to establish a heterogeneous platform for olefin polymerization.The chemisorption methods for α-diimine Ni and Pd complexes could be classified based on the types of immobilization techniques: i.e i) covalent attachment or ii) surface-bound anions as shown conceptually in Fig. 37 . The former method involved the covalent bonding between the silica supports and α-diimine Ni and Pd complexes, with the help of a linker (R in Fig. 37). The latter approach involves the treatment of the silica particles with cocatalysts (MAO, TAM, or others) to form surface-bound Al compounds. Then, the Al compounds served as the initiator to convert the complexes into active cationic metal-alkyl complexes. In this way, the complex was successfully immobilized onto the support particles through the ionic attraction to the surface-bound anions (electrostatic interactions) [150]. Wegner et al. has developed an efficient synthetic strategy for novel 2,5- and 2,6-phenyl substituted α-diimine Ni complexes (C53 in Fig. 38 ) [152]. The Ni complexes were directly supported during catalyst synthesis, without any further chemical links. These surface-bound catalysts were applied for ethylene polymerization in the gas phase. Unsupported C53a and C53b Ni complexes were benchmarked in the homogeneous solution for the comparison study, also with the gas-phase polymerization. The supported catalysts presented strong chemical stability. Even after several months, no decomposition was observed. The supported catalysts exhibited moderate activities in ethylene polymerization, which was up to 1.36 × 105 g of PE (mol of Ni) -1h−1 (30 °C, 60 min) (Table 5 ). The produced polyethylene exhibited the features typical for α-diimine Ni catalyst; i.e. ranging from HDPE to LLDPE with narrow PDI and low branching densities. In general, polyethylene produced in the gas phase exhibited higher molecular weights than those obtained from solution polymerization. The growth of single polyethylene particles was investigated in gas-phase polymerization using video microscopy. All the Ni complexes exhibited strong tolerance towards hydrogen additions during ethylene polymerization. Huang et al. has developed an ionic immobilized C19d Ni complex (Fig. 39 ) onto the silica via different Al organic compounds (Et3Al and Et2AlCl) (C19d@SiO2 ). The immobilized Ni complexes displayed moderate activities [0.68 × 106 g of PE (mol of Ni) -1 bar−1 h−1] (50 °C, 60 min) towards ethylene polymerization, initiated by the co-catalyst either i Bu3Al or Et2AlCl [153]. The synthesized polyethylene was characterized as medium branching densities (30–50 CH3/1000C), ultrahigh molecular weights (up to 2.2 × 106 g/mol), and narrow molecular weight distributions (2.1–2.4) (Table 5). The supported nickel complexes produced spherical particles of polyethylene upon slurry-phase polymerization, while noteworthy, there was no reactor fouling.Based on the initial complex A, Fevero et al. recently developed a series of functionalized α-diimine Ni complexes with covalent tethers (C54-58 in Fig. 40 ) [154]. The Ni complexes were covalently attached to the mesoporous silica (MCM-41) which were applied in ethylene heterogeneous polymerization. The Ni precatalysts were directly attached as the single precursor (C56 and C57) or a binary precursor (C58) to the TMA treated silica. Under optimized conditions, the binary catalysts exhibited similar catalytic performance as the homogeneous catalysts [as high as 3.97 × 106 g of PE (mol of Ni) -1h−1] (30 °C, 20 min) (Table 5). The analytical evidence from GPC and the thermal test indicated that both catalysts of the binary precursor were highly active during ethylene polymerization. C56 catalysts delivered high linear polymers while the C55 and C57 yielded branched polymers. The use of binary C58 catalyst resulted in polymers as the different functions, which was assumed to be due to the polymers' mixture. The functionalized tethers and covalent attachment both from the backbone and N-aryl groups were proven as a successful strategy to immobilize the α-diimine Ni(II) complexes on the solid support. Tafazolian et al. reported a unique form of immobilizing the α–diimine Ni(II) complex on the inorganic ZrO2 [155]. The calcined ZrO2 and dilute sulfuric acid solution formed firstly the sulfated zirconium oxide (SZO) sites, which contained Brønsted acids. These active sites allowed the ionic support of α–diimine Ni complex on the ZrO2, which was already partially dehydroxylated at 300 °C. In the MeCN/Et2O solution, the immobilizing reaction took place between the (α-diimine)NiMe2 complex and SZO, generating the ionic support and methane (C59 in Fig. 41 ). Under 45-psi ethylene pressure, C59b polymerized ethylene monomer in toluene with the TOF of 21000 h−1 at 40 °C (15 min) (Table 5). The high molecular-weight polyethylenes were produced in catalyzed polymerization (M n = 1.53 × 105 g/mol). The polymers exhibited moderate branching as 71B/1000C with a narrow PDI around 1.8. The elevated temperatures gave rise to a decrease in catalytic activities, polymer molecular weights, and molecular weight distribution. A steady decrease in the catalytic activity was observed with the increased reaction time in slurry-phase polymerization. Ethylene copolymerization with 10-undecenoate was also performed in this study. Compared to the polyethylenes, the synthesized copolymers presented a moderate molecular weight (M n = 2.97 × 104 g/mol) with broad dispersity (PDI = 5.17). Bahuleyan et al. have reported a leaching-free strategy to directly support the α-diimine Ni complexes on nonporous silica (C60and C61 in Fig. 42 ) [156]. This method avoids any tedious process, such as the chemical and thermal treatments of the silica substrate. The reactive amino groups on the α-diimine Ni complex provided the functionality to prepare the covalently supported catalysts. Firstly, amino-functionalized ligands reacted with the organosilane, 3-(triethoxy-silyl)propylisocyanate, which played the role of linkers on the ligands. Using the Stober method, the silica-supported ligands or complexes (after metalation with (DME)NiBr2) were prepared by the reaction between linkers with SiO2. Instead of the MAO-supported forms, the catalytic activity of the supported systems could be up to 106 g of PE (mol of Ni) -1 bar−1 h−1 (30 °C, 60 min) with activation of a very small amount of aluminum compounds (Al/Ni 100) like EASC, MeAlCl2, and Et2AlCl (Table 5). No visible leaching from the active metal centers or pollution of the reactor was observed in the heterogeneous ethylene polymerizations. The types of cocatalysts, catalytic activities, and metal loadings were considered the main factors influencing the morphology of polyethylenes. Electron microscopic investigations indicated a fibrous architecture for the polymers.In the field of coordination-insertion polymerization catalyzed by the late-transition metal complexes, the polymeric microstructures are largely governed by catalyst structures and reaction conditions. The control over microstructure has a direct influence on mechanical and other properties of the polymers [157]. Due to the chain walking process, α-diimine Ni complexes produced mainly branched polymers using only ethylene as the monomer feedstock [158,159]. The structural modifications on the ligand backbone and N-aryl substituents of the α-diimine Ni complexes played a crucial role in controlling the chain-walking behavior, which had a significant influence on the polymeric branching densities. The branching nature of the polymers were determined by the high temperature 1H and 13C NMR, where the branching degree (per 1000C) were calculated (Fig. 43 ) [54,107,160]. In the high temperature 13C NMR spectra, the types and percentage of the various branches could be characterized clearly, like methyl, ethyl, propyl, butyl, amyl and even longer chains. These branched polymers are part of the polyolefin thermoplastic elastomers (P-TPE) [157]. Compared to the thermoplastics polymers (like HDPE), a low Young’s modulus, high elongation at break, and high elastic recovery is typical for P-TPE-type materials. The stress–strain curves indicated mechanical properties like tensile strength and elongation at break that are typical for P-TPE materials (Fig. 44 A). The elastic recovery of the P-TPE materials was also confirmed (Fig. 44B) [82]. For example, Sui et al. synthesized a series of the elastic polyethylenes catalyzed by the unsymmetrical α-diimine Pd complexes, which displayed very nice mechanical properties [160]. High tensile strengths (18 MPa) and great elastic properties, like elongation at break close to 500 %, were observed. The molecular weight and broad PDI were considered as the main factor, influencing the mechanical performance of the polymers. In 2017, Lian et al. reported the synthesis of PE-based TPEs catalyzed by the α-diimine Ni complexes [82]. The tensile strength (3 to 28 MPa) and elongation-at-break (300 % to 1800 %) values of these polymers could be modified by different Ni complexes and polymerization conditions. An exceptional elastic recovery as high as 1605 % was noted. One year later, Fang et al. developed the newly synthesized elastomeric materials via ethylene polymerization, which exhibited good tensile strength ranging from 7.4 to 16.3 MPa and elongation-at-break values ranging from 450 % to 700 % [158]. Notably, the mechanical properties of the polymers were significantly influenced by their molecular weights and branching densities. In addition, an elastic recovery up to the strain recovery values of 83 % was observed. More recently, Liu et al. reported a series of highly branched polyethylene (161 branches /1000C) catalyzed by the trifluoromethoxy-substituted α-diimine Ni catalysts [107]. The lowest tensile strength was up to 3.3 MPa, while the highest elongation at break was observed as 984 %. The elastic recovery were determined as 71 %. The molecular weight, crystallinity, and the alkyl-branching architectures significantly influenced the tensile strength of these polymers. The distinctive properties of the produced polymers was believed to be the promising alternatives to the commercial thermoplastic elastomers.Besides the unique elastic properties, the copolymers incorporated with the polar monomers catalyzed by the α-diimine Ni and Pd complexes exhibited remarkable hydrophilicity (Fig. 45 ) [79]. This hydrophilicity could be easily determined via the water contact angles (WCA) measurements. For instance, Dai et al. synthesized the E-MA copolymers with good surface properties obtained from α-diimine Pd complexes [161]. The WCA values gradually decreased from 104° to 54° with increasing incorporation of MA monomer. The hydrophilicity of polar-functionalized polymers was reported to strongly related to the structures and contents of the polar groups. Normally, the presence of the hydrogen bond donors in the polar groups (such as –COOH and –OH units) brought about a higher hydrophilicity than the rest of polar monomers [65,78,162]. The stereochemistry of the polar monomers and the microstructures of the copolymers also affected the polymeric surface property (hydrophilicity). In contrast, with the use of α-diimine Pd complexes it was much more challenging to directly achieve the P-TPE materials due to the superior chain-walking tendency [161 163]. It led to the formation of the highly branched (amorphous) polymers with poor mechanical properties. Recently, Dai et al. reported the synthesis of polar functionalized P-TPE via the ethylene copolymerization with 10-undecenoic acid catalyzed by the α-diimine Pd complexes [163]. The polar copolymers displayed characteristics of thermoplastic elastomers with great elastic recovery (SR = 72 %-80 %) (Fig. 44B), while the good surface properties enabled by polar monomers was well pronounced at the same time.This review summarizes the recent advances in the synthesis of the α-diimine Ni and Pd complexes applied in ethylene (co)polymerization. These α-diimine metal complexes achieved very high catalytic activities and thermal stability in ethylene (co)polymerization. The molecular weight, dispersity, branching densities, and melting points could be directly modulated via the finely controlled chain-walking mechanism. The synthesis of high-value polymers such as elastic polymers, LLDPEs, UHMWPEs, and functionalized copolymers can be realized by the ethylene monomer as the main feedstock. Although there are lots of advances in homogenous polymerization, the heterogeneous domains the industrial application. It is clear that the majority of current efforts in the development of α-diimine Ni and Pd complexes are more focused on homogenous (co)polymerization systems. Nevertheless, many fails and errors are inevitable along with the catalysis developments. The reports on heterogeneous polymerization catalyzed by these complexes are still rare. As the homogenous platform is well studied and established, the successful heterogenization of the well-defined α-diimine Ni and Pd complexes could become the primary research emphasis in the future. Due to the unique catalytic performance, these late-transition metal complexes can compete with or outperform present-day catalysts. This heterogeneous strategy will ensure the further control of the polymeric microstructure, morphology, and macro performance on an industrial scale.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was partially supported by Subitex (2020-2025) and China Scholarships Council (No. 201904910562).

α-Diimine Ni and Pd complexes are one of the most examined late-transition organometallics in the application of catalyzed ethylene (co)polymerization. These organometallic catalysts provide unique advantages and particular opportunities to tailor the architectures, composition, and topology of synthesized polymers through catalyzed polymerization. Two decades after their initial discovery, they are still drawing extensive attention in both academia and industry. More recently, researchers have studied the effect of structural modifications on the α-diimine Ni and Pd complexes and their catalytic behaviors in ethylene (co)polymerization. This review highlights the recent progress in the developments of α-diimine Ni and Pd complexes achieved in the last decade. The chain-walking mechanism as a unique catalytic behavior of α-diimine Ni and Pd complexes is also addressed. The versatile synthesis of ligands and complexes enables researchers to tailor the catalytic performance and the microstructures of polyethylene. Correlations between their structural tunes and catalytic behaviors, polymer properties, and the ethylene copolymerization with polar monomers are comparatively presented and discussed. The heterogenization study of α-diimine Ni and Pd complexes on a solid support for heterogeneous catalysis is also comprehensively summarized. The review is broadly classified into four sections which includes i) the coordination-insertion chemistry in ethylene (co)polymerization, ii) the modification of ligand structure, iii) their application in the field of heterogeneous polymerization, iv) and the properties of the synthesized polymers, followed by the short summary and outlook for their potential studies and applications.

Precious metal nanoparticles supported on metal oxides are used for chemical conversions in heterogeneous and electrocatalysis. Reducing the size of the particles increases the fraction of atoms present at the nanoparticles’ surface, and thus the per-atom efficiency. When the particles enter the subnano regime, quantum size effects can affect the catalytic activity. [1,2] In the ultimate limit, single metal atoms can be anchored directly onto the oxide support, and so-called single-atom catalysis (SAC) has emerged as a key strategy in heterogeneous and electrocatalysis in the last decade. [3–7] Unravelling how catalytically active metal atoms bind to the metal oxide support and interact with reactants is essential for understanding their properties. The local coordination environment has been shown to strongly influence the reactivity and stability of SACs, [8–11] but the structural details of the active sites are difficult to obtain from experiment. This is partly due to the structural inhomogeneity of powder supports, and partly due to the limitations of analytical techniques. In the absence of this critical information, the reaction mechanism is typically modelled computationally assuming an idealized low-index facet of the support material with the catalyst atom located at a high-symmetry site. Such models almost certainly do not represent the active catalyst, particularly for electrochemical applications, because the presence of water and/or hydroxyl groups is neglected.One approach to investigate the validity of the assumptions made in the computational modelling of SACs is to synthesize analogous systems experimentally. Such experimental modelling is achieved using single-crystalline metal-oxide supports where the atomic structure is well known. The metal of interest is evaporated directly onto the pristine surface in ultrahigh vacuum, which allows the most stable adsorption site to be determined. One can also selectively introduce molecules that might affect the stability of the system, and determine their individual impact unambiguously. For systems ultimately utilized in an aqueous solution, water is the obvious candidate. Recently, we demonstrated that Rh atoms sinter rapidly after deposition on a pristine α-Fe2O3 ( 1 1 ¯ 02 ) model support in ultrahigh vacuum (UHV) at room temperature but are stabilized as “single atoms” when the same experiment is performed with 10−8 mbar water in the background pressure. The enhanced dispersion occurs because the adatoms are stabilized by additional coordination to two OH ligands. [12] On the other hand, adsorbates can also induce sintering, as observed for Pd/Fe3O4(001) in the presence of CO. [13] In this work, we turn our attention to TiO2 as a model support. The thermodynamically stable rutile phase of TiO2, especially the (110) surface, has been widely investigated in surface science studies. [14,15] In SAC, the anatase polymorph (a-TiO2) is of particular interest because it becomes more stable in the nanoparticle form typically used as a support [16]. The reactivity of a-TiO2-supported SAC systems has been heavily investigated in recent years [17–23]. DeRita et al. [24], for example, have convincingly demonstrated that Pt adatoms are active for CO oxidation. The possibility that agglomerates might be responsible for the observed activity was ruled out by using a very low Pt loading; each support particle hosted on average just one Pt atom. DFT-based calculations accompanying the experiments suggested that Pt atoms are probably not stable on the bare a-TiO2(101) surface. Moreover, a single Pt adatom on top of the bare a-TiO2(101) surface also failed to reproduce the experimentally observed binding energy and vibrational frequency of Pt-adsorbed CO molecules. These benchmark parameters were best reproduced when the Pt atoms were assumed to be coordinated to two additional oxygen atoms originating from hydroxyl groups on the surface. [24] Another study by the same group [25] reported that the coordination of Rh adatoms on a-TiO2 is sensitive to the composition of the reducing gas that was used to activate the catalyst. When Rh was pre-treated in CO at 300°C and further exposed to CO, Rh(CO)2 species formed with Rh being bound to two O2c from the lattice. (For a sketch of the a-TiO2 surface structure and the adsorption site, see Fig. 1 , below) When the system was pre-treated with H2 at 100°C, hydroxyls formed. These coordinated to the Rh(CO)2 species by adding an additional neighbouring surface OH group which substantially changed the CO binding energy. It was concluded from CO FTIR-TPD and DFT that the presence of hydroxyl groups can alter the local metal coordination and molecular desorption significantly. [25] Here, we present a surface science investigation study of four different metals – Pt, Rh, Ni and Ir – vapor-deposited directly onto an a-TiO2(101) single crystal support at room temperature in UHV. We find that Ir is the only metal that exhibits atomic dispersion under UHV conditions. However, the presence of water de-stabilizes the Ir adatoms, which leads to the formation of large clusters anchored at step edges. Pt, Ni, and Rh all form mostly clusters even at very low coverages, suggesting diffusion is facile on the regular terrace at room temperature. For Pt and Ni, small protrusions are observed in the STM images that we tentatively assign as isolated adatoms immobilized at defects.Room-temperature scanning tunnelling microscopy (STM) was performed in a two-vessel UHV chamber consisting of a preparation chamber (base pressure p < 10−10 mbar) and an analysis chamber (p < 5 × 10−11 mbar). The analysis chamber is equipped with a nonmonochromatic Al Kα X-ray source (VG) and a SPECS Phoibos 100 analyzer for XPS, and an Omicron μ-STM. The STM measurements (positive sample bias, empty states) were conducted in constant current mode with an electrochemically etched W tip. The natural a-TiO2(101) single crystal was prepared in UHV by sputtering (Ar+, 1 keV, 10 min) and annealing (610 °C, 20 min). Every fifth cycle the sample was annealed at 500 °C for 20 min in 5 × 10−7 mbar of O2 and then in UHV at 610 °C. [26] Pt, Rh, Ni and Ir were deposited using an e-beam evaporator (FOCUS), with the flux calibrated using a water-cooled quartz microbalance (QCM). One monolayer (ML) is defined as one metal atom per surface unit cell. (The areal density of unit cells is 5.15 × 1018 m− 2) The STM images were corrected for distortion and creep of the piezo scanner as described in ref [27]. The gray scale of each image is set individually to ensure that the possible adatoms and other small adsorbates are easily distinguishable. For measurements of the apparent height of adatoms or clusters, the average height of the surrounding substrate in the STM images was defined as a height of zero. Fig. 1a shows STM images of the a-TiO2(101) surface after several cleaning cycles. As is typical for this well-studied surface, a clean sample exhibits rows of bright, oval-shaped protrusions running in the [010] direction. These are attributed to the surface Ti5c and O2c atoms shown in Fig. 1c, d. [28] The [10 1 ¯ ] direction cannot be easily determined from these images; it was inferred from the preferred step directions. [28] Isolated dark features (highlighted with a blue arrow in Fig. 1a between the rows are inhomogeneously distributed over the surface. These have previously been attributed to extrinsic Nb dopants, which are often present in natural anatase TiO2 samples. [29] Our XPS survey spectra did not show a peak that would allow us to confirm or debunk this assignment, likely because their average coverage is too low (0.02–0.03 ML as measured by STM). For what follows, it is important to note that surface oxygen vacancies (VOs) are not present at the surface of a-TiO2(101). Even when formed artificially, they quickly diffuse to the subsurface at room temperature [30]. This is in stark contrast to rutile TiO2(110), where VO sites are prevalent and active sites for adsorption [14]. Fig. 1b shows the surface after 2 hours of exposure to the residual gas of the preparation chamber at room temperature (with the evaporator turned on but the shutter closed). Bright protrusions are observed; they appear identical to those observed after water adsorption in low-temperature studies [31]. Since water is known to desorb from regular a-TiO2(101) surface sites below 250 K [32], we conclude these water molecules are adsorbed at surface defects. The concentration agrees with that of the dark defects highlighted in Fig. 1a. Interestingly, the water molecules are mobile at room temperature (Fig. 2 ), but do not leave a visible defect behind when diffusing. This suggests that water and defect probably diffuse together, which makes it unlikely that the defect can be a cation substituting Ti in the anatase lattice. It could conceivably be an interstitial lattice species, or perhaps a surface site above a subsurface defect such as an oxygen vacancy. The images also exhibit a low concentration of molecular O2 [33] species, which are in the residual gas as left-over from the oxidation step during sample preparation. This species is also most likely bound at defects, and a few examples are highlighted in orange and marked as (O2)extr in Fig. 1b, consistent with the labelling in ref. [29]. These (O2)extr are also mobile at room temperature, and in a rare case we observed one of them to hop onto a dark defect, without leaving a similar defect behind (Fig. 3 ). This suggests that there may be defect sites that are not visible in STM images, where adsorbates can bind more strongly than at regular surface sites. Overall, these data show that the regular anatase surface is inert at room temperature, but that defects (both visible and invisible in STM) can act as binding sites for molecular adsorbates. In what follows, we will show that these defects can also stabilize metal adatoms.A previous STM study of the Pt/a-TiO2(101) system [34] revealed that small clusters form predominantly on the terrace, with some species tentatively assigned to adatoms. Our data (Fig. 4 a for a coverage of 0.05 ML) are similar to those in presented in ref. [34], and we also observe the coexistence of larger clusters and smaller features that have a uniform apparent height of 150–160 pm. At a lower coverage of 0.01 ML (Fig. 4b), the density and average size of the clusters is lower, and the 150–160 pm species are observed again. These smallest Pt species are easily distinguished from adsorbed water by their larger apparent height at our imaging conditions (60–80 pm for water, see Fig. 4c for a comparison), and because they are immobile in room-temperature STM movies. Given their relatively small apparent height, we tentatively assign these protrusions to Pt1 species. From the experiment alone, we cannot discount that these species could be dimers (or trimers) if such species were significantly more stable. For the higher coverage (0.05 ML), ≈7 % of the deposited Pt (according to the QCM calibration) can be attributed to possible single atoms, whereas at 0.01 ML this increases to ≈17 %. Fig. 5 a shows a high-resolution image, in which orange dots mark the approximate position of surface Ti5c atoms. Assuming that the substrate maxima imaged by STM are closer to the Ti5c than the O2c, the Pt-related protrusion is close to the position predicted by DFT calculations in ref. [34] (in between two adjacent O2c atoms, grey circles in Fig. 1d). We also note that the Pt adsorption site is equivalent to the sites where the dark defects are seen in STM (Fig. 5f), so it is possible that these defects stabilize the Pt atoms. Adsorbed water and O2 are also labelled in Fig. 4a and 4b for ease of comparison. Fig. 4d shows an STM image of Pt deposited in a water vapor background of 2 × 10−8 mbar. Again, a mixture of clusters and possible adatoms is observed, and the ratio of clusters and possible single atoms is comparable to that obtained in UHV. We thus conclude that water has no significant effect on the dispersion of Pt on the a-TiO2(101) terraces, at least in this low-pressure regime. Fig. 5b shows that the possible Pt adatom adsorption site is the same, independent of whether deposition was done in a water vapor background or in UHV. Fig. 6 shows the a-TiO2(101) surface after deposition of 0.02 ML Rh (a) in UHV and (b) in a water vapor background of 2 × 10−8 mbar. Unlike Pt, Rh forms exclusively small clusters on the surface despite the presence of the dark defects. We did not observe any features that we would attribute to single atoms, irrespective of the environment. We conclude that Rh1 species are not stable on the a-TiO2(101) surface at room temperature under our conditions. This is similar to our experience with r-TiO2(110) [37], where Rh1 species were found to sinter already at 150 K. Fig. 7 shows the surface after deposition of 0.02 ML Ni in a) UHV and a water vapor background of 2 × 10−8 mbar. Like Pt, Ni forms a mixture of clusters and small, uniform features that could be attributed to adsorbed single atoms. The coverage of these small species is relatively high: assuming that they are Ni1 they would account for ≈20 % of the deposited Ni, with the rest contained within larger clusters. The smallest species are easily distinguished from adsorbed water, partly by their apparent height (150–170 pm), and also because they are immobile on the a-TiO2(101) surface at room temperature. Thus, in analogy to Pt, we presume that the smallest Ni species are most likely trapped at defect sites. Fig. 5c and d show that these species are adsorbed at a different location on the surface than the features attributed to Pt atoms.After deposition in a water background of 2 × 10−8 mbar, the concentration of the Ni1 species doubles from 20 % of the deposited Ni to 40 %. There is no discernible difference between the protrusions in the two experiments, so it seems that water has a significant effect on the dispersion of Ni and may play a role in stabilizing Ni at defect sites.The last metal investigated in this study was Ir. Fig. 8 shows 0.02 ML Ir deposited in UHV. Unlike Pt, Rh and Ni, Ir forms mostly uniform features with an apparent height of 130–170 pm. All these features occupy the same site on the surface and are immobile at room temperature. Like Pt, Ir is located approximately between two O2c surface atoms (Fig. 5). We assign these features to single Ir adatoms, which appear at a coverage of 0.011 ML on the surface. In addition to the single atoms, a small number of clusters can also be recognized. The apparent height of all features is depicted in a histogram in Fig. 8b. A clear peak exists at 150 pm due to the features attributed to single atoms, with the shoulder at larger apparent heights originating from clusters. Considering that each cluster contains several Ir atoms, the coverage of the ≈150 pm high features agrees nicely with the assignment to single atoms. If the smallest species were dimers, our QCM calibration would have to be significantly underestimating the amount of Ir deposited. Increasing the Ir coverage to 0.05 ML (Fig. 8c) increases the density of clusters but does not affect the density of adatoms. Fig. 8d shows the influence of water on the Ir/a-TiO2(101) system. Deposition in water at room temperature leads to complete sintering of the single Ir atoms and the formation of large clusters at the step edges. This de-stabilizing effect of water is different to all the other metals studied here.We also performed XPS measurements on the four metals deposited in UHV and in water vapor. Fig. 9 shows an overview. For Pt, Rh and Ni, the peaks are shifted towards higher binding energy than those of the respective pure metals in the bulk. Water did not cause any drastic peak shifts, but intensity changes consistent with the propensity of dispersion/cluster formation observed in STM. The peak maxima are marked with a dotted line.Overall, this study shows that Pt, Rh and Ni readily sinter after deposition onto the a-TiO2(101) surface in UHV conditions. Rh is particularly unstable, and forms small clusters even at the lowest coverage studied with no evidence of any adatoms. Pt and Ni exhibit a mixture of small clusters and small, uniform features, which we assign as single atoms. Ir, in contrast, is highly dispersed at low metal coverages, but clusters begin to form when the coverage is increased. Our analysis of the adsorption site suggests that the adatom-assigned protrusions are between two surface O2c atoms for Pt and Ir, which is consistent with the site predicted for Pt by several DFT studies [34,35,38]. If the metal atoms bind to O, it is clear that the behaviour of the different metals can be understood in terms of the different oxygen affinities. Campbell and co-workers [39] recently studied adsorption of several late transition metals on MgO(110) and CeO2−x(111), and reported the trend Ir > Ni > Pt > Rh for the oxygen affinity, which matches the relative stability observed for the UHV experiments here.One issue with the assignment of adsorption at a regular lattice site is that the diffusion barrier for Pt atoms along the [010] direction has been calculated to be 0.86 eV [38]. Such a value means that Pt atoms could diffuse at room temperature on the ideal surface, which is likely why the majority of Pt atoms form small clusters before our STM measurements are conducted. Consequently, we infer that the immobile adatoms we observe at room temperature must be trapped at defect sites.TiO2 is sometimes considered synonymous with oxygen vacancies, because the behaviour of rutile TiO2(110) [37] in UHV is dominated by VO sites. DFT calculations suggest that Pt atoms would indeed be highly stable at surface VO sites (4.71 eV, compared to 2.20 eV on the pristine surface) [34], but it is known that VOs are preferentially accommodated in the subsurface layers. It is possible that such a large energy difference could cause a Vo to diffuse to the surface [40] in the presence of Pt atoms, but this is inconsistent with our STM results for Pt and Ni: In this case, the adatom would sit on an O2c site, not in-between as is consistently observed with Pt and Ir (Fig. 5). Consequently, we conclude that the immobile Pt and Ir species are stabilized by another defect type. Ni on the other hand is slightly shifted from the Pt and Ir adatoms and could therefore possibly be stabilized by a Vo.The dark defects observed in STM are a primary candidate for the stabilization of metal adatoms because the defect is also located between two O2c atoms (compare Fig. 5a and f). However, since the density of these defects is very inhomogeneous, which hinders any analysis of the number of defects covered by other species, we cannot exclude that another defect also plays a role. In any case, the nature of the dark defect is not clear. The previous assignment to substitutional Nb dopants [29] seems unlikely given the diffusion behaviour observed in the presence of water (Fig. 2). Similar logic leads us to exclude that the defect is linked to substitutional Fe cations, although we do observe a small Fe2p signal in XPS survey spectra as this metal is a common contaminant in natural crystals. Nevertheless, Fe tends to be localized in patches on the surface, and its appearance differs significantly from the dark defects at standard imaging conditions. [26] Finally, we can also exclude that the dark defects are mistaken for an adsorbate, because the appearance of most candidate molecules present in the residual gas (water, O2, CO, OH) has already been established on the a-TiO2(101) surface. [29,31] We propose that the defect is most likely a dopant atom present in an interstitial site in the lattice. Hebenstreit et al. speculated that it could possibly be a Ti interstitial.[41] While we cannot positively identify the chemical nature of the dark defect at this stage, our results suggest that extrinsic doping of the oxide could be a strategy to provide stronger binding sites capable of immobilizing expensive metals on a-TiO2(101).Turning now to the effect of water, we first note the completely different behaviour of Rh on a-TiO2(101) compared to our previous study on α‑Fe2O3 ( 1 1 ¯ 02 ) . In that work, depositing the metal in a background of water led to complete dispersion because Rh adatoms were stabilized by additional OH ligands [12]. One possible difference here is that water is already partially dissociated on α‑Fe2O3 ( 1 1 ¯ 02 ) at room temperature [42], so OH ligands are more readily available than on a-TiO2(101) where water adsorbs molecularly. Another difference is the surface geometry: On α‑Fe2O3 ( 1 1 ¯ 02 ) , OH groups adsorbed on nearby surface Fe cations can create a square planar environment for the Rh1 adatom [12], which is known to be energetically favourable. [43] On a-TiO2(101), this is not possible because OH groups adsorbed on surface Ti cations can at best create a threefold coordination, assuming the Rh1 remains coordinated to two surface O atoms.The only case where the dispersion seems to be aided by water is Ni, and our data suggest that ≈40 % of the deposited metal can be stabilized as isolated adatoms. The apparent height and adsorption site is the same as it was in the absence of water, so it is possible that some of the Ni1 were already partially stabilized by the water inadvertently present in the residual gas in the UHV experiments. The complete opposite effect was seen after the deposition of Ir in a water vapor background. The presence of water promotes a dramatic sintering of the dispersed species, leading to mobile clusters, which finally get trapped at the steps. Clearly, then, the effect of water is difficult to predict, and given that water and hydroxyl groups are always present on metal-oxide surfaces, its omission from computational modelling of SAC systems is likely a major oversimplification. It is also important to recognise that water can have a significant effect on the reactivity, and there is evidence that water can play a role in SAC reaction mechanisms [44,45].Finally, one of the goals of this study was to assess the suitability of the a-TiO2(101) surface as a model support for surface science studies of SAC mechanisms. While it would be possible to study adsorption at the adatoms by STM/nc-AFM, the ambiguity over the nature of the defect sites that stabilize the adatoms precludes reliable modelling of the system. In any case, the presence of clusters at low coverage will make it difficult to distinguish the reactivity of single atoms from clusters using area averaging-techniques. At present, it is difficult to recommend this system as a suitable model system for studies of single-atom catalysis.We have carried out room-temperature STM measurements of the a-TiO2(101) surface after deposition of Pt, Rh, Ni and Ir in UHV and in a water vapor background. Pt and Ni form a mixture of small clusters and, possibly, single atoms. Rh exclusively forms clusters, while Ir is highly dispersed at a low coverage. The influence of water strongly varies from metal to metal. No influence is discernible for Pt and Rh, but the dispersion of Ni is increased when deposition is performed in a water vapor background. The exact opposite effect occurs in the case of Ir, which rapidly sinters after deposition in a water vapor background. The adsorption site of the species attributed to Pt and Ir atoms is the same as calculated for Pt1 on the pristine surface; nevertheless, there is evidence that the single metal atoms are trapped at defect sites on the a-TiO2(101) surface. As such, we conclude that doping of oxide surfaces could be a viable strategy to provide strong adsorption sites for single metal atoms. Lena Puntscher: Data curation, Investigation, Writing – original draft. Kevin Daninger: Data curation, Investigation. Michael Schmid: Writing – review & editing. Ulrike Diebold: Writing – review & editing. Gareth S. Parkinson: Funding acquisition, Supervision, Conceptualization, Writing – review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.LH, KD, and GSP acknowledge funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No. [864628], Consolidator Research Grant “E-SAC”). UD acknowledges funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme grant agreement No. [883395], Advanced Research Grant ‘WatFun’). The authors acknowledge TU Wien Bibliothek for financial support through its Open Access Funding Programme.

Understanding how metal atoms are stabilized on metal oxide supports is important for predicting the stability of “single-atom” catalysts. In this study, we use scanning tunnelling microscopy (STM) and x-ray photoelectron spectroscopy (XPS) to investigate four catalytically active metals – platinum, rhodium, nickel and iridium – on the anatase TiO2(101) surface. The metals were vapor deposited at room temperature in ultrahigh vacuum (UHV) conditions, and also with a background water pressure of 2 × 10−8 mbar. Pt and Ni exist as a mixture of adatoms and nanoparticles in UHV at low coverage, with the adatoms immobilized at defect sites. Water has no discernible effect on the Pt dispersion, but significantly increases the amount of Ni single atoms. Ir is highly dispersed, but sinters to nanoparticles in the water vapor background leading to the formation of large clusters at step edges. Rh forms clusters on the terrace of anatase TiO2(101) irrespective of the environment. We conclude that introducing defect sites into metal oxide supports could be a strategy to aid the dispersion of single atoms on metal-oxide surfaces, and that the presence of water should be taken into account in the modelling of single-atom catalysts.

Biomass is a promising renewable source for the production of fuels, chemicals and H2 [1–4] due to its availability and CO2 neutral contribution. Besides, H2 is a clean fuel whose potential as an energy carrier makes it a promising choice for its transformation into any form of energy for diverse end-use applications.In recent decades, the progress of technological strategies for H2 production from biomass has gained increasing attention in the literature [5–9]. Two types of routes are used for biomass conversion into H2, as are thermochemical and biological processes [10]. Thus, biomass may be converted into H2 through the following routes: i) water biophotolysis using micro-algae and cyanobacteria, ii) photofermentation, iii) dark-fermentation, and iv) hybrid reactor system [5]. The biological H2 production (biohydrogen) from microorganism metabolism is a promising technology under development, in which renewable sources can be used for the sustainable production of H2 [11,12]. Amongst the different routes for biomass valorization, thermochemical processes have merited especial consideration in the literature [8,13,14], particularly biomass steam gasification [15–19], fast pyrolysis [20,21], and the steam reforming of the bio-oil obtained in the pyrolysis process [22–25]. More recently, the alternative route of biomass pyrolysis and in-line catalytic steam reforming has attracted remarkable attention for the production of H2 from biomass [26–31]. Most pyrolysis-reforming studies conducted in the literature have been performed in discontinuous mode using batch reactors. However, a great effort has been made in the recent years in order to implement a continuous feeding system, and therefore to step further in the scaling-up of this process.The choice of a suitable catalyst for these processes is of uttermost significance for the viability of their industrial implementation. Accordingly, primary catalysts, such as dolomite, olivine, γ-Al2O3 or spent fluid catalytic cracking (FCC) catalyst, have been widely investigated in biomass gasification [32–35]. Thus, several authors have reported the activity of dolomite and olivine for reforming and cracking reactions [36,37], whereas γ-Al2O3 is effective in tar decomposition and promoting H2 production [38]. The utilization of a spent FCC catalyst is of special relevance, since it increases the lifetime of a refinery waste material [39,40].Besides, commercial Ni-based catalysts have been extensively used in steam reforming processes, since these types of catalysts involve several advantages, such as their lower cost compared to noble based catalysts, as well as their high activity for breaking C − C and O − H bonds. However, their fast deactivation, mainly by coke deposition on the active sites, is a great challenge to face up. Accordingly, different strategies have been proposed with the aim of improving the activity and stability of the reforming catalyst, as are the use of different reactor configurations, the selection of suitable operating conditions or the optimization of the catalyst design based on supports and promoters [8,41–43].The strategy proposed in this study to attenuate the fast catalyst deactivation lies in the modification of the feed into the reforming step, which may be conducted in the pyrolysis reactor itself or downstream by upgrading the bio-oil produced in the biomass pyrolysis. Thus, it is well established that certain bio-oil compounds reaching the reforming catalyst bed significantly influence the mechanisms of catalyst deactivation, particularly coke formation, and therefore the performance of the catalyst in the reforming step [44]. Although huge effort has been made to assess the catalyst performance and its deactivation causes in the steam reforming process, the highly complex bio-oil composition and the problems associated with its handling boosted use of bio-oil model compounds, such as ethanol, phenol, acetic acid, toluene, or their mixture [45–47]. Therefore, knowledge of the reforming catalyst performance and the main species responsible for catalyst deactivation by coke deposition under real process conditions is still limited.Based on this background, this study pursues a dual objective. On the one hand, to enhance catalyst activity and stability in the reforming step in a tandem pyrolysis-reforming reactor by conditioning the biomass pyrolysis volatile stream using highly available and inexpensive catalysts. On the other hand, to provide a further understanding of the reactivity of the main bio-oil oxygenate compounds and their role in the deactivation of the reforming catalyst. The results obtained in this study will contribute to progressing towards the understanding of the main coke precursors in the reforming step and promoting the proposal of new strategies for improving catalyst stability, which are the main challenges to be faced in the industrial implementation of the pyrolysis-reforming process.Accordingly, the production of H2 from biomass (pinewood sawdust) has been carried out in a conical spouted bed reactor (CSBR) for the fast pyrolysis, and an in-line fixed bed reactor for the reforming of the volatiles produced [48]. For the conditioning of this stream, different low cost materials (inert sand, γ-Al2O3, a spent FCC (fluid catalytic cracking) catalyst and olivine) have been located prior to the reforming catalyst. Continuous biomass pyrolysis has been conducted in a conical spouted bed reactor (CSBR) and the volatiles formed have been transferred into a fixed bed reactor for their conditioning and reforming. Thus, the fixed bed reactor includes two reaction sections, the first one with the guard catalyst (conditioning step) and the second one with the steam reforming one. In a previous study, this reactor configuration revealed a high efficiency for the conversion of the volatiles derived from biomass pyrolysis into a hydrogen rich syngas, with catalyst deactivation being lower than when a fluidized bed reactor is used in the reforming step [48].The biomass used in this process was pinewood waste (pinus insignis), which was crushed and sieved to a particle size in the range from 1 to 2 mm. The ultimate and proximate analyses were determined in previous studies in a LECO CHNS-932 elemental analyzer and in a TGA Q5000IR thermogravimetric analyzer, respectively [26,49]. An isoperibolic bomb calorimetry (Parr 1356) was used to determine the higher heating value. Table 1 summarizes the main biomass features. Fig. S1 in the Supplementary Information shows the TG profile of the pinewood sawdust.Four different low-cost materials have been used in this study as guard catalysts for the conditioning of the pyrolysis volatile stream, as are: i) inert silica sand (Minerals Sibelco), ii) olivine (Minerals Sibelco), with basic character and active for reforming biomass-derived oxygenates, iii) spent FCC catalyst, supplied by Petronor Refinery in Muskiz, Spain, and iv) γ-Al2O3 (Alfa Aesar). The spent FCC catalyst and the γ-Al2O3 are of acid character and active for cracking reactions [39].Prior to use, the spent FCC catalyst was regenerated by calcination with air at 575 °C for 1 h in order to burn the coke deposited during its utilization in the refinery. The FCC catalyst particles were agglomerated by wet extrusion to obtain a particle size suitable for use in the fixed bed. Bentonite (50 wt%) was used as binder to confer mechanical and thermal resistance upon this catalyst. Subsequently, the extruded sample was dried overnight and calcined with air at 575 °C for 2 h. Then, the FCC catalyst was ground and sieved to a particle size in the 0.8–1.6 mm range. Similarly, olivine and γ-Al2O3 were also ground and sieved to attain the desired particle size (0.8–1.6 mm). The fraction of inert silica sand was also within this range.These conditioning catalysts were characterized by N2 adsorption–desorption, X-ray Fluorescence (XRF) spectrometry and NH3-TPD. The characterization procedure has been described in the Supplementary Information.The commercial catalyst used in the reforming reactor (ReforMax® 330 or G90-LDP), denoted as G90, was supplied by Süd Chemie. The selection of this catalyst is based mainly on its availability and reliability (without reproducibility problems in its preparation), since a significant amount of catalyst is needed in all the experimental runs. Moreover, commercial G90 and other similar commercial Ni/Al2O3 catalysts have been extensively used in the literature about the steam reforming of tar compounds [50,51], biomass and sewage sludge pyrolysis volatiles [52–56] and pyrolysis oils produced from waste plastics [57].This commercial catalyst for CH4 reforming was provided as perforated rings (19 × 16 mm), which were ground and sieved to a particle size in the 0.4–0.8 mm range. This particle size range showed a suitable performance in previous studies operating in fixed bed regime [48]. The chemical composition of G90 catalyst is based on NiO, whose nominal content is 14 wt%, apart from CaAl2O4 and Al2O3. The textural properties of the fresh catalyst determined by N2 adsorption–desorption have been shown in previous studies [58,59]. Accordingly, the catalyst is a mesoporous material, with mean pore diameter of 12.2 nm. The results of BET surface area (19.0 m2 g−1) are rather low.The Ni based catalyst reduction temperature was ascertained by temperature programmed reduction (TPR), and the results are provided elsewhere [60,61]. Accordingly, the TPR profile revealed two main peaks with the prevailing one located at 550 °C, which was ascribed to the reduction of NiO which is interacting with Al2O3 support. The peak observed at higher temperature (700 °C) was associated with NiAl2O4 spinel phase. Moreover, prior to the pyrolysis-reforming experiments, in-situ catalyst reduction was carried out by feeding a stream of 10 vol% H2 with N2 at 710 °C for 4 h.The biomass pyrolysis-reforming has been carried out in a bench scale laboratory plant, whose scheme is shown in Fig. 1 . The pyrolysis step was conducted in a conical spouted bed reactor (CSBR), whereas a fixed bed reactor was selected to perform the catalytic reforming of the volatiles from the pyrolysis step. In the latter reactor, different conditioning beds (sand, γ-Al2O3, FCC and olivine) were placed prior to the reforming catalyst bed (see Fig. 1).Previous studies conducted by the research group have proven the good performance of the CSBR for the pyrolysis of different materials, such as biomass [49,62,63], waste plastics [64,65] or tires [66–68]. Moreover, the design of the CSBR is based on previous hydrodynamic studies [69], and its dimensions are as follows: conical section height, 73 mm; cylindrical section diameter, 60.3 mm; conical section angle, 30°; bed bottom diameter, 12.5 mm; and gas inlet diameter, 7.6 mm. Continuous removal of the char particles in this reactor is carried out by means of a lateral outlet pipe located above the bed surface (Fig. 1). The gas stream is heated to the desired temperature prior to entering the reactor by means of a preheater. Both the reactor and the gas preheater are located inside a radiant oven of 1250 W.The temperature in the fixed bed reactor, which is located inside an oven (550 W), is controlled by a thermocouple located in the catalyst bed. The pilot plant is provided with a cyclone, which retains the char and sand particles entrained from the pyrolysis bed. With the aim of avoiding steam and heavy compounds condensation, both reactors (the CSBR and the fixed bed reactor), all interconnection pipes and the cyclone are placed inside a forced convection oven, wherein the box temperature is maintained at 300 °C. Avoiding the condensation of heavy compounds before and after the fixed bed reactor is essential to carry out the analysis of the products.The solid feeding device consists of a vessel equipped with a vertical shaft connected to a piston placed below the material bed. At the same time as the piston raises, the biomass feeder is vibrated, which ensures continuous discharge of the biomass into the reactor.The water required in the conical spouted bed and in the reforming step was fed by a high precision pump (Gilson 307). It was vaporized by means of an electric heater prior to entering the pyrolysis reactor. Moreover, the plant is provided with three mass flow-meters for N2, (used as fluidizing agent in the process of heating the reaction system), air and H2 (used for the reforming catalyst reduction prior to the reforming step).The product condensation system is provided with a condenser and a coalescence filter, which ensure the collection of the non-reacted steam and bio-oil compounds prior to analysis.The pyrolysis step was carried out at 500 °C, as the condensable volatile fraction (liquid fraction) obtained in the biomass pyrolysis is maximized at this temperature [49]. Based on previous hydrodynamic studies, the CSBR contained 50 g of silica sand with a particle size in the 0.3–0.35 mm range. Besides, a water flow rate of 3 mL min−1 was chosen for all the runs, which corresponds to a steam flow rate of 3.73 NL min−1. These conditions ensure a vigorous movement in the CSBR.The reforming temperature was fixed at 600 °C, which was established as the optimum one in previous biomass pyrolysis-reforming runs [26]. Thus, higher temperatures (700 °C) showed a limited effect on the reforming results [58] and may favour sintering of the metallic Ni active sites. [70].In the fixed bed reactor, the bed was divided into two sections: i) the conditioning step with the guard bed, (silica sand, γ-Al2O3, spent FCC catalyst or olivine), which is located in the upper section of the reactor, and ii) the reforming catalyst bed, which is composed of a mixture of inert sand (1–2 mm) and a commercial Ni/Al2O3 (G90) catalyst (0.4–0.8 mm). A steel mesh was used to divide both fractions with the aim of easing their separation for further characterization when each experimental run was finished.The conditioning catalysts used in this study are significantly different concerning density, i.e., sand: 2600 kg m−3; olivine: 3300 kg m−3; FCC: 1246 kg m−3, and γ-Al2O3: 1666 kg m−3. The bed mass of these materials was chosen in order to have the same bed volume in all the experiments (30 mL), which was that corresponding to a GHSVvolatiles of 3100 h−1, and the particle size of all guard catalysts was in the 0.8–1.6 mm range. The corresponding masses were 44.2 g of silica sand, 46.2 g of olivine, 17.3 g of spent FCC catalyst and 19.9 g of γ-Al2O3. Moreover, the same bed volume of the mixture of commercial Ni/Al2O3 catalyst (G90) and inert sand was used (30 mL), which corresponds to 9.4 g and 29.0 g of reforming catalyst and inert sand, respectively.In order to compare the influence of the conditioning step on the reforming one, the same operating conditions were established. Accordingly, all the runs were conducted in continuous regime by feeding 0.75 g min−1 of biomass, with the S/B ratio being 4 and the space time 15 gcat min gvolatiles −1. These conditions allow attaining conversion values close to thermodynamic equilibrium without involving high energy requirements to vaporize the water supplied. Thus, given that the steam flow rate required for attaining a suitable spouting regime in the CSBR has been set at 3.73 NL min−1 (3 mL min−1 of water), the corresponding biomass flow rate was 0.75 g min−1.Prior to the pyrolysis-reforming runs, all the elements of the plant were heated using N2 as fluidizing agent. Then, the fluidizing gas was switched to water and once temperature had been stabilized, biomass feed started.The analysis of the volatile stream was carried out at three different locations: i) after the pyrolysis step, ii) once the stream had passed the guard bed (conditioning step) and, finally, iii) at the outlet of the steam reforming reactor. Furthermore, experimental runs using different reactor configurations have been conducted under the same conditions. Thus, in the pyrolysis runs, the volatile stream is directly connected to the condensation system without passing through the fixed bed. In the conditioning runs, both reactors (pyrolyser and conditioning fixed bed) were used, with the fixed bed containing the conditioning catalyst, i.e., the reforming catalyst G90 was not introduced in the reactor. Finally, the configuration for the pyrolysis-reforming runs was made up of a CSBR and a fixed bed reactor containing both the guard and the reforming catalysts. This latter configuration corresponds to the scheme shown in Fig. 1.The mass balance in the pyrolysis runs (biomass pyrolysis and biomass pyrolysis + guard catalyst) was closed by weighting the char particles, i.e., the amount of char remained in the reactor plus those collected through the lateral pipe and retained in the cyclone and filter, and combining this information with that obtained by on-line chromatographic analysis. In these pyrolysis runs, cyclohexane was used as an internal standard to validate the mass balance closure. Thus, the pyrolysis product stream leaving the fixed bed reactor was analyzed in-line in a GC Agilent 6890 provided with a HP-Pona column and a flame ionization detector (FID) by means of a line thermostated at 280 °C, and the non-condensable gases were analyzed in a GC Varian 4900. Besides, the identification of the bio-oil compounds was conducted by condensation of the liquid sample and further analysis by means of a GC–MS spectrometer (Shimadzu 2010-QP2010S) provided with a BPX-5 column. In the pyrolysis-reforming runs, the overall and elemental mass balances (C, H and O) were closed based on the information about the volatile stream that reached the reforming catalyst (which has been determined as mentioned above) and the information obtained in the GC and microGC analyses of the stream at the outlet of the reforming step. The mass balances closure was above 95 % in all cases, and runs were repeated at least 3 times in order to ensure reproducibility. The chromatographic analyses were conducted in-line by means of a GC Agilent 6890 provided with a HP-Pona column and a flame ionization detector (FID). In order to avoid the condensation of non-converted oxygenate compounds, the sample from the reforming reactor outlet stream has been injected into the GC by means of a line thermostated at 280 °C. The permanent gases, i.e., H2, CO2, CO, CH4 and C2-C4 hydrocarbons, were analyzed in a GC Varian 4900 once the outlet stream of the reforming reactor was condensed and filtered.The textural properties of all conditioning beds were analyzed after each experimental run by N2 adsorption–desorption technique in a Micromeritics ASAP 2010 following the procedure described in the Supplementary Information.The coke formed on the spent catalysts, both guard and reforming ones, was analyzed at the end of each continuous run. The coke content was measured by Temperature Programmed Oxidation (TPO) in a Thermobalance (TGA Q5000 TA Instruments) coupled to a mass spectrometer (Thermostar Balzers Instrument). Given that the Ni active phase is oxidized at the same time as the coke combustion occurs, the CO2 formation is monitored throughout the TPO runs, according to the following procedure: i) Signal stabilization with a N2 stream at 100 °C, and, ii) oxidation with air (heating rate of 5 °C min−1 to 800 °C maintaining this temperature for 30 min to ensure complete coke combustion. Similarly, the amount of coke deposited on each guard catalyst after the pyrolysis-reforming runs was determined following the same procedure. The deactivated G90 catalysts were analyzed by Scanning Electron Microscopy (SEM) in a JEOL JSM-6400 apparatus.In order to evaluate the influence the different guard catalysts have on the subsequent reforming step, volatile conversion and individual product yields have been taken as the key reaction indices. It should be noted that the definition of these reaction indices is based on the volatile products that reach the commercial Ni/Al2O3 catalyst (G90) bed (gases and bio-oil derived compounds), i.e., once the volatile stream from the pyrolysis step had passed the guard bed. Thus, the carbon contained in the char produced in the pyrolysis step was not considered, given that this product was removed from the CSBR (through the lateral pipe) prior to the conditioning step.Accordingly, the volatile conversion in the reforming step is determined as the ratio between the C moles in the product stream leaving the reforming step (Cgas) and the C moles in the volatile stream reaching the reforming catalyst (Cvolatiles): (1) X = C gas C volatiles  · 100 Similarly, the yield of each individual product, i, has been calculated based on the pyrolysis volatile stream. (2) Y i = F i F volatiles  · 100 where Fi and Fvolatiles are the molar flow rates of each compound i and the volatile stream at the inlet of the reforming reactor, respectively.The hydrogen yield is defined based on the maximum allowable by stoichiometry: (3) Y H 2 = F H 2 F H 2 0  · 100 where FH2 is the H2 molar flow rate and F0 H2 the maximum allowable by the following stoichiometry: (4) C n H m O k + 2 n - k H 2 O → n C O 2 + ( 2 n + m / 2 - k ) H 2 Finally, H2 production is defined by mass unit of the biomass in the feed: (5) P H 2 = m H 2 m biomass 0 · 100 where mH2 and m0 biomass are the mass flow rates of the H2 produced and biomass fed into the process, respectively.The textural properties (BET surface area, pore volume and pore diameter) of the conditioning catalysts are shown in Table 2 . As observed, FCC and γ-Al2O3 are mesoporous materials, with an average pore size of around 17.0 nm, whereas inert sand and olivine are non-porous materials with negligible BET surface area and pore volume. Apart from the characteristic features of the conditioning catalysts, their physical properties may significantly influence the pyrolysis volatile composition to be fed into the reforming catalyst bed. Thus, mesoporosity would enhance the diffusion of bulky reactants, i.e., phenolic compounds, such as guaiacol and their derivatives [71]. In the case of the spent FCC catalyst, which is based on the HY zeolite, the use of bentonite as binder provides meso and macropores to the catalyst, which minimize external blockage of the channels [72]. However, the microporous structure of this zeolite is also evident, with a microporous surface area of 57 m2 g−1.The chemical composition of each conditioning catalyst was determined by XRF analysis, and the results are set out in Table 2. As observed, inert sand is mainly composed of SiO2; olivine is a mineral containing MgO, SiO2 and Fe2O3; the γ-Al2O3 used in this study contains a small amount of SiO2, apart from Al2O3; and the FCC catalyst agglomerated with bentonite (50 wt%) is a mixture containing Al2O3, SiO2, Fe2O3 and P2O5, among other metal oxides. Futhermore, incorporation of bentonite greatly influences the composition of the spent FCC catalyst (used in a previous gasification study [39]), as it leads to a significant increase in the amount of SiO2. Moreover, it has been widely reported that the chemical composition of the olivine plays a positive role in tar decomposition and reforming reactions [73,74] due to the presence of Fe0 on its surface [75]. Table 2 also shows the total acidity of the conditioning catalysts determined by NH3-TPD analysis. The results obtained revealed that only the spent FCC and the γ-Al2O3 catalyst contain acid sites, with a total acidity of 47 and 106 µmolNH3 gcat −1, respectively, whereas in the case of olivine, a negligible acidity is observed (6 µmolNH3 gcat −1). The acidity of these materials enhances cracking reactions involving bio-oil oxygenated compounds, leading to a higher amount of aromatics and paraffins.Biomass pyrolysis was conducted at 500 °C using steam as fluidizing agent. Moreover, the catalytic conditioning of fast pyrolysis volatiles was carried out at 600 °C in the fixed bed reactor (see Fig. 1). A previous study detailed the biomass pyrolysis products obtained using these low-cost conditioning catalysts, and the main mechanisms of bio-oil transformation [76].The pyrolysis products obtained were grouped into three different fractions: i) a gaseous fraction made up of CO2, CO, H2 and small amounts of C1–C4 hydrocarbons, ii) a condensable volatile fraction (liquid fraction or bio-oil) composed of water and a complex mixture of oxygenated compounds, and iii) a solid residue or char, which is the non-volatilized biomass fraction. Table 3 shows the product yields obtained once the pyrolysis stream obtained at 500 °C in the CSBR had passed the conditioning beds in the fixed bed reactor at 600 °C.As observed in Table 3, a high bio-oil yield was obtained in the biomass pyrolysis at 500 °C (75.36 wt%), which evidences the good performance of the CSBR for biomass pyrolysis due to the characteristic features of this reactor, as are high heating rates, short vapour residence times and rapid char removal from the hot reaction environment [49,77,78]. Conditioning catalysts led to a significant reduction in the bio-oil yield at the expense of gaseous product formation. In the case of inert sand, this drop is a consequence of thermal cracking reactions, whereas the decrease with the other catalytic materials is due to simultaneous thermal and catalytic cracking reactions. These results evidence that the features of the conditioning catalysts, i.e., physical properties, chemical composition and catalyst acidity (see Table 2), greatly influence the composition of the volatile stream to be fed into the reforming catalyst bed. Thus, although all conditioning catalysts tested are active for cracking (with an increase in the non-condensable gas yield in detriment of the condensable fraction (bio-oil)), the basic nature of olivine, as well as its limited porous structure (which hinders the diffusion of bulky oxygenated compounds into the bed material), led to a lower extension of cracking reactions. Besides, the mesoporous structure of FCC and Al2O3 guard catalyst, and especially the higher acidity of FCC and γ-Al2O3 catalysts, Table 2, led to higher gas yields (26.11 and 32.45 wt%, respectively) due to the promotion of bio-oil cracking.The char produced is continuously extracted from the pyrolysis reactor by a lateral outlet (see Fig. 1), and its yield was not therefore affected by conditioning catalysts. Accordingly, it remained constant in all the experimental runs (17.34 wt%).In the biomass pyrolysis at 500 °C, CO and CO2, and small amounts of H2 and C1–C4 hydrocarbons account mainly for the gaseous product stream. Besides, when inert sand, spent FCC and γ-Al2O3 were used, CO was the main compound in the non-condensable gaseous stream, which is evidence that decarbonylation reactions prevailed rather than decarboxylation ones. The higher extent of cracking reactions when spent FCC and γ-Al2O3 catalysts were used was also evidenced in the higher yield of CH4 and C2-C4 hydrocarbons in the volatile stream. However, in the case of olivine, the yield of CO2 was higher than that of CO and the other gaseous compounds due to the basic nature of this mineral, which enhanced ketonization and aldol condensation reactions leading to the formation of CO2 and water [79]. Besides, the highest H2 yield was observed in the experimental runs conducted with the olivine guard bed. Thus, the chemical composition of olivine, which contains Fe0 on its surface, promotes reforming and WGS reactions [73,75].The detailed bio-oil composition obtained in each experimental run is shown in Table 4 . The inert nature of steam as fluidizing agent in the biomass pyrolysis at 500 °C has been demonstrated in previous studies [26,80], and has been confirmed in this one by comparing the results obtained when steam was used as fluidizing agent (see Tables 3 and 4) with those reported for N2 [49]. Besides, this inert nature of steam is a great advantage for process viability, as the cost of the gases is reduced and the problems related to inert gas separation prior to the catalytic reforming step are avoided. However, the use of conditioning beds significantly modified the composition of the bio-oil products. As aforementioned, all conditioning catalysts were located in a fixed bed reactor at 600 °C (above that of pyrolysis), which may also have certain influence on the bio-oil composition.In comparison with the bio-oil obtained in the pyrolysis at 500 °C, the use of inert sand led to a significant drop in the amount of light alcohols, saccharides, and mainly in that of the phenolic fraction (particularly the catechol fraction), as a result of thermal cracking reactions. Moreover, the concentration of polycyclic aromatic alcohols increased from 0.27 to 2.47 wt%, respectively. The acidity of the spent FCC and γ-Al2O3 conditioning catalysts (Table 2) promoted deoxygenation reactions, as well as cracking, oligomerization, alkylation, isomerization, cyclization and aromatization, leading to a considerable increase in the hydrocarbon fraction (to 6.12 and 8.49 wt%, respectively). Significant differences were observed in the distribution of the components in the phenolic fraction, with catechols being the major components with the FCC conditioning bed, whereas the phenolic fraction obtained with the Al2O3 catalyst only contained alkyl-phenols. The higher selectivity of the Al2O3 catalyst for the production of alkyl-phenols revealed the effective dealkoxylation of guaiacols and cathecols [81].In the case of the olivine conditioning bed, a decrease in the acid and furan fractions was observed, with the ketone fraction remaining almost constant. Thus, the basic nature of this material promoted the ketonization of acids over basic catalysts, as well as the aldol condensation of small ketone and aldehyde molecules to larger chain ketones by carbon–carbon coupling reactions [79,82]. A reduction in the guaiacol fraction, and therefore in the overall phenolic fraction was also evidenced, whereas the yield of hydrocarbons (mainly naphthalene compounds) increased because of secondary cracking reactions.The effect of using conditioning catalysts prior to the reforming step has been analyzed, i.e., the influence the composition of the volatile stream fed into the reforming bed has on the catalyst activity and stability. Accordingly, the evolution of the volatile conversion (Fig. 2 ) and product yields (Fig. 3 ) was monitored with time on stream based on the following main reactions:Oxygenate steam reforming: (6) CnHmOk+(n-k)H2O → nCO + (n + m/2-k)H2 Water Gas Shift (WGS): (7) CO + H2O ↔ CO2 + H2 Oxygenate cracking (secondary reaction): (8) CnHmOk → oxygenates + hydrocarbons + CH4 + CO + CO2 + C Methane (and hydrocarbons) steam reforming: (9) CH4 + H2O ↔ CO + 3H2 Fig. 2 shows that the volatiles derived from biomass pyrolysis are completely reformed independently of the stream composition at the inlet of the reforming catalyst bed, which is explained by the high activity of the commercial Ni/Al2O3 (G90) catalyst.It is to note that the volatile stream reaching the reforming catalyst in any experimental run is made up of a complex mixture of oxygenate compounds. Most of the research studies dealing with the mechanisms involving reforming reactions and/or catalyst deactivation have been conducted with model compounds and synthetic mixtures simulating bio-oil and tar. However, different compounds reactivity has been reported depending on whether they are reformed alone or in a mixture of different oxygenated compounds [83]. Therefore, detailed and laborious studies are required to analyse, on the one hand, the reactivity of the compounds in the biomass pyrolysis volatile stream modified by the use of conditioning catalysts (due to the high amount of species contained and their interactions) and, on the other hand, their further contribution to catalyst deactivation.As observed in Fig. 2, the experiments conducted without any conditioning catalyst showed a stable catalyst performance for the first 50 min on stream (conversion values up to 99.8 %), and it then decreased to 56.6 % after 86 min on stream, which is evidence of the deactivation of the catalyst. The runs conducted with inert sand as guard bed showed similar results of conversion, which decreased to 58.2 % for 87 min on stream. Despite the differences obtained in the volatile composition when no conditioning catalyst and inert sand were used (see Table 4), similar catalytic performance in the reforming step was observed. It seems that, in the pyrolysis-reforming experiment without any conditioning material, the volatile stream leaving the pyrolysis reactor undergoes thermal cracking reactions in parallel with catalytic ones on the reforming catalyst, with the thermal cracking leading to a volatile composition similar as the one obtained with inert sand.Regarding the experiments conducted using olivine guard bed, almost full conversion was attained for the first 30 min on stream, and it then gradually decreased to 51.6 % for 98 min on stream. However, despite the fact that the reforming catalyst begins to lose activity earlier when there is an olivine guard bed than without any conditioning catalyst or inert sand, its deactivation rate is significantly lower, which leads to greater stability over time. This fact is explained by the different nature of the volatile stream to be reformed. The presence of more refractory compounds in the volatile stream when olivine was used as conditioning catalyst, namely the hydrocarbon fraction (1.06 wt%), led to a faster initial loss of activity. Although the hydrocarbon fraction’s reactivity is low for reforming reactions, its low concentration did not involve a fast catalyst deactivation with time on stream. Other authors have also reported a higher reactivity of oxygenated compounds derived from biomass pyrolysis in comparison with the hydrocarbon compounds due to the presence of C═O bonds that enhance the formation of carbon oxides in the reforming step [84,85]. The lower amount of phenolic compounds when olivine was used, especially the guaiacol fraction, resulted in a greater stability over time. The lower aldehyde fraction than with inert sand may also contribute to attenuating catalyst deactivation. Thus, several authors have reported that the main coke precursors, and therefore the main responsible for catalyst deactivation are aldehydes, phenols, and saccharides [83,86]. Gayubo et al. [86] state phenolic compounds (as well as aldehydes) are the main contributors to catalyst deactivation by coke formation. Valle et al. [87] analyzed the influence the presence of phenolic compounds has on catalyst stability in the reforming of raw bio-oil. Accordingly, they approached the removal of these phenolic compounds from the raw bio-oil by accelerated aging and liquid–liquid extraction methods. They observed that catalyst deactivation was lower than when raw bio-oil was used in the steam reforming, since phenols removal from the bio-oil significantly reduced coke content. Ochoa et al. [88] associated the composition of oxygenate compounds in the reaction medium with the composition of the coke formed using Fourier transform infrared (FTIR) spectroscopy. They concluded that methoxyphenols (guaicols) and levoglucosan (saccharides) have greater influence on coke formation than acids, ketones and aldehydes.The performance of the commercial Ni/Al2O3 (G90) catalyst when the spent FCC guard catalyst was used showed full conversion for the first 30 min on stream, and then sharply decreased attaining a conversion value of 38.2 % after 76 min on stream. This fast catalyst deactivation is mainly associated with the high concentration of the phenolic fraction, especially the high amount of catechol compounds (11.56 wt%). Besides, the high concentration of polycyclic aromatic alcohols (mainly composed of indenol and naphthalenol derived compounds) may also contribute to the fast catalyst deactivation in the reforming step. These results are consistent with those obtained in other literature studies. Thus, Artetxe et al. [47] investigated the steam reforming of different tar model compounds (phenol, toluene, methyl naphthalene, indene, anisole and furfural), and reported that the lowest conversion was attained when phenol was used as model compound. Trane-Restrup and Jensen [89] studied the steam reforming of cyclic model compounds in the bio-oil (2-methylfuran, furfural, and guaiacol) over Ni-based catalysts, revealing that the phenolic compound guaiacol was the most difficult to convert to synthesis gas. Likewise, Wang et al. [90] reported the lower reactivity of phenol over a Ni based catalyst compared to other oxygenated compounds, such as furfural, hydroxyacetone or acetic acid.The high concentration of hydrocarbons (6.12 wt%) in the bio-oil as a result of the acidity of the FCC catalyst also influenced the reforming reaction. This hydrocarbon fraction is mainly composed of polycyclic aromatic hydrocarbons (PAHs), namely, indene (0.12 wt%), naphthalene (2.40 wt%), fluorene (0.74 wt%), anthracene (0.65 wt%) and phenanthrene (1.89 wt%). Thus, several authors have reported the lower reactivity of large cyclic hydrocarbons with higher molecular weights than oxygenate compounds in the steam reforming reactions [47,84]. Regarding hydrocarbon reactivity in steam reforming reactions, several studies in the literature report lower reactivity of naphthalene than other bio-oil model compounds like toluene, benzene, pyrene or anthracene [41,91,92]. In fact, these aromatic hydrocarbons are well known because they undergo condensation reactions to form coke, which accelerates catalyst deactivation [93,94].The best catalytic performance was observed when γ-Al2O3 was used as conditioning catalyst, with a stable volatile conversion for the first 30 min on stream and then decreasing to 39 % subsequent to 112 min on stream. Similarly to other conditioning catalysts, the initial loss of catalyst activity occurred earlier than in the runs conducted without conditioning bed or with silica sand, which is a consequence of the total acidity of the γ-Al2O3, as it promotes secondary cracking reactions leading to a higher concentration of olefins and aromatic hydrocarbon compounds. However, although there is a high concentration of this hydrocarbon fraction (the highest one is obtained when Al2O3 is used as conditioning bed), these compounds do not involve a sharp decrease in volatile conversion. The reduction in the concentration of phenols in the volatile stream to be reformed, particularly that of guiacol and catechol fractions, significantly attenuated the Ni/Al2O3 (G90) catalyst deactivation. Total removal of acids and saccharides and a significant reduction in the aldehyde fraction may also contribute to attenuating the fast deactivation of the reforming catalyst. Several authors have reported that the main coke precursors, and therefore the main responsible for catalyst deactivation, are aldehydes, phenols, and saccharides [83,86].The great differences in the performance of the reforming catalysts when the spent FCC catalyst and γ-Al2O3 conditioning catalysts were used are mainly due to the differences observed in the composition of phenolic compounds. Thus, whereas alkyl-phenols (15.96 wt%) are only obtained with the γ-Al2O3, catechols are the major fraction with the FCC conditioning bed (11.56 wt%). A comparison of the reforming performance when olivine and Al2O3 conditioning beds are used reveals that both total concentration of the phenolic fraction and component distribution in this faction influence the performance of the reforming catalyst. Accordingly, the presence of guaicols, and especially catechols led to a fast catalyst deactivation in the reforming step.Despite the stability of G90 catalyst is greatly improved when an Al2O3 conditioning catalyst is used, the rapid deactivation of the reforming catalyst will entail working under reaction-regeneration cycles when the operation is performed at large scale. Fig. 3 shows the evolution of the yields of gaseous product and non-converted bio-oil compounds with time on stream for the experiments conducted without conditioning (Fig. 3a), and with conditioning catalysts (Fig. 3b-e). As observed, H2 and CO2 decreased with time on stream in all cases due to the lower extension of reforming, Eqs. (6) and (9), and WGS reaction, Eq. (7), as the catalyst is being deactivated. The runs conducted without conditioning catalyst and inert sand showed a similar trend in the evolution of gaseous product yields, which is evidence of the similarities in the volatile stream reaching the reforming bed catalyst in both experiments.When olivine and spent FCC catalysts were used, the evolution with time on stream followed a similar trend as conversion, with a stable H2 and CO2 yield for the first 30 min. Regarding the FCC guard bed, a sharp decrease in H2 and CO2 yields was observed due to the attenuation of the reforming and WGS reaction (Eqs. (6) and (7), respectively) as the catalyst was being deactivated. In the case of the γ-Al2O3 conditioning bed, the yields of H2 and CO2 decreased from the beginning of the reaction. Thus, H2 yield decreased from 95.9 to 28.1 % and CO2 from 93.9 % to 33.1 % for 112 min on stream.In all cases, the deactivation rate of the catalyst is faster as the concentration of the non-converted bio-oil compounds in the reaction medium is higher. In fact, an autocatalytic deactivation behaviour has been reported in the reforming of biomass pyrolysis volatiles [58,95]. These results clearly show that the oxygenated compounds, particularly the phenolic fraction (mainly catechols) produced in the pyrolysis of biomass using FCC conditioning bed, cause a much faster deactivation, which reveals that these compounds are the main coke precursors.The results obtained at zero time on stream for H2 production (based on the biomass mass unit in the feed) revealed a high performance of the overall pyrolysis-reforming strategy, with values between 9.5 and 10.2 wt%. Similar H2 production values were reported in a previous study conducted under the same experimental conditions, but using a fluidized bed reactor in the reforming step [26]. Xiao et al. [96] studied the pyrolysis-reforming of pinewood chips on a Ni/coal char catalyst, obtaining a H2 production of 10 wt% at a reforming temperature of 650 °C. Ma et al. [97] reported a H2 production value of 7.6 wt% in a three-step process (biomass pyrolysis, gas–solid simultaneous gasification and catalytic reforming) using a Ni/MgO commercial catalyst. Akubo et al. [98] investigated the pyrolysis-catalytic steam reforming of six agricultural biomass waste samples obtaining a H2 production in the 3.3–5.1 wt% range.As concerns CO yield, a slightly decreasing trend is observed in all the experimental runs, which is a consequence of a balance involving its production by the reforming reaction (decreasing with time on stream), Eq. (6), formation by cracking reactions, Eq. (8), and deactivation of the catalyst for the WGS reaction, (Eq. (7)). As the reaction proceeded, the yields of CH4 and light hydrocarbons increased slightly, which is evidence of cracking reactions, although to a minor extent (CH4 yields lower than 2 %) due to the operating conditions used in the reforming reactor, i.e., rather low temperature and residence time.With the aim of evaluating the influence of different conditioning beds on the pyrolysis volatile composition and their relationship with catalysts performance and deactivation, the cokes deposited on the guard catalysts as well as on the commercial Ni/Al2O3 (G90) catalyst have been characterized by temperature programmed oxidation (TPO). The main causes of catalyst deactivation in the reforming processes are metal sintering and coke deposition [99]. However, previous pyrolysis-reforming studies conducted by the research group using the commercial Ni/Al2O3 (G90) catalyst evidenced that metal sintering did not occur [100]. Thus, the low reforming temperature used in this study (600 °C) is slightly higher than Ni Tamman temperature (590 °C), and metal sintering is therefore avoided. Accordingly, coke deposition is the main responsible for catalyst deactivation.TPO analyses have been conducted to the guard materials in order to ascertain the influence textural properties and their characteristic features have on the coke deposited. It should be noted that the coke is formed due to the contact of the volatile stream with these materials prior to reaching the reforming catalyst. However, due to the different duration of the runs (which depends on the stability of the G90 catalyst in the reforming step), and in order to compare the amount of coke deposited on each conditioning catalyst, the average coke deposition rate per biomass mass unit has been defined as follows: (10) r - coke = W coke / t W catalyst m biomass with Wcatalyst and Wcoke being the catalyst and coke masses, respectively, mbiomass the biomass mass flow rate in the feed and t the reaction time in each run.The results of coke content and average coke deposition rate per biomass mass unit are set out in Table 5 , and the TPO profiles are shown in Fig. 4 .As observed in Table 5, significant differences are observed in the results for the conditioning catalysts used. Thus, the runs conducted with silica sand and olivine led to a negligible amount of coke deposition (0.13 and 0.38 wt%, respectively), which corresponds to average coke deposition rates per biomass mass unit of 0.02 and 0.05 mgcoke ggc -1 gbiomass -1. The limited porous structure of these materials (see Table 2) hindered coke deposition. Their characteristic features, i.e., the inert nature of silica sand, and the basic nature of olivine, as well as the capability of the latter to enhance reforming reactions, may contribute to attenuating coke deposition.As concerns γ-Al2O3 and spent FCC conditioning catalysts, the amounts of coke deposited and the average coke deposition rates per biomass mass unit were considerable higher (0.98 and 2.04 mgcoke ggc -1 gbiomass -1, for FCC and γ-Al2O3, respectively). It is well-established in the literature that the acid properties of these materials promote the formation of coke deposits due to dehydration, cracking and polymerization reactions, which take place on the acid sites [22,71,94,101,102]. Accordingly, the higher total acidity of the γ-Al2O3 compared to the FCC catalyst (See Table 2) led to a higher amount of coke deposited on this conditioning bed. The selective coke deposition on the former conditioning catalyst surface attenuates the subsequent coke formation on the reforming catalyst, and therefore improves its stability. Thus, the coke precursors are deposited on the acid sites of Al2O3 conditioning catalyst, leading to a volatile stream less prone to coke formation on the G90 reforming catalyst surface. Moreover, the mesoporous structure of these materials favors coke deposition on their surface. Similarly, Li et al. [93] analyzed the main reaction pathways occurring in the catalytic cracking of different bio-oil model compounds (acetic acid, cyclopentanone and guaicol). They reported a higher coke production when guaicol was used, and ascribed it to the coke chemical structure and hydrogen to carbon effective ratios of the feedstock. Besides, these authors describe coke formation as a sequence of polymerization and polycondensation reactions involving bulky aromatic compounds formed in the catalytic cracking of guaicol, which lead to carbon deposits on the catalyst surface.The textural properties of the fresh and spent conditioning catalysts are also shown in Table 5. As observed, the coke deposited on the conditioning materials also influenced their textural properties. In the case of inert sand and olivine, no significant differences were observed prior and subsequent to use in the pyrolysis-reforming runs due to the limited porous structure of these materials. However, a sharp decrease in the SBET area was observed for the FCC catalyst (from 81.3 to 15.1 m2 g−1) due the full blockage of the pores, especially the micropores of the HY zeolite structure. Concerning the γ-Al2O3 catalyst, the specific surface area and average pore diameter decreased from 100.6 to 83.9 m2 g−1 and from 16.9 to 12.3 nm, respectively, which is evidence of the partial blockage of the pores in this catalyst. Fig. 4 shows the TPO profiles of all conditioning catalysts, wherein significant differences are revealed concerning their coke combustion temperature. As regards the γ-Al2O3 catalyst, one main peak located at 475 °C is observed, whereas the main peak in the profile of the spent FCC catalyst is located at 535 °C. Thus, cokes of different nature are deposited, with the one deposited on the γ-Al2O3 catalyst being more hydrogenated (higher proportion of aliphatic compounds than aromatic ones), whereas that on the FCC catalyst has a more structured aromatic composition [103,104]. Table 6 shows the amounts of coke deposited (CC) and the average coke deposition rates per biomass mass unit fed (rC) on the deactivated Ni/Al2O3 catalysts when different conditioning beds are used prior to the reforming. As in the previous section, this average coke deposition rate has been determined by Eq. (10). As observed, a similar average coke deposition rate was obtained in the experiments conducted without conditioning catalyst and inert sand (0.67 and 0.66 mgcoke gcat -1 gbiomass -1, respectively), which is consistent with their similar evolution of conversion and product yields with time on stream (Figs. 2 and 3, respectively).The poor performance of the reforming process observed in Fig. 2 when the FCC catalyst was used is supported by the high average coke deposition rate obtained (0.70 mgcoke gcat -1 gbiomass -1). Thus, the composition of the volatile stream reaching the reforming catalyst is responsible for this coke formation rate, and therefore for catalyst deactivation. Accordingly, the high concentration of phenolic compounds (catechols) in the volatile stream when the FCC guard bed is used increases the coke deposition rate.The lowest average coke deposition rate was obtained when γ-Al2O3 was used as guard catalyst, which is consistent with the better performance observed in Fig. 2. The amount of coke deposited on the conditioning catalyst has an influence on the amount of coke deposited on the reforming catalyst. As aforementioned, the higher acidity of the γ-Al2O3 favored coke promoters deposition on its surface (see Table 5), and so decreased the coke formation rate on the subsequent Ni/Al2O3 (G90) catalyst used in the reforming step. Besides, the composition of the volatile stream to be reformed, with a high fraction of hydrocarbons and the phenolic one containing only alkyl-phenols, attenuates coke deposition. Thus, several authors have reported that oxygenated compounds are more prone to form carbon deposits than aromatic hydrocarbons [47,84].Regarding the experiments conducted with the olivine conditioning catalyst, a similar average coke deposition rate as in the runs without conditioning bed and with silica sand (0.67 mgcoke gcat -1 gbiomass -1) was observed on the reforming catalyst. Thus, although the conversion of pyrolysis volatiles decreased faster during the initial minutes on stream due to the presence of refractory compounds in the volatile stream, a similar coke formation rate was observed on the Ni/Al2O3 reforming catalyst.It should be noted that use of a fixed bed regime in the reforming step may lead to severe coke formation on both the conditioning and the reforming catalysts, and therefore to operational problems, such as bed plugging [60,105]. Fig. 5 shows the TPO profiles of the deactivated commercial Ni/Al2O3 catalyst when different conditioning beds were used prior to the pyrolysis-reforming process. In these profiles, two main peaks can be distinguished in all the catalyst samples, with the first one located at 435 °C (coke I), and the second one at 525 °C (coke II). The different combustion temperatures observed are closely related to the location and/or composition of the coke deposited. Accordingly, the lower combustion temperature (<475 °C) is ascribed to the coke deposited on Ni metallic sites (encapsulating coke with an amorphous nature). This coke fraction hinders the access of reactants to the active sites due to Ni particle encapsulation, and is therefore the main responsible for catalyst deactivation. Besides, this type of coke (coke I) is more hydrogenated, has a higher content of aliphatic compounds than the other one at higher temperature and is stemmed from the decomposition of oxygenates derived from biomass pyrolysis and the re-polymerization of phenolic compounds [58]. The peak located at the higher temperature (coke II) has been related to the coke deposited on the support aside from Ni active sites (and therefore, having less influence on catalyst deactivation), and is composed of highly ordered and condensed aromatic compounds [94,100,106]. Moreover, the SEM images shown for the deactivated G90 catalysts in Fig. S2 in the Supplementary Information, in which no specific morphology of the coke formed is observed, confirm the results obtained by TPO analysis.In the case the spent FCC catalyst is used as guard bed, the TPO profile of the Ni/Al2O3 catalyst reveals a higher proportion of coke I than coke II, which is evidence that it promotes the formation of encapsulating coke, thereby hindering the access of reactants to the metallic sites. Therefore, although coke content has a great influence on catalyst deactivation, coke structure and location also play an essential role on deactivation [94,107]. A minor peak appears at 660 °C in the TPO profile of the G90 catalyst used without any conditioning bed. As aforementioned, G90 catalyst is doped with Ca, and previous studies show that this high combustion temperature peak must be ascribed to the decomposition of CaCO3, which is formed by carbonation of the CaO contained in the commercial G90 catalyst [48]. Furthermore, all the conditioning catalysts used in the pyrolysis-reforming process led to a small peak at 610 °C, which arose from the thermal cracking of hydrocarbons or oxygenates in the reaction medium [94].Improvements have been carried out in an integrated reaction system for H2 production from biomass consisting of a conical spouted bed reactor for the fast pyrolysis and an in-line fixed bed reactor for the steam reforming of the oxygenate volatile stream. Thus, it has been proven that the provision of low cost conditioning catalysts (γ-Al2O3, spent FCC catalyst and olivine) prior to the reforming reactor allows tempering the volatile stream, and therefore modifying oxygenate composition, which enables the attenuation of the fast deactivation of the reforming catalyst (G90).Coke deposition is the main cause of catalyst deactivation, which leads to the blockage of the Ni active sites. Phenolic compounds in the oxygenate stream, and especially the presence of guaiacols and catechols, have a considerable influence on coke formation due to the repolymeration of these compounds on the Ni sites. This coke has been identified by TPO analysis due to its low combustion temperature.Based on the results and features of the conditioning catalysts, it has been proven that the high total acidity of γ-Al2O3 (with a high density of centers and moderate acid strength) is suitable for the selective cracking of the volatile fraction responsible for coke formation. Thus, although the initial H2 production decreases when γ-Al2O3 is used as conditioning catalyst, the stability of the reforming G90 catalyst is enhanced, and therefore a longer duration of the reaction stage is feasible.Although the deactivation of the catalyst is notably attenuated with this strategy, the deactivation by coke deposition of the conditioning catalyst is also observed. Consequently, the scalability of this two-step pyrolysis-reforming process provided with a conditioning step will require regeneration strategies for both the G90 reforming catalyst and the γ-Al2O3 conditioning catalyst. Enara Fernandez: Investigation, Visualization, Writing – review & editing. Laura Santamaria: Writing – original draft, Visualization, Writing – review & editing. Maite Artetxe: Writing – original draft, Conceptualization, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Maider Amutio: Conceptualization, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Aitor Arregi: Validation, Visualization, Writing – review & editing. Gartzen Lopez: Conceptualization, Validation, Writing – review & editing, Visualization, Supervision, Project administration. Javier Bilbao: Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Martin Olazar: Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was carried out with the financial support of the grants RTI2018-101678-B-I00, RTI2018-098283-J-I00 and PID2019-107357RB-I00 funded by MCIN/AEI/ 10.13039/501100011033 and by “ERDF A way of making Europe” and the grants IT1218-19 and KK-2020/00107 funded by the Basque Government. Moreover, this project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 823745.Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuel.2021.122910.The following are the Supplementary data to this article: Supplementary data 1

The fast deactivation of the reforming catalyst greatly conditions H2 production from biomass. In order to alleviate this problem, use of conditioning catalysts in a previous conditioning step has been proposed to modify the pyrolysis volatile stream reaching the reforming catalyst. The experimental runs have been conducted in a two-step reactor system, which includes a conical spouted bed reactor for the continuous pinewood sawdust pyrolysis and an in-line fixed bed reactor made up of two sections: the conditioning and the reforming steps. Biomass fast pyrolysis was conducted at 500 °C and the reforming step at 600 °C. Different conditioning beds (inert sand, γ-Al2O3, spent fluid catalytic cracking (FCC) catalyst and olivine) were used for the conditioning of biomass pyrolysis volatiles and the influence their composition has on the performance and deactivation of a commercial Ni/Al2O3 reforming catalyst has been analyzed. Considerable differences were noticed between the conditioning catalysts, with the reforming catalyst stability decreasing as follows depending on the type of material used: γ-Al2O3 > olivine > inert sand ≈ no guard bed > spent FCC catalyst. The high acidity of γ-Al2O3 (with a high density of weak acid centers) is suitable for the selective cracking of phenolic compounds (mainly guaiacol and catechol), which are the main precursors of the coke deposited on the Ni active sites. Although H2 production is initially lower, the reforming catalyst stability is enhanced. These results are of uttermost significance in order to step further in the scaling up of the in-line pyrolysis-reforming strategy for the direct production of H2 from biomass.

Global anthropogenic CO2 emissions were reduced about 8% (2.6 Gt of CO2) in 2020, as a result of the Covid-19 crisis [1]. This fact, although results from lockdown measures and economy slowdown, may turn into the starting point from which CO2 emissions progressively decline in the future if adequate actions are taken. According to IEA, governments have now the chance to accelerate the transition into a more resilient and cleaner energy system, while rebooting their economies and creating new jobs. Making the right investments, the economic growth can work together with a sustainable recovery plan, which might lead to air pollution emissions decrease of 5% by 2023 [2]. This plan, among other objectives, contemplates: (i) accelerating the installation of low carbon energy sources (such as renewable wind and solar PV) along with the expansion and modernisation of electricity grids; (ii) turn fuels production and utilization more sustainable; and (iii) boost innovation in crucial technology areas including hydrogen, batteries, CO2 utilisation and small modular nuclear reactors. In this context, Power to Gas (PtG) process is presented as an interesting alternative. This process targets the production of synthetic natural gas (SNG) through the catalytic conversion of renewably produced H2 and CO2 from flue gases according to the Sabatier reaction: CO2 + 4H2 ⇄ CO2 + 2 H2O (ΔH° = - 165 kJ mol−1). Thus, CO2 is used as raw material instead of being emitted as a waste and renewable energy is stored in form of a low-carbon fuel such as SNG or methane. Besides, as H2 is produced via water electrolysis in low electricity demand periods, renewable power is better exploited, which promotes its development and expansion. The produced CH4 can be easily stored or widely distributed in the current gas grid and, afterwards, can be used again for power and heat generation in private homes, mobility sector or industry [3,4].The complete hydrogenation of CO2 into methane (popularly named as CO2 methanation) is a process with considerable kinetic limitations (8 electron reduction) which can only be achieved with a suitable catalyst; commonly, supported Ni or Ru highly dispersed over a basic mesoporous support. In recent years, Ni and Ru catalysts with increasingly smaller and, a priori, more active metallic particles have been designed mostly due to advances in nanomaterials synthesis techniques, which allow increasing the surface/volume ratio and the number of active sites [5,6]. The reduction of particle size not only leads to higher metallic surface areas but also to changes in particle’s morphology, which according to its structure sensitivity could modify Turnover Frequency (TOF) numbers. It has been reported that low coordinated Ni nanoparticles contain more surface defects that act as surface hydrogen traps facilitating its dissociation and improving Ni specific activity [7]. On the contrary, other authors have reported that, in the case of Ru catalysts, low coordinated or monolayer sites induce lower CO2 methanation rates than larger nanoclusters, since they suffer from poisoning by the adsorption of stable carbonyls during reaction [8–10]. In order to obtain small particles or change their structure, diverse preparation methods have been alternatively employed, such as Incipient Wetness Impregnation (IWI) [11], one-pot Evaporation-Induced Self-Assembly (EISA) [12], Microwave-Assisted (MA) [13], Deposition-Precipitation (DP) [14], Co-Precipitation (CP) [15] and polyol method [16] or equivalent Glycerol Assisted Impregnation (GAI) [17].The main disadvantage of Ni catalysts with respect to those of Ru is their considerably lower activity at low temperature due to their inferior H2 dissociation capacity [18]. Instead, the main drawback of Ru catalysts is their exorbitant price. Nevertheless, designing appropriate Ni-Ru bimetallic systems could be a solution to balance those handicaps. Generally, these bimetallic catalysts are known to exhibit better catalytic properties compared to their monometallic counterparts such as higher conversion, fewer side reactions (selectivity) and more stability due to a synergistic effect [19–21]. This synergy happens as a result of specific electronic interactions and geometric positional relationships between the two metals (combination of “ligand” and “ensemble” effects) [20]. By adding an appropriate secondary metal (Ru) to the catalytic formulation, the electronic properties of the main metal (Ni) are usually altered leading to changes in reagents adsorption and reaction intermediates formation. These changes, in turn, can modify the reaction pathway and the activation energy so that the activity of the catalyst is improved [21]. Recently, Ni-Ru bimetallic systems have proven to be very effective specifically for CO methanation [22–25]. Liu et al. [26] also reported enhanced catalytic activity for CO2 methanation over 10Ni-1Ru-2CaO/Al2O3 formulation due to a significant increase in H2 and CO2 chemisorption capacities, whereas Wei et al. [27] did not achieve activity improvement by adding Ru to Ni-zeolite (13X and 5A) catalysts. However, the analysis of Ni-Ru systems in terms of physicochemical and catalytic properties for CO2 methanation has not been the focus of many systematic studies so far.Regarding CO2 methanation reaction mechanism, several studies have been carried out by means of Operando FTIR or DRIFTS with the aim of determining the reaction intermediates and elementary reaction steps over supported Ni [20,28–30] and Ru [9,10,16,31] catalysts. Although there is still controversy regarding the identification and the role or place of some adsorbed species in the reaction pathway, two widely accepted routes have been proposed: the so-called dissociative and associative mechanisms [32]. The former assumes the dissociative adsorption of CO2 into adsorbed CO or carbonyl followed by its hydrogenation into CH4. In the latter, by contrast, CO2 is molecularly adsorbed in form of carbonates or bicarbonates, which are progressively reduced by H spillover into formate, methoxy species and, finally, methane [33]. Noteworthy, CO2 methanation mechanism over bimetallic catalysts has scarcely been studied.The main goal of this work has been to sequentially improve the low temperature activity of the conventional and industrial Ni/Al2O3 formulation by modifying the preparation method and incorporating a secondary metal, such as Ru. Additionally, this work has aimed to identify which factors are responsible for such improvement. Firstly, the influence of Glycerol Assisted Impregnation (GAI) method on the dispersion and structural characteristics of Al2O3-supported Ni and Ru particles was examined. These materials were catalytically compared with equivalent ones prepared by the conventional Incipient Wetness Impregnation method (IWI). After that, the effect of Ru incorporation on the physicochemical properties and catalytic performance of Ni/Al2O3 formulation was studied. To our knowledge, we pioneer operando FTIR analysis of CO2 methanation reaction on Ni-Ru bimetallic system, identifying the type and evolution of reaction intermediates and determining the roles of both Ni and Ru in the reaction pathway.For this work, a series of alumina-supported Ni and Ru catalysts as well as bimetallic Ni-Ru/Al2O3 samples were prepared. The two pairs of monometallic Ni and Ru catalysts were obtained by two synthesis procedures that consisted of the following steps: impregnation and calcination. According to the first procedure, the metal (Ni or Ru) nitrate solution was incorporated into Al2O3 support by Incipient Wetness Impregnation (IWI) and the resulting catalyst precursor was calcined in a muffle under air (uncontrolled atmosphere). IWI method as well as calcination procedure was the same as followed and explained in detail in our previous work [34]. However, in the second synthesis route, the metal solution is introduced by Glycerol Assisted Impregnation (GAI) method and the precursor is calcined under a controlled atmosphere. The GAI method, which was developed by Gudyka et al. [17], also consists of the typical dry impregnation but employs a glycerol/water solution as solvent instead of bare H2O. In our case, a 30 wt% C3H8O3/water solution was used. After impregnation, samples were dried overnight and calcined ex situ in a tubular reactor under 50 mL min−1 of 20 %H2/N2 (controlled atmosphere) at 550 °C for 2 h (with 10 °C min-1 heating rate). In both cases, the required amounts of Ni(NO3)2·6H2O (Sigma Aldrich, 99.99 %) and Ru(NO)(NO3)3 (Sigma Aldrich, Ru = 1.5 % w/v) precursors were employed in order to attain 12 wt% Ni and 3 wt% Ru nominal metal contents and the calcination temperature was chosen according to thermogravimetric results of catalysts precursors. These four catalysts were labelled according to their composition and preparation method as follows: NiAlIWI, RuAlIWI, NiAlGAI and RuAlGAI.On the other hand, once results of monometallic catalysts were analysed, three additional bimetallic catalysts were prepared by GAI method varying the Ru content from 0.5 to 1.5 wt%. In all cases, the nominal Ni content was set at 12 wt% and small amounts of Ru were incorporated by co-impregnation. After that, samples were also dried overnight at 120 °C and calcined under the same conditions described above. These samples were named Ni-xRuAl, where variable x represents the Ru content (0.5, 1.0 or 1.5 wt%).In order to determine the suitable calcination temperatures, thermogravimetric analysis was carried out in a Setaram Setsys Evolution apparatus connected in series with a Pfeiffer Prisma mass spectrometer (TGA-MS). In all cases, around 100 mg of catalyst precursor was placed in a 30 μL Al2O3 crucible and was firstly dried in situ at 125 °C. After that, the temperature was increased from 125 to 625 °C with 5 °C min−1 heating rate and continuously recorded along with mass loss. The catalysts precursors prepared by IWI method were calcined under 50 mL min−1 oxidative stream (5% O2/He), whereas the ones prepared by GAI were analysed under reductive atmosphere (5% H2/Ar). The exit gas stream composition was analysed by mass spectrometry following the 16 (CH4), 17 (NH3), 18 (H2O), 28 (CO/N2), 30 (NO), 44(CO2/N2O) and 46 (NO2) mass signals.The textural properties and crystalline phases of the supported catalysts were determined by N2 physisorption and X Ray Diffraction (XRD). The protocols for these analyses are detailed in the former work [34].The micrographs of the monometallic catalysts were obtained by a TECNAI G2 20 TWIN microscope which operates at 200 kV and is equipped with a LaB6 filament, EDAX-EDS microanalysis system and Transmission Electron Microscopy (TEM). The micrographs together with elemental maps of the bimetallic catalysts, instead, were taken by a FEI Titan Cubed G2 60–300 microscope with much higher resolution. This microscope is equipped with a high-brightness X-FEG Schottky field emission electron gun, monochromator, CEOS Gmbh spherical aberration corrector and Super-X EDX system with High-Angle Annular Dark-Field (HAADF) detector for Z contrast imaging in Scanning Transmission Electron Microscopy (STEM) configuration. All powder samples were mixed with ethanol solvent and kept in and ultrasonic bath for 15 min in order to attain a good suspension. After that, a drop of suspension was spread onto a TEM copper grid (300 mesh) covered by a holey carbon film for each sample. Finally, the grids were dried under vacuum to remove the solvent. The particle size distribution of monometallic catalysts was determined by measuring the diameter (d) of at least 200 particles. After that, the mean metal dispersion (DMe ) was estimated applying the d-FE model [35] as follows: (1) D Me ( % ) = 5.01 d a t ∑ j n j d j 2 + 2.64 d a t 0.81 ∑ k n k d k 2.19 ∑ i n i d i 2 × 100 where, di, dj and dk are the diameters of the “i”, “j” and “k” particles, ni is the number of particles with diameter di , nj is the number of particles with diameter dj (dj > 24·dat.), nk is the number of particles with diameter dk (dk ≤ 24·dat) and dat. is the atomic diameter of Ni or Ru.The resistance against oxidation of catalysts prepared by GAI method was determined by three consecutive RedOx cycles in a Micromeritics AutoChem 2920 apparatus. Previously, the samples were exposed to a 50 mL min−1 stream of 5%H2/Ar in order to reduce the passivated nickel layer. For each RedOx cycle, 15 oxidative pulses (5 cm3 of 5%O2/He) were injected followed by another 15 reductive pulses (5 cm3 of 5%H2/Ar). Note that between pulse injections an inert gas stream of He or Ar was continuously fed depending on the step (oxidative or reductive, respectively). The resistance to oxidation of NiAlGAI and Ni-1.0RuAl catalysts was measured at 325 °C, while that of RuAlGAI catalyst at 550 °C. The temperatures were chosen considering that Ni and Ru are oxidized at around 300 °C and 500 °C, respectively [19]. The resistance against oxidation, defined as the cycle reversibility, was calculated by the following expression: (2) Reversibility of cycl e i ( % ) = n i ( N i 0 ) n i ( N i 2+ ) + n i − 1 rem . ( N i 2+ ) × 100 where, ni (Ni2+) are the moles of Ni oxidized in cycle i, n i − 1 rem . N i 2+ are the moles of Ni that remain oxidized from previous cycles and ni (Ni0) are the moles of Ni reduced or recovered in cycle i. Note that ni (Ni2+) as well as ni (Ni0) were calculated from total O2 and H2 uptakes of oxidative and reductive steps, respectively.Hydrogen Temperature Programmed Desorption (H2-TPD) experiments were also performed on a Micromeritics AutoChem 2920 apparatus. These experiments allowed us determining the hydrogen chemisorption capacity as well as chemisorption strength of monometallic and bimetallic catalysts. In a first step, the metal surface of samples was reduced and cleaned up by 5%H2/Ar gas stream at 500 °C for 30 min and then cooled down to 50 °C. After that, a 50 mL min−1 stream of pure hydrogen was fed long enough for complete adsorption or saturation (around 1 h). Subsequently, catalysts were flushed out with Ar for 30 min in order to remove physisorbed H2. Finally, the desorption was conducted increasing temperature up to 850 °C at 10 °C min−1 heating rate. From integration of TPD profiles at T < 450 °C, the metallic surface area (S Ni) of NiAlGAI catalyst was estimated according to this equation: (3) S Ni m 2 g -1 = N A V m × V des . × S F × a t A Ni where, Na is the Avogadro number, V m is the molar volume in cm3 mol−1, Vdes. is the volume in cm3 of desorbed H2 per gram of catalyst, SF is the stoichiometric factor and atA Ni is the effective atomic area of Ni. In this work, a SF (Ni/H2) of 2 and atA Ni of 6.49 × 10-20 m2 atom−1 were assumed.CO2 methanation reaction was performed in a downstream fixed bed reactor (ID =9 mm). In all cases, the stainless-steel reactor was loaded with 0.5 g of catalyst particles (dp = 300−500 μm), which were diluted to 50 % (v/v) with quartz particles in order to avoid hot spots. The Ni and Ru catalysts prepared by IWI were firstly reduced at 500 and 400 °C for 1 h with 20 % H2/He, respectively. The samples prepared by GAI were also reduced but at 250 °C in order to remove the passivated oxide layer formed by having been in contact with air. After cooling down the samples to 200 °C with He (inert gas), the temperature was raised up to 400 °C in steps of 25 °C under reactant stream. This gaseous mixture was composed of 16 % CO2 and 64 % H2 (H2/CO2 = 4), balanced up to 100 % with He (total flow of 250 cm3 min−1). The outlet gas stream composition was analysed by GC (Agilent 7890B) once steady state was reached at each temperature. H2, He, CH4 and CO concentrations were monitored by MolSieve type columns, while that of CO2 by HayeSep type column. The produced water was retained by a Peltier cooling module upstream of the gas chromatograph to avoid molecular sieve column degradation. All reactions were carried out at atmospheric pressure and at WHSV of 30,000 mL h−1 gcat −1.The catalytic performance was evaluated by CO2 conversion ( X C O 2 ), CH4/CO products selectivity ( S C H 4 or S CO ) and yield ( Y C H 4 and Y CO ) which were calculated from reactor inlet and outlet molar flows according to the following equations: (4) X C O 2 % = F C O 2 in − F C O 2 out F C O 2 in × 100 (5) S C H 4 % = F C H 4 out F C O 2 in − F C O 2 out × 100 (6) S CO % = F CO out F C O 2 in - F C O 2 out × 100 (7) Y C H 4 % = X C O 2 × S C H 4 100 = F C H 4 out F C O 2 in × 100 (8) Y CO % = X C O 2 × S CO 100 = F CO out F C O 2 in × 100 where Fi is the inlet or outlet molar flow of component “i” in mol s−1.Finally, the Turnover Frequency (TOF) numbers, which indicate the number of CO2 molecules converted per second and per active site, were calculated as follows: (9) T O F M e s -1 = − r C O 2 mol C O 2 g cat . -1 s -1 S M e mol Me g cat -1 = F C O 2 in × X C O 2 × M W M e W × D M e × F M e where MW is the mass weight of the metal in g mol−1, W is the catalyst weight in g, DMe is the metallic dispersion and FMe is the mass fraction of metal in the catalyst. Operando FTIR spectra were collected using an IR cell from In-Situ Research Instruments, coupled to a Nicolet 6700 spectrometer equipped with a MCT detector and using a spectra resolution of 4 cm−1. Powdered samples were pressed at 1.5 tons into 10 mg cm-2 wafers which, prior to the experiments, were in situ activated/reduced at 500 °C for 1 h under a 5% H2/Ar flow of 20 mL min−1. After pretreatment, wafers were cooled down under Ar flow to 150 °C, being background spectra collected every 25 °C. CO2 adsorption tests were carried out by exposing samples to a 20 mL min−1 stream of 5% CO2/Ar, whereas in CO2 methanation experiments a 5% CO2: 20 % H2: 75 % Ar gas mixture was used. In both cases, experiments were carried out in two steps. Firstly, the used gas mixture was stabilised during 30 min and a series of spectra were collected at 0, 1, 3, 5, 10, 15 and 30 min. Secondly, temperature programmed adsorption (CO2/Ar flow) or temperature programmed surface reaction (TPSR, CO2/H2/Ar flow) was run from 150 to 450 °C using a heating rate of 2 °C min−1. Note that the depicted spectra were obtained by subtraction of those recorded under reaction/adsorption conditions every 25 °C and those corresponding to backgrounds.To determine how catalysts precursors are decomposed and the temperature required for their complete calcination, thermogravimetric analysis (TGA) was carried out (Fig. 1 ). Additionally, the gaseous products from precursors calcination were analysed by a mass spectrometer connected at the exit of the thermobalance (Figs. S1 and S2, supplementary material). Fig. 1a and b show both TG and dTG profiles of supported Ni catalysts precursors calcined under oxidative (5% O2/He, IWI catalyst) and reductive (5% H2/Ar, GAI catalyst) atmospheres, respectively. In general, the mass loss takes place in different consecutive steps that can be identified by the dTG profiles. In the case of NiAlIWI precursor calcined under O2/He (Fig. 1a), the dTG profile presents a main mass loss rate peak at 265 °C and two shoulders at 200 and 350 °C. The first shoulder can be attributed to structural water desorption from Al2O3 or water released during dehydration steps of nickel precursor (Ni(NO3)2·6H2O), whereas the broad peak and the second shoulder are due to nitrate decomposition/oxidation into NOx (NO and NO2), as confirmed by MS signals (Fig. S1b). Mass loss is observed up to 475 °C approximately, suggesting that a calcination temperature of 500 °C is enough for the complete precursor decomposition into NiO/Al2O3.The TG profile of the NiAlGAI precursor (Fig. 1b) is somewhat different due to the presence of an organic compound which could be a metal alkoxide from coordination nickel cations (Ni2+) with glycerol solution [5,6,17]. In this case, the dTG profile shows 4 differentiated negative peaks among 125 and 400 °C. In agreement with MS spectra (Fig. S1d), the first one at 170 °C could be attributed to NO3 − reduction into NO and the next two, centered at 290 and 325 °C, to the reduction of the organic template. It can be suggested that the glycerolate is decomposed into smaller molecules (such as ethylene glycol and ethanol) and surface carbon by hydrogenolysis reactions. In fact, the last mass loss rate peak centered at 360 °C matches with the appearance of methane (m/z = 15) in the product stream, suggesting that the remaining surface carbon is being reduced. In this case and according to the TG profile, a calcination temperature of 550 °C is needed to complete NiAlGAI precursor reduction.Regarding Ru/Al2O3 precursors, the TGA profiles of their respective calcinations are shown in Fig. 1c and d. By comparing those figures with the above described, it can be observed that the mass loss profile of RuAlIWI precursor (Fig. 1c) is similar to that of NiAlIWI (Fig. 1a). In fact, the same calcination steps are identified and confirmed by MS spectra (Fig. S2b): a first peak at 205 °C due to water release followed by a more intense negative peak together with a shoulder at 335 °C, which are attributed to nitrate and nitrosyl groups oxidation into NOx. Although the precursor is completely removed at 450 °C, a calcination temperature somewhat lower (400 °C) was employed in order to avoid excessive growing of RuO2 crystallites [34]. Finally, TGA profiles of RuAlGAI precursor calcined under 5%H2/Ar are shown in Fig. 1d. Note that the dTG profile presents a broad band which could be divided into 3 negative peaks at 185, 255 and 320 °C, which correspond to several calcination steps. According to MS spectra (Fig. S2d), nitrate groups and organic compounds are partially reduced and water is released as product in a first step (negative peak at 190 °C). In a second step (from 250 to 350 °C), the organic compound continues being reduced and carbon monoxide (m/z = 28) is observed in the products stream. Additionally, a small broad peak can be appreciated at around 540 °C. This peak matches with methane appearance from hydrogenation of remaining surface carbon. In this case, a temperature of 550 °C was used for precursor calcination. It must be highlighted that in all cases the observed total mass loss is similar to that expected for complete calcination of catalyst precursors: 19.9 vs. 17.2 % for NiAlIWI, 23.5 vs. 27.1 % for NiAlGAI, 10.9 vs. 9.9 % for RuAlIWI and 18.9 vs. 19.6 % for RuAlGAI.Once catalysts precursors were calcined according to TGA results, the resulting catalysts were characterized by several techniques. Some physicochemical properties are shown in Table 1 . It should be noted that the metal content of all catalysts is close to the nominal, indicating that Ni and Ru were successfully incorporated by the two methods (IWI and GAI). In addition, the high specific surface area and pore volume of all catalysts indicate that the textural properties of starting γ-Al2O3 (SBET = 214 m2 g−1 and Vpore = 0.563 cm3 g−1) were not considerably affected by the different impregnation and calcination processes. As expected, supported Ni catalysts presented lower SBET and Vpore than RuAl ones, mainly due to their higher metal content. On the other hand, the catalysts prepared by GAI method exhibited slightly lower values of such textural properties than those prepared by IWI, probably due to the higher calcination temperature.In regard to XRD analysis of reduced catalysts (not shown), both elemental Ni (XRD peaks at 2θ = 44.5, 51.8 and 76.4 ˚) and Ru (XRD peaks at 2θ = 38.4, 42.2 and 44 ˚) were clearly identified on NiAlGAI and RuAlIWI samples, respectively. However, broad and low-intensity peaks of Ni0 and no peaks of Ru0 were detected in NiAlIWI and RuAlGAI XRD patterns, suggesting that the crystalline phases are better dispersed than on NiAlGAI and RuAlIWI catalysts. This fact was confirmed by crystallite size calculation according to Scherrer equation (τ, Table 1).The effect of the preparation method on the morphology as well as on the particle size distribution was determined by TEM. In addition, the mean metal dispersion and metal surface area (Table 1) were calculated by d-FE model [35]. The micrographs of Ni catalysts along with their corresponding particle size distribution histograms are displayed in Fig. 2 . In both cases, quasi-spherical supported Ni particles (circled in yellow) were observed. It can be appreciated that the particle size distribution (calculated from measurement of at least 200 particles) is wider in the case of the sample prepared by GAI method. In fact, NiAlIWI catalyst presents Ni particles sizes from 2 to 10 nm, whereas the distribution of NiAlGAI sample ranges from 3 to 19 nm. In this line, the average particle sizes are 5.8 nm (D Ni = 19.8 %) and 11.2 nm (D Ni = 11.5 %) for NiAlIWI and NiAlGAI catalysts, respectively. Note that these values are in agreement with crystallite sizes estimated by XRD, indicating that the active phase is better dispersed on NiAlIWI catalyst. However, this catalyst presents a reduction degree of 38 % at 500 °C, i.e., less than the half of total nickel is reduced before the reaction, as determined in our previous work [34]. For that reason, the Ni reactive surface area is slightly higher for the catalyst prepared via GAI method (see Table 1).Such differences in dispersion and amount of reducible nickel are related with the calcination step. In the case of NiAlIWI catalyst, the precursor is calcined in air favouring mainly the formation of NiO highly interacting with the support or even NiAl2O4 inert phase. After reduction treatment at 500 °C, small and well distributed Ni particles are obtained but not all nickel is reduced due to the high metal-support interaction observed by H2-TPR. This high interaction between NiO and Al2O3, which was extensively studied in the literature [36,37], is also confirmed by examination and comparison of several TEM micrographs: far fewer Ni particles are visualized on NiAlIWI than on NiAlGAI catalyst, indicating a lower Ni reduction extent. On the other hand, the NiAlGAI precursor is calcined under reductive atmosphere (20 % H2/N2), avoiding the formation of Ni2+ species able to react with γ-Al2O3 and assuring that all nickel will be reduced after the preparation procedure. Besides, the presence of non-volatile organic compounds apparently prevents Ni crystals from excessive growing. As the temperature increases during the calcination, it seems that incipient nickel nanocrystals are embedded in an organic matrix that acts as a barrier preventing them from sintering [17]. As a result, all Ni is reduced and quite well dispersed in form of 11 nm size particles. Noteworthy, Ding et al. [38] observed a similar Ni particle size distribution for a Ni/SiO2 prepared by the glycerol assisted impregnation and reported that glycerol resulted to be the best alkanol solvent among those studied.Analogously, Fig. 3 shows TEM micrographs together with particle size histograms of monometallic Ru/Al2O3 samples. In both catalysts, Ru particles with different morphology were easily visualized (within yellow circles or rectangles). Ruthenium was homogeneously dispersed in form of spherical particles on RuAlGAI while a much more heterogeneous distribution was verified on RuAlIWI. The latter presents both oval and hexagonal Ru particles or even aggregates formed by several particles. In this case, the particle size distribution seems to be quite affected by the preparation method. On one side, RuAlIWI catalyst has an unimodal particle size distribution with a long tail ranging from 4 to 32 nm and a corresponding average particle size of 14.8 nm (D Ru = 7.2 %). On the contrary, the particle size distribution of RuAlGAI sample, shown in Fig. 3b, is symmetric and much narrower. It should be noted that this catalyst presents an average particle size of 2.7 nm, which correspond to a dispersion of 34.4 %. These results clearly indicate that GAI is a more appropriate method to disperse Ru over Al2O3.In our former studies based on thermo-XRD results, we reported that RuO2 crystals tend to grow fast and agglomerate under oxidative calcination conditions due to the formation of volatile RuOx [34]. That would explain why bigger particles and so long tail are observed in the histogram of the catalyst prepared by IWI method. This fast growth is clearly avoided by GAI method, which includes a non-oxidative calcination. Furthermore, even more uniform and smaller particles are created due to the organic enclosing effect above explained. Yan et al. [10] obtained similar metallic dispersion (DRu = 32.2 %) in a 3%Ru/Al2O3 prepared by incipient wetness impregnation of Ru(III) acetylacetonate precursor and performing the calcination treatment under 10 %H2/Ar flow. As a result, the RuAlIWI catalyst contains a Ru surface area of 0.79 m2 g−1 while that of RuAlGAI is 3.90 m2 g−1.In a final step, the catalytic performance of the catalysts was evaluated in order to determine the effect of the preparation method on activity. Fig. 4 shows the CO2 conversion (above) along with product selectivity (below) as a function of the reaction temperature for Ni/Al2O3 and Ru/Al2O3 catalysts, respectively. As previously reported [18], Ru-based catalysts were more active than Ni-based ones due to the higher ability of the former to dissociate hydrogen at lower temperature. Thus, the catalytic activity order is as follows: RuAlGAI > RuAl IWI > NiAlGAI > NiAlIWI. The activity profiles of Ni/Al2O3 samples are not so different, as shown in Fig. 4a. In both cases, the CO2 conversion (reaction rate) increases exponentially with temperature from 225 °C (onset reaction temperature) to 325 °C and then, this increase slows down as the reagents are depleted and equilibrium conversion is approached. It must be noted that the CO2 conversion is slightly higher for NiAlGAI catalyst in the studied temperature range, resulting in a T 50 (temperature at which 50 % of CO2 conversion is obtained) only 5 °C lower. However, a more significant difference can be observed in selectivity (Fig. 4b): NiAlIWI produces more CO than NiAlGAI catalyst at mild temperatures (T ≈ 300 °C), although never more than 3.5 % of converted CO2. In fact, the CO selectivity of NiAlIWI at 300 °C is around 2.5 times higher than that of NiAlGAI catalyst (3.2 vs. 1.3 %). The small amount of carbon monoxide is produced either from reverse water gas shift (RWGS) or reforming reactions.On the other hand, the higher CO2 conversion and CH4 selectivity observed for NiAlGAI catalyst are probably related to a higher Ni surface area (8.25 vs. 5.62 m2 g−1). This hypothesis was supported by calculations of TOFs at 250 °C. Note that by definition, TOF assumes that reaction takes place at any point of metal surface. However, under CO2 methanation conditions, the partial H2 pressure is at least four times higher than that of CO2, which disfavors the adsorption of the latter. Consequently, metal particles will be largely covered by H2. Also, considering that the support (γ-Al2O3 in this study) is able to adsorb or active CO2, it can be assumed that CO2 methanation takes place at the perimeter of metal-support interface rather than on surface, as reported in a previous work [39]. Therefore, for more realistic comparison, TOF was normalized with respect to interfacial length or perimeter (TOF/I 0, Table 1). The total metal-support perimeter per metal surface area (I0 ) was calculated by Eq. (10), which was proposed by Kourtelesis et al. [40] and is based on developments reported by Duprez et al. [41]. (10) I 0 m interface / m Me 2 = S Me 2 × β × ρ Me × A W Me N A × a t A Me where, S Me is the metallic surface area in m2 gMe −1, β is a particle shape factor (33.3 for hemispherical particles), ρ Me is the density of the metal in g m-3, AW Me is the metal atomic weight of the metal in g mol−1, NA is the Avogadro number, and atA Me is the area occupied by a single metal atom (6.49·10-20 m2 Ni atom-1 and 6.13·10-20 m2 Ru atom-1). It can be observed that TOF/I0 values are of the same order of magnitude, suggesting that the CO2 methanation rate per metal atom at the interface for supported catalysts with average Ni particle perimeters of 18.2 nm (NiAlIWI catalyst) and 35.2 nm (NiAlGAI catalyst) is quite similar. Recently, the structure sensitivity of CO2 methanation over supported metals has been studied by various authors. For instance, Vogt at al. [29] clearly reported structure sensitive CO2 methanation over Ni/SiO2 catalysts with small particle sizes ranging from 1 to 6 nm, concluding that the more active Ni atoms are those forming clusters of 2−3 nm. The high TOF of these clusters was attributed to an intermediate adsorption strength of CO on Ni, which was reported to be a reaction intermediate of CO2 methanation on Ni/SiO2 catalyst. However, Beierlein et al. [14] demonstrated that the specific activity does not depend on Ni particle size within a range from 6 to 91 nm, observing a linear correlation between the activity and Ni surface area and concluding that CO2 methanation on highly loaded Ni/A2O3 catalysts is a structure insensitive reaction. Therefore, it seems that structure sensitivity clearly depends on the range of Ni particle size studied as well as the metal-support combination used. In our case, the results are in agreement with the findings of the second authors, since the observed specific activity barely increase when decreasing particle size from 11 to 6 nm.Analogously, the light-off and selectivity curves of Ru/Al2O3 catalysts are displayed in Fig. 4c and d. In this case, the onset temperature for both samples is 200 °C and the equilibrium CO2 conversion is reached at the same temperature ( X C O 2 at 400 °C = 82 %). Nevertheless, the increase in CO2 conversion with temperature for RuAlGAI is faster than for RuAlIWI catalyst, which leads to a notable 20 °C left shift of the light-off curve (i.e., superior activity at low temperature). Regarding the selectivity towards CH4, it was higher than 99.5 % in the range of studied temperatures and only trace amounts of CO in terms of ppm were detected for RuAlGAI catalyst (Fig. 4d). Considering that metal particles of RuAlGAI are five times smaller than that of RuAlIWI catalyst, one could expect a bigger difference in catalytic performance. This suggests that metal-support interface of the former is less active, as revealed by TOF/I0 values also summarized in Table 1. Note that TOF/I0 value is around one order of magnitude lower for RuAlGAI catalyst, suggesting that CO2 methanation is structure sensitive on RuAl catalysts. Indeed, a lower specific methanation activity on small Ru particles or clusters had already been reported by several authors [8–10]. According to them, CO formation via r-WGS is favoured rather than CO2 methanation on atomically dispersed or low coordinated small Ru particles. Despite that fact, a considerable T 50 value gradient of 20 °C is observed, which evidences that a more active catalyst is obtained by GAI method.Monometallic Ni and Ru catalysts prepared by Glycerol Assisted Impregnation (GAI) achieved better methanation activity compared to those prepared by incipient wet impregnation. In a second step, bimetallic Ni-based catalysts with small Ru contents (< 2 wt%) were prepared following the GAI coimpregnation procedure. The physicochemical properties of NiAlGAI and bimetallic catalysts (Ni-0.5RuAl, Ni-1.0RuAl and Ni1.5RuAl) are shown in Table 2 .As observed, the metal contents determined by ICP are very close to the nominal values, indicating that no relevant amount of metal was lost during the synthesis. Interestingly, the specific surface area slightly increases with Ru content: 5, 9 and 10 %, respectively. This unexpected trend can be explained by analysing the pore size distribution of the catalysts (Fig. S3, supplementary material). The monometallic catalyst (NiAlGAI) presents a narrow unimodal mesopore size distribution centered at 7.3 nm, whereas bimetallic catalysts exhibit wider and bimodal distributions with maxima between 6 and 10 nm. As already discussed, NiAlGAI catalyst presents similar particles with sizes probably above 7 nm, which partially or completely block the mesopores of the support. However, the bimodal distribution verified for bimetallic catalysts might be due to the presence of particles with well differentiated size or morphology, which might penetrate into the small pores of γ-Al2O3. Ru incorporation widens the distribution but decreases its intensity, which finally results in a slight increase of SBET from 168 to 175 m2 g−1 and similar pore volume of 0.42 cm3 g−1. Thus, introduction of Ru makes some improvement in textural properties of Ni/Al2O3 catalyst.XRD analysis was also performed for bimetallic catalysts (not shown). However, no characteristic peaks of both metals were detected (crystallites sizes < 5 nm). This observation is in agreement with N2 physisorption results and indicates that Ni and Ru are finely dispersed.Concerning catalysts’ resistance against oxidation, NiAlGAI and Ni-1.0RuAl samples were exposed to three consecutive RedOx cycles at 325 °C. Each RedOx cycle consisted of feeding 15 oxidative pulses (5 cm3 of 5%O2/He) followed by another 15 reductive pulses (5 cm3 of 5%H2/Ar). On that way, the effect of O2, fed in a concentration similar to that typically presented in flue gases, on the catalysts was estimated and their reversibility was determined. Note that the resistance to oxidation of 3RuAlGAI catalyst was measured at 550 °C in order to ensure that its oxidation was effective. The reversibility values of NiAlGAI and Ni-1.0RuAl catalysts, defined as the percentage of Ni reduced per cycle after sample being exposed to 15 oxidative pulses (Eq. 2), are shown in Fig. 5 .It can be clearly observed that the reversibility values of the bimetallic catalyst are superior to those of monometallic one in all cycles, observing the major difference in the third cycle: 60 vs. 42 %, respectively. This indicates that incorporation of Ru provides higher resistance to oxidation and/or higher capacity to recover the reductive state than the monometallic NiAlGAI. The observed higher reversibility is due to ruthenium role as promotor of nickel reduction in the bimetallic catalyst, i.e., H2 is firstly dissociated on Ru surface and then can migrate to neighbouring NiO particles facilitating their reduction [23]. Furthermore, it can be noticed that the reversibility of monometallic catalysts decreases from 52 % (cycle 1) to 42 % (cycle 3), while that of bimetallic catalyst remain stable around 60 %. Although the decrease from second to third cycle is not so pronounced (- 2%), it seems that the reversibility value of NiAlGAI sample could keep decreasing in further consecutive cycles due to a progressive formation of NiO that is no longer able to be reduced by remaining Ni°. The fact that reversibility of bimetallic system is apparently stable, suggests that Ni particles are near to and surrounded by Ru ones, which avoids or at least slows down the formation of non-reversible NiO particles. However, as reported by Rynkowski et al. [19], the presence of Ru does not prevent the formation of spinel type oxides at long term and high temperatures. Finally, it must be highlighted that the H2 uptake was around 2 times the O2 uptake at 550 °C for RuAlGAI catalyst, suggesting, as expected, 100 % reversibility.On the other hand, the hydrogen adsorption capacity was determined by TPD. Thus, H2-TPD profiles of the samples are depicted in Fig. 6 .It can be observed that all profiles exhibited two bands, before and after 450 °C. While the band below 450 °C can be generally attributed to H2 chemisorbed on metal particles (type I), the one at higher temperature is associated with H2 desorption from the subsurface alumina layers or with the spillover phenomenon (type II) [42]. Indeed, the H2-TPD profile of bare γ-Al2O3 does not show any signal variation below 400 °C but an intense band at higher temperature, which might be related to a dehydroxylation process (Fig. 6). Likewise, the band at low temperatures can be divided into several peaks. For instance, the monometallic NiAlGAI catalyst, presents a main peak at 375 °C and additional H2 desorption below 250 °C. According to Ewald et al. [4], the main peak corresponds to hydrogen chemisorbed on Ni surface while the TCD signal at low temperatures can be ascribed to hydrogen adsorbed on the corners of large Ni particles or on better dispersed particles. Noteworthy, the main peak position shifts towards lower temperatures and its intensity increases with Ru content, suggesting that the amount of exposed Ni atoms grows accordingly. Such increase in Ni dispersion was also reported by other authors who incorporated Ru [26], Cr [12] or Fe [21]. The amounts of desorbed H2 calculated from TPD profiles integration are summarized in Table 2. Note that this parameter duplicates with co-impregnation of 1.5 % Ru on Ni/Al2O3 formulation, i.e., the fraction of exposed metal notably rises. Accordingly, the ability to supply dissociated hydrogen under methanation reaction conditions remarkably increases with the Ru content. Based on H2 desorption data, Ni dispersion on the monometallic catalyst was also estimated, resulting a value of 7.9 % (11.5 % by TEM, Table 1). In the case of bimetallic catalysts, dispersion cannot be estimated since exposed atoms of both Ni and Ru, in major and minor extent respectively, contribute in the total H2 desorption below 450 °C. Anyway, the amount of desorbed hydrogen compared to that of NiAlGAI catalyst is more than twice for Ni-1.5RuAl catalyst and hence, this suggests that its metal surface could be around double.In order to determine the morphology, size and distribution of the supported bimetallic particles, HAADF-STEM analysis was conducted. The high-angle Z-contrast annular field imaging together with EDX mapping allowed us differentiating between two or more elements, such as Al (Z = 13), Ni (Z = 28) and Ru (Z = 44). STEM micrographs together with EDX maps of bimetallic catalysts are shown in Fig. 7 . It can be observed that Ni (red coloured) and Ru (green coloured) are homogenously dispersed as individual spherical particles, which means that no alloy is formed during the calcination at 550 °C [26]. Noteworthy, the Ni-0.5RuAl catalyst presents an average Ni particle size of 7.4 nm (calculated from around 50 particles), 4 nm lower than that obtained for monometallic NiAlGAI catalyst. This parameter is even lower for Ni-1.0RuAl and Ni-1.5RuAl, with values of 6.3 and 5.9 nm, respectively. Therefore, Ni particle size is lowered by increasing the amount of co-impregnated Ru. Regardless the metal loading (0.5, 1.0 or 1.5 %), the Ru particle size resulted to be around 4−5 nm for all bimetallic catalysts. Note that some of these particles are located near to Ni ones, especially for catalysts with higher Ru contents (see Fig. 7b and c). The fact that Ni and Ru particles are next to each other or in intimate contact is in agreement with the enhanced reducibility observed by H2-TPR: the neighbour Ru particle acts as H supplier via spillover mechanism favouring the reduction of Ni2+ [23].As is in the case of monometallic catalysts, Ni dispersion on bimetallic catalysts was also estimated by d-FE model and the results are summarized in Table 2. As already observed by H2-TPD, the Ni dispersion is significantly enhanced with Ru loading. In fact, Ni dispersion increases 9.4, 15.6 and 20.0 % by adding 0.5, 1.0 and 1.5 % of Ru, respectively. This behaviour might indicate that both Ru and glycerol solvent act as structural promoters during the calcination process, avoiding the excessive growing or sintering of Ni. Based on the characterization results properly discussed above, it is expected that Ni/Al2O3 catalysts performances are improved with the incorporation of small percentage of Ru in the formulation.Thus, once bimetallic catalysts were characterized and the effect of Ru on physicochemical properties of Ni/Al2O3 determined, their catalytic performance was studied. The conversion-temperature as well as the selectivity-temperature curves of bimetallic catalysts are shown in Fig. 8 . For comparison purposes, the light-off curves obtained for NiAlGAI and RuAlGAI catalysts are also displayed. It can be clearly observed that the addition of increasing amounts of co-impregnated Ru leads to a notable increase of the sigmoid curve slope, especially at mild temperatures (from 275 to 325 °C). Accordingly, the T 50 value is lowered 40 °C by only co-impregnating 1.5 %Ru, which indicates that the presence of Ru considerably improves the activity of Ni/Al2O3 formulation. Although different trends are observed depending on the temperature, all catalysts exhibit selectivity to CH4 higher than 98.5 %. The slightly lower S C H 4 (or higher CO production) observed for bimetallic catalysts at low temperature compared to that of NiAlGAI catalyst may be related to some desorption of CO from low coordinated and inactive Ni and Ru particles. Even so, the methane yield clearly increases with Ru content, being the productivity order at 300 °C as follows: Ni-1.5RuAl ( Y C H 4 = 51 %) > Ni-1.0RuAl ( Y C H 4 = 44 %) > Ni-0.5RuAl ( Y C H 4 = 32 %) > NiAlGAI ( Y C H 4 = 20 %). It should be noted that Ni-1.0RuAl catalyst (T 50 = 305 °C) shows almost the same activity as 3RuAlGAI catalyst, whose noble metal content is three times higher.According to the characterization results, co-impregnation of Ru increases Ni dispersion. Besides, the presence of small Ru particles close to Ni ones considerably improves reducibility and hydrogen chemisorption capacity of nickel. Under reaction conditions, this leads to a greater amount of dissociated H2, which is an essential reaction intermediate, and hence to a superior activity. Thus, the great enhancement observed in the catalytic performance can be attributed to a synergistic effect between Ni and Ru, as also reported by Liu et al. [26].The catalytic behaviour of alumina supported Ni and Ru catalysts proved to be stable for 24-h-on stream and at stoichiometric feed ratio in the former work [34]. Then, in order to accelerate the aging of the catalyst, the stability of monometallic NiAlGAI and bimetallic Ni-1.0RuAl catalyst was evaluated for 50 h-on-stream under harsher reaction conditions: at 325 °C (far for equilibrium conversion) and under sub-stoichiometric feed ratio (H2/CO2 = 3). Noteworthy, the activity of NiAlGAI catalyst resulted to be stable during the evaluated period, observing CO2 conversion values within 35 and 37 %, as shown in Fig. 9 a. This indicates that the catalyst did not suffer from any type of deactivation such as particle sintering or poisoning [4] even though more CO was produced (Y CO = 1.33 %) as consequence of sub-stoichiometric feed. Besides, the CH4 selectivity also remained stable, observing values within 96.3 and 97.1 %.In the case of the bimetallic catalyst, the stability test also included three wet periods (t =2 h) in which increasing amounts of water (10, 20 and 30 mL/min) were fed interspersed by dry periods. It can be observed that, before 25 h-on-stream, the catalytic performance remained stable as observed for NiAlGAI catalyst, obtaining CO2 conversion and CH4 selectivity average values of 60 and 99 %. However, the feed of increasing amounts of water, led to a CO2 conversion drop of around 3 ( y H 2 O = 0.04), 6 ( y H 2 O = 0.08) and 9% ( y H 2 O = 0.12) without a remarkable CH4 selectivity decrease (0.1, 0.25 and 0.5 %). This behavior indicates that water is strongly adsorbed on part of active sites, temporally rendering them unavailable for the reaction. Nevertheless, the activity was completely recovered when switching to dry conditions, indicating that water adsorption or inhibition effect is reversible at short term. Thus, based on the above activity and stability results, it can be concluded that glycerol assisted impregnation is a viable catalyst preparation method.Although it has been shown that bimetallic catalyst have enhanced catalytic properties based on characterization as well as activity results, the individual roles of both Ni and Ru on the CO2 methanation reaction mechanism are not clear yet. Such roles, as well as the identification of the reaction intermediates, will be analysed in this section by Operando FTIR study. Fig. 10 shows the evolution of CO2 adsorption FTIR spectra with temperature for bare γ-Al2O3. Immediately after 5%CO2/Ar exposure at 150 °C (see black spectrum), three clearly distinguishable bands appeared at 1653, 1437 and 1228 cm−1, whose intensity grows with time up to 30 min. These bands, already identified by many authors in the literature [9,43–46], correspond to asymmetric as well as symmetric OCO stretching (νa(OCO) and νs(OCO)) and OH deformation (δ(OH)) vibration modes of bicarbonate species, respectively. Besides, two negative bands can be observed in the hydroxyl region (3800−3600 cm−1) at 3765 and 3665 cm−1 together with a narrow positive peak at 3620 cm−1. The negative ones are attributed to the vibration of OH– groups adsorbed along alumina surface whereas the positive one corresponds to ν(OH) vibration mode of bicarbonates. The presence of negative bands clearly indicates that bicarbonates are formed from CO2 chemisorption on OH– groups of γ-Al2O3, which are partially consumed after 30 min CO2 adsorption [43]. Additionally, other wide and weak bands appear at 1575 and ≈ 1330 cm-1, which might be assigned to νa(OCO) and νs(OCO) vibration modes of (chelating) bidentate carbonates. It is expected that carbonates are formed from CO2 chemisorption on surface O2- of γ-Al2O3 acting as Lewis basic sites [44].The intensity of bicarbonate bands along with those of bidentate carbonates progressively decreases with temperature until practically disappearing at 400 °C, indicating that these species are not strongly attached to alumina. In fact, the weak-medium bond strength of bicarbonate has already been observed by CO2-TPD [34,47]. However, the increase of temperature gives rise to small bands at 1515 and 1457 cm-1, which might be related to formation of more stable organic compounds. Furthermore, additional discrete bands are observed at 1393 and 1375 cm-1, suggesting the presence of formate species. The formation of formates on alumina have already been reported and we suggest they come from reaction between bicarbonate or carbonate and residual H chemisorbed during the pre-treatment [9].After studying CO2 adsorption over the bare support, the CO2 methanation was analysed by means of Operando FTIR over monometallic Ni formulations (NiAlIWI and NiAlGAI catalysts). FTIR spectra recorded under reaction conditions from 150 to 450 °C along with their respective C-species evolution for NiAlIWI and NiAlGAI catalysts are shown in Fig. 11 . Starting by the analysis of NiAlIWI catalyst results (Fig. 11a), note that the black spectrum, which was recorded at 150 °C after 30 min under reaction stream exposure, shows more intense bands in the carbonate region (1800-1200 cm−1) than bare alumina (Fig. 9). Specifically, the bands assigned to bidentate carbonates at 1574 and 1330 cm−1 overlap with additional new ones at 1545 and 1380 cm−1, which might be assigned to vibration of monodentate carbonates [3,45]. This greater number of surface carbonates could be associated with the presence of non-reducible Ni2+O2- or even NiAl2O4 able to adsorb CO2 [37]. As the temperature increases, these bands disappear giving rise to clear and intense bands at 1595, 1395 and 1375 cm−1, characteristic of 3 vibration modes of formates: asymmetric OCO stretching (νa(OCO)), CH deformation (δ(CH)) and symmetric OCO stretching (νs(OCO)), respectively [9,31,47]. Complementary, the band corresponding to CH stretching (ν(CH)) is observed at 2900 cm-1 (not shown), confirming the formation of formate species. After that, new increasing bands appear at 3016 cm-1 (νa(CH)) and 1305 cm-1 (δ(CH)), indicating the formation of methane gas [33]. Note that no bands were verified in the carbonyl region (2100−1800 cm−1) but the characteristic bands of CO gas were observed at 2175 and 2105 cm−1, suggesting that no detectable amount of COads could have formed on Ni° by CO2 disproportionation.Analogously, Fig. 11b displays FTIR spectra of NiAlGAI catalyst. As expected, the same bands and/or species were identified in the carbonate region but with different concentration. In fact, the bands corresponding to carbonates are less intense at the starting temperature (150 °C) probably due to the absence of NiO or NiAl2O4 acting as basic sites in the catalyst prepared by GAI method. Notably, unlike NiAlIWI, NiAlGAI catalyst presents 3 bands in the carbonyl region (2100−1800 cm−1) located at 2020, 1920 and 1860 cm−1. The first is ascribed to the stretching vibration of terminally or linearly adsorbed CO on top single Ni atoms, whereas the other two can be attributed to weakly and strongly attached bridged carbonyls on neighbouring Ni atoms, respectively [20,29,48]. Interestingly, the band corresponding to linearly adsorbed CO shifts with temperature, while the others remain at the same frequency. This shift is associated with changes in CO covering on Ni surface and suggests that these CO species participate in the CO2 methanation mechanism. On the contrary, bridged carbonyls are more stable and may not react with hydrogen [29]. Furthermore, it is wide known that the ν(CO) frequency (in wavenumbers) is associated with the metallic dispersion: the higher the frequency, the higher the dispersion or the lower the Ni particle size. Thus, according to the 3 ν(CO) bands, NiAlGAI catalyst presents particles with different sizes indicative of highly, moderately and poorly dispersed Ni° [48]. This observation is consistent with TEM results, according to which a particle size distribution ranging from 3 to 20 nm is observed. Noteworthy, the lack of adsorbed carbonyls on the catalyst prepared by IWI suggests that there are differences in the Ni electronic state when comparing Al2O3 supported Ni catalysts. In the case of NiAlIWI, it seems that Ni, after reduction pretreatment, is partially oxidized or positively charged (Niδ+) due to the interaction with remaining non reducible Ni2+ species or with Al3+ cations exposed on the alumina surface. As the exposed Ni has electron deficiency, NiAlIWI presents lower affinity to dissociate CO2 by H-assistance or adsorb CO and, hence, no bands are detectable within 2100−2000 cm−1. Although Ni2+ is also able to adsorb CO, no bands were observed among 2300 and 2100 cm−1 assignable to CO on Ni2+ sites. NiAlGAI, by contrast, has much more affinity to CO adsorption since all nickel is in reduced state (Ni°) after being calcined under reductive atmosphere (GAI method).The evolution of the main reaction intermediates and methane with temperature is clearly shown in the attached figures (Fig. 11c and d). In the case of NiAlIWI catalyst (Fig. 11c), it can be observed that the relative concentration of bicarbonates decreases as that of formates increases, following a symmetric evolution (T < 250 °C). This suggests that formates mainly arise from bicarbonates although it cannot be excluded that, in minor extent, carbonates are also reduced into formates [47]. After that, from 250 °C to 325 °C, adsorbed bicarbonates disappear and the formation rate of formates slows down up to zero, i.e., its relative concentration reaches a maximum. This slowdown or depletion matches with methane appearance, whose relative concentration increases exponentially in agreement with activity results. Finally, at higher temperatures (T > 350 °C), the relative concentration of formates decreases, while that of methane slowly increases up to 425 °C approaching to the limited thermodynamic equilibrium of an exothermal reaction. Thus, it can be assumed that formates at the metal-support interface could participate in methane formation. However, it cannot be claimed that formates are directly hydrogenated following the associative mechanism, since not bands characteristic of methoxy species (reaction intermediates) or methanol have been detected by FTIR, as reported by Solis-García et al. [28]. Finally, the appearance of COgas from 300 °C together with the absence of adsorbed carbonyls indicates that this by-product could be formed from decomposition of formates as follows: (11) HCO O ads → HC O ads + O ads → C O ads + O H ads → C O gas + H 2 O gas On the other hand, the corresponding species evolution of NiAlGAI sample is displayed in Fig. 11d. Note that, in general, the relative concentration curves for adsorbed species follow the same trend but are clearly shifted towards lower temperatures. In fact, bicarbonates are depleted or transformed into formates faster (at 275 °C) and the maximum of formates concentration curve, which is also volcano-shaped, is clearly shifted 50 °C into the left (275 vs. 325 °C). Carbonyls relative concentration, in turn, increases with temperature up to 300 °C and then starts depleting. We suggest that carbonyls, which appear from 200 °C, might arise from formates decomposition (Eq. (11)) or, less probably, from CO2 dissociative adsorption. Wang et al. [9] also studied CO2 methanation by FTIR on a 5%Ru/Al2O3 catalyst and concluded that formates are reactive towards the formation of adsorbed CO when it is close to metal particles. From 225 °C, the linearly bonded and, in minor extent, weakly attached bridged carbonyls may be hydrogenated into methane, whereas the strongly attached bridged ones remain stable. From 300 °C, some of the bridged carbonyls could be desorbed as COgas, as revealed by bands at 2175 and 2105 cm−1. Noteworthy, the general shift of adsorbed species evolution indicate that NiAlGAI catalyst has a greater capacity to dissociate H2 and provide H, which is essential to carry out the successive steps of reaction mechanism. This leads to a higher activity at mild temperatures, as evidenced by the higher CH4 relative concentration of NiAlGAI catalyst at 300 °C (0.59 vs. 0.48).The FTIR spectra as well as evolution with temperature of adsorbed species over RuAlGAI and Ni-1.0RuAl catalyst are shown in and Fig. 12 . Additionally, CO2 methanation FTIR spectra of RuAlIWI catalyst are included in Fig. S4 (supplementary material). Mainly, the same species as in the case of Ni catalysts are observed in carbonate region with similar evolution. However, the position and intensity of bands appearing at carbonyl region are different, i.e., the type and distribution of carbonyl species are not the same. In fact, FTIR spectra of Ru/Al2O3 catalysts show a main band at 2015 cm−1 at 150 °C that can be attributed to vibration of linearly adsorbed CO over reduced Ru atoms (Ru-CO) [10,31,49]. This band is more intense to that observed for NiAlGAI catalyst, indicating that Ru has a major capacity or more affinity to adsorb CO than Ni. However, unlike RuAlIWI catalyst, RuAlGAI presents a shoulder at 1970 cm−1 (Fig. 12a) related to stretching vibration of terminal CO species located at metal-support interface ((Al2O3)Ru-CO) [49]. Note that the main band on both Ru catalysts red shifts with temperature from 2015 to 1990 cm−1 due to a decrease in Ru surface coverage by CO, whereas the position of the shoulder observed for RuAlGAI catalyst does not shift and it vanishes above 350 °C along with appearance of CO gas in the cell. Based on these observations, it can be concluded that on-top CO species participates in the reaction but the same cannot be stated for CO species adsorbed at the interface. It seems that this species may not participate in the reaction but eventually desorbed, indicating that RuAlGAI presents a higher fraction of inactive Ru atoms in agreement with the lower TOF/I 0 value obtained.In the case of the bimetallic catalyst, it should be considered that bands appearing at 2100-1800 cm−1 region correspond to carbonyl species adsorbed on both Ni and Ru particles. Thus, what Fig. 12b shows is a combination of bands previously observed for NiAlGAI and RuAlGAI catalysts, characteristic of above-mentioned CO species. The difference is that a new peak is observed at 2056 cm−1 attributed to geminal di-carbonyls on low coordinated Ru [9,10,49], which disappear above 250 °C. According to Panagiotopoulou et al. [50], this species disappears with temperature since it is converted into linearly adsorbed CO due to H2-induced agglomeration of low coordination Ru sites into bigger Ru clusters. Noteworthy, the combination band at 2030 cm−1 corresponding to linearly adsorbed carbonyls is significantly more intense than on NiAlGAI catalyst, indicating that CO adsorption is promoted by the co-impregnation of 1% Ru. On the other hand, the band corresponding to weakly attached bridged carbonyls (at 1910 cm−1) is clearly more intense compared to that observed on NiAlGAI catalyst, which confirms that the bimetallic catalyst presents a higher Ni dispersion (26.3 vs. 11.5 % according to TEM results). As the temperature increases, bands at 2030 and 1910 cm−1 first blue shift up to 250 °C and then red shift to 2010 cm−1 and 1905 cm−1, respectively. The red shift matches with the appearance of CH4 band at 1305 cm−1, suggesting that both species could be reaction intermediates.Regarding to C-species evolutions of RuAlGAI and Ni-1.0RuAl catalysts (Fig. 12c and d), it can be seen that they are quite similar (except to that of CO), observing a shift of curves towards lower temperatures with respect to those of monometallic Ni catalysts. The shift is due to an enhanced catalytic activity, as demonstrated by H2-TPD runs. In fact, the bands corresponding to bicarbonate species vibration at 150 °C are much less intense than those observed for NiAlGAI catalyst in both cases, suggesting that bicarbonates are more easily hydrogenated into formates, which reach maximum concentration value at 175 and 200 °C, respectively. After that, formates at the interface are decomposed into carbonyls and, subsequently, part of carbonyls (most probably linear carbonyls) are hydrogenated into CH4, which relative concentration at 300 °C is 0.64 (for RuAlGAI) and 0.72 (for Ni-1.0RuAl).Finally, it should be highlighted that RuAlGAI presents a considerable higher amount of potentially reactive carbonyls (linearly bonded) but a CH4 yield similar to that of bimetallic Ni-1.0RuAl catalyst, as can be deduced by comparing its respective spectra and C-species evolution at different temperatures. This suggests that the fraction of carbonyls effectively converted into CH4 is lower in the monometallic catalyst. In fact, although Ni-1.0RuAl adsorbs less CO, it disposes of an enhanced dissociated hydrogen supply to reduce CO as a result of the Ni-Ru synergetic interaction. Based on these results, it can be concluded that an effective CH4 formation not only depends on the type and number of adsorbed carbonyls but also on the availability of adjacent H atoms to carry out the CO bond hydrogenation.To sum up, Scheme 1 proposes and depicts the proposed reaction pathways on bimetallic Ni-1.0RuAl catalyst deduced from operando FTIR results shown in this section.Firstly, CO2 is mainly adsorbed on hydroxyl groups (OH−) of γ-Al2O3 to give monodentate bicarbonates (HCO3 −), whereas H2 is dissociated and adsorbed on metal surface. After that, dissociated H2 (H atoms) spillovers and reacts with bicarbonates close to metal particles yielding bidentate formates (HCOO−), which are considered potential reaction intermediates in alumina supported catalysts. Specifically, formates adsorbed at the interface are decomposed into hydroxyls (OH−) on γ-Al2O3 support and carbonyls (CO), which, in the case of monometallic catalysts, are adsorbed either on Ni or Ru surface. However, in the bimetallic system, CO is expected to preferentially adsorb over Ru nanoparticles due to a higher affinity, whereas H2 is adsorbed on neighboring Ni particles acting as H atoms reservoir. Then, carbonyls are reduced by adjacent H atoms into formyl (COH, not observed), which are subsequently hydrogenated into CHXO species (hydroxycarbene (CH2O) or hydroxymethyl (CH2OH)). At certain hydrogenation degree (x = 1–3), the CO bond cleavage of CHxO species occurs (rate determining step), finally releasing CH4 and H2 molecules [16,31,51].In this work, the low temperature activity of Ni/Al2O3 formulation is systematically improved through the use of efficient synthesis (IWI vs. GAI) and the addition of Ru. Overall, catalysts prepared by GAI method presented better catalytic performance than those prepared by IWI. In the case of Ni catalysts, the formation of Ni2+ strongly interacting with the support was avoided by GAI synthesis route, resulting in a higher Ni surface area available for the reaction. Instead, GAI method led to a notable increase in the metal dispersion on RuAlGAI catalyst due to the glycerol enclosing effect but, in return, the specific activity (TOF/I 0) of Ru nanoparticles resulted to be two order of magnitude lower since reaction is structure sensitive. On the other hand, the activity of Ni/Al2O3 was improved even more by co-impregnation of small amounts of Ru as a result of a synergistic combination. In fact, the bimetallic Ni-1.0RuAl catalyst showed remarkably higher Ni dispersion, reducibility, and CO adsorption capacity than NiAlGAI catalyst, observing a methane yield equal to that of 3RuAlGAI. Operando FTIR experiments revealed that CO2 methanation over alumina supported Ni and Ru catalysts proceeds via formation of carbonyl species mainly arising from intermediate formates decomposition, followed by its hydrogenation into CH4. In the bimetallic system, the potentially most reactive species is CO linearly adsorbed over Ru, which is more easily hydrogenated by H atoms supplied from adjacent Ni particles. We conclude that the enhanced CO2 methanation activity of bimetallic catalyst is not only due to a promoted CO adsorption but also to a higher supply of dissociated H2.Adrián Quindimil: Methodology, Investigation, Data curation, Writing - Original Draft, Visualization.M. Carmen Bacariza: Methodology, Investigation, Data curation, Writing - Review & Editing, Visualization.José A. González-Marcos: Verification, Resources, Data curation, Writing - Review & Editing, Visualization.Carlos Henriques: Conceptualization, Resources, Funding acquisition, Validation, Supervision.Juan R. González-Velasco: Conceptualization, Resources, Funding acquisition, Validation, Supervision, Project administration.The authors report no declarations of interest.The support from the Economy and Competitiveness Spanish Ministry (PID2019-105960B-C21), the Basque Government (IT1297-19) and the SGIker (Analytical Services) at the University of the Basque Country are acknowledged. One of the authors (AQ) also acknowledges University of the Basque Country by his PhD grant (PIF-15/351).Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.apcatb.2021.120322.The following is Supplementary data to this article:

Conventional Ni/Al2O3 catalyst, currently used for COx removal in ammonia production, admits room for improvement as catalysts for application in low temperature CO2 methanation, which is the aim of this work. The Incipient Wetness Impregnation (IWI) has been replaced by Glycerol Assisted Impregnation (GAI) method and, afterwards, a secondary metal (Ru) has been co-impregnated forming a bimetallic system. The monometallic as well as bimetallic systems have been characterized by several techniques (TGA, XRD, N2-physisorption, TEM, H2-TPR, H2-TPD, STEM-EDX and operando FTIR) and tested for CO2 methanation reaction in a downflow fixed bed reactor (conditions: P =1 bar, H2: CO2 ratio = 4 and WHSV = 30,000 mL h−1 g−1). GAI method together with a reducing calcination atmosphere (20 %H2/N2) results effective to avoid the formation of large metal particles during the synthesis, especially for Ru/Al2O3 formulation. In fact, the Ru dispersion of the catalyst prepared by GAI (RuAlGAI) is around 5 times higher than that of RuAlIWI catalyst. On the other hand, NiAlGAI presents larger population of reduced particles but bigger in size than NiAlIWI catalyst, which finally provides the former with slightly higher metal surface and superior catalytic performance. By co-impregnating small amounts of Ru (0.5, 1.0 or 1.5 wt%) the Ni surface is considerably increased which, together with Ru synergistic collaboration, results in a methane yield rise from 20 to 44 % at 300 °C. The operando FTIR results show no differences in the reaction pathway with GAI preparation method and incorporation of Ru, but different evolution of reaction intermediates concentration with temperature. The bimetallic Ni-RuAl system presents much higher capacity to adsorb CO and hydrogenate the reaction intermediates (adsorbed formates and carbonyls) by dissociated H2 than its monometallic counterparts.

The wide application of traditional fossil fuels not only produces amounts of green-house gases CO2, but also causes the emission of harmful substances such as SO2 and NOx, both of which endanger the environment [1,2]. The demand for clean energy is increasing. Solar energy, wind power, tidal and geothermal energy are important energy sources today, but they are intermittent and regionally limited. As an energy carrier, hydrogen not only has the characteristics of greenness and easiness of preparation, but also has high specific energy [3,4]. So, solar and wind energy in nature may be stored in the form of hydrogen through water splitting. The chemical energy in hydrogen can be further converted into electric energy through a hydrogen–oxygen fuel cell when necessary. This strategy not only effectively stores natural resources, but also owns the facilitation of energy transport. During the whole process, hydrogen–oxygen fuel cell is one of the keys [4-6].Because it is not affected by the Carnot cycle [7,8], fuel cells have high efficiency in the energy conversion. Compared with other power generation devices, fuel cells have also several advantages, for example, they do not make noise and also do not emit any harmful gases during operation [9,10]. In particular, they can provide continuous electrical energy as long as the fuel is enough. Fuel cells have become one of the most promising and clean cell systems and get wide applications in various fields [11,12]. Several important vehicle manufactures such as Ford, Toyota, BMW have realized the scale up application of hydrogen in automobiles with fuel cells.The hydrogen–oxygen fuel cell converts chemical energy in hydrogen into electric energy; the cell device consists of cathode, anode and electrolyte (Fig. 1 ). During the discharge process in acidic electrolyte, the fuel H2, loses electrons, undergoes an oxidation process and then forms protons at the anode. The lost electrons arrive at the cathode via an external circuit; the oxidant O2, gets the electrons, results in a reduction reaction and then forms water at the cathode [13].Hydrogen-oxygen fuel cells involve two half reactions: the hydrogen oxidation reaction (HOR) at the anode and the oxygen reduction reaction (ORR) at the cathode. In the system of proton-exchange membrane fuel cell (PEMFC), acidic electrolyte was used. With acidic electrolyte, the overpotential of the HOR is relatively lower. In contrast, the overpotential of the ORR at the cathode is high, which highly influences the overall performance of fuel cells [14]. Precious Pt metal is needed as a catalyst for the ORR, which significantly promotes the fuel cells cost [13,15,16]. The cost of a PEMFC highly depends on the usage of Pt catalyst amount. Up to date, the specific powder density of Pt PEMFC locates at 4–5 kW gpt -1, much lower than the cost target demands (>8–10 kW gpt -1) that was proposed by the Department of Energy of the USA (total cell cost of <$30 kW−1).In recent years, new effective catalysts used in alkaline electrolyte for ORR have been developed [17-22]. The typical new non-precious ORR catalysts, present comparable or even better catalytic performance than Pt catalysts under alkaline conditions, are Fe-N-C catalysts [23-28]. Along with the development of anion exchange membrane, researchers then pay attentions to anion-exchange membrane fuel cells (AEMFCs) with alkaline electrolytes, with the expectation of developing non-platinum fuel cells [29-31].In alkaline electrolytes, although the non-precious Fe-N-C catalysts can be used. The overpotential of HOR on the anode end is relatively higher. Therefore, developing of an economical and efficient catalyst for HOR under alkaline conditions is highly required. Generally, the HOR was proposed to proceed with a combination of three elementary steps: Tafel, Heyrovsky, and Volmer steps. These elementary steps can be presented as follows (* is a hydrogen adsorption site on catalyst surface, Had in the Equation denotes hydrogen atoms in adsorbed state) [32-34]: (1) H 2 + 2 ∗ → 2 H ad Tafel step (2) H 2 + ∗ + OH - → H ad + H 2 O + e - Heyrovsky step (3) H ad + OH - → H 2 O + e - + ∗ Volmerstep The Tafel step is a chemical dissociative adsorption process of hydrogen molecules with the formation of two adsorbed hydrogen atoms. The dissociation energy for one H2 molecular is 4.52 eV. Due to the diatomic molecular, two adjacent empty adsorption sites are needed on catalyst surface. This was confirmed by the results that no HOR activity of Pt atoms were shown when they are atomically dispersed on a sulfur doped carbon substrate [35]. Heyrovsky step involves not only the chemical dissociative adsorption of hydrogen molecules, but also the transfer of one electron. In this process, OH– species in the vicinity of catalyst surface are needed, which was also argued to be OHad adsorbed on catalyst surface but not OH–. Volmer step reveals the electron transfer from adsorbed hydrogen atoms to form one molecular water.The HOR either follows Tafel-Volmer or Heyrovsky-Volmer mechanisms [36]. In alkaline electrolyte, the catalyst should provide at least dual adjacent sites for all of the three elementary steps. In steps of Heyrovsky and Volmer, one site for hydrogen adsorption, another for OH or OH– adsorption; while in acid solution, the only required sites are those for chemisorb hydrogen. This is commonly believed to be the reason for the much lower HOR activity of catalysts in alkaline electrolyte [36,37]. All of the three steps involve Had. Thus, the hydrogen binding energy between hydrogen atoms and active sites (noted as HBE) is a critical factor for HOR, which is also used as a descriptor of HOR activity for a catalyst. Smaller HBE is more favorable for the Volmer step; while if HBE is too small, Tafel or Heyrovsky steps would be difficult to occur [32,38,39]. Especially, it should be noted that the hydrogen binding energy on catalyst surface would vary with electrolytes of varying pH values and potentials for HOR [40-43].According to the calculation of density functional theory (DFT), the metal-hydrogen bond strength on various metal surfaces were estimated with free energy of hydrogen adsorption (ΔGH) that is correlation with the HBE. The relationship between the exchange current density of hydrogen evolution reaction and ΔGH is present in Fig. 2 . Pt locates at the top of the volcanic-type relationship. It indicates that Pt catalyzing the reaction results the highest exchange current density, while the binding of hydrogen on W, Fe, Co, Ni, and Pd metals is too strong, which locates on the left side of the volcanic-type relationship. Accordingly, Cu, Au and Ag metals are on the other side of the relationship. Weaker hydrogen binding will lead to weaker adsorption of H atoms on the active sites, which is not favorable for the Tafel step. In contrast, stronger hydrogen binding will not favorable for Volmer step [39,44,45]. Therefore, HBE (or ΔGH) is usually used as a descriptor of HOR activity for a catalyst [32,38,39,46,47], an excellent catalyst requires active sites with suitable HBE value.Except the HBE, the OH– species, that is, the surface adsorbed OH/OH– species also significantly affect the HOR catalytic activity in alkaline electrolyte [49,50], since for steps of Heyrovsky and Volmer, bifunctional catalytic surface that can adsorb hydrogen atom and OH– (oxophilic surface) is needed (Fig. 3 ) [51]. Pt is an excellent component for the dissociative adsorption of molecular hydrogen, while monometallic Pt surface is not a good plat for the adsorption of OH– at the conditions relevant for HOR [52]. Ramaswamy et al. suggested that during the HOR reaction, Had is formed on Pt surface. A voltage penalty must be paid to get over repulsive force due to the transfer of negatively charged OH– anions from outer-Helmholtz plan (OHP) to negatively charged Pt surface, which causes the formation of Pt-Had···OHq-ad cluster and then comes to the formation of water molecule (Fig. 3A) [51]. On a bimetallic surface, quasi-specifically absorbed hydroxide species (OHq-ad) on Mp surface is able to react with Pt-Had forming water. Similar effect appears when the alloying elements come to Cu, Ni, Co, Nb, etc. It should be noted that these elements do not form Hupd duo to the oxide/hydroxide on its surface. Specially absorbed hydroxide species react with Pt-Had forming MOx(OHy) instead of M−OHad (Fig. 3C). Introducing of Ru in Pt catalyst can induce more than five-fold increase of HOR catalytic activity in 0.1 M KOH [53]. It is believed that the introduction of Ru increases site amounts for the adsorption of hydroxyl species.Similar cases are Ni(OH)2-metal composite catalysts. It was found the presence of Ni(OH)2 could promote the adsorption of OH–, which endows the catalyst surface with bifunctions and promotes HOR [55]. Ddekel [49] and Strmcnik et al. [56] also indicated that adjusting the OHad on the catalyst surface is an effective method to optimize HOR activity. Their studies suggested that the catalytic activity for HOR can be improved by hybriding or alloying of two metals sites together. In this bimetal system, the Metal-H intermediates react with the OHad on the adjacent metal sites that can bond of OH species with Metal-OHad or Metal-HUPD-OHad configurations. Thus, Davydova [57] and Koper et al. [58] proposed a new description of catalytic activity for HOR catalysts, i.e. binding energy of OH species on active sites (OHBE). So far, however, whether the OHBE can be used as a descriptor of HOR activity is still under debate, since there are also some studies showing that the presence of second metal sites influences the catalytic activity through modulating the HBE [59].Rotating disk electrode (RDE) with a standard three-electrode system is usually used to check the electrocatalytic activity for a HOR catalyst. Be noted that the rotating electrode is needed to weaken the influence of hydrogen diffusion in the electrolyte on the catalytic current. H2-saturated 0.1 M KOH aqueous solution is often used as the electrolyte. An appropriate amount of catalyst ink that was usually prepared by dispersing catalysts in Nafion-ethanol solution was dropped onto a glassy carbon disk electrode and was allowed to be dry, which was used as the working electrode. To avoid the contamination of Pt on the catalyst, the counter electrode cannot be Pt-based materials. Carbon rods and glassy carbon pieces are encouraged to be used. To check the catalytic activity, the linear sweep voltammetry method is used. With the increasing of potentials, hydrogen will be oxidized and then so will water. Be noted during the operation, the catalyst surface will possibly be oxidized.During the HOR process, the influence of hydrogen mass transfer on current density cannot be ignored. The influence of hydrogen mass transfer on the polarization curve should be corrected by the Koutecky-Levich formula (Equation (4)), from which the kinetic current ik can be calculated by Equation (4). (4) 1 i = 1 i k + 1 i d = 1 i k + 1 B c o ω 0.5 In Equation (4), i is the total current actually measured, ik is the kinetic current density, id is the diffusion limited current that is related to the Nernstian diffusion potential. B, Co , and ω are the Levich constant, the solubility of hydrogen in 0.1 M KOH, and the speed of the RDE, respectively. Plotting 1/i vs. 1/ω0.5 will help the deduce of kinetic current [60]. (5) i k = i 0 exp α F RT η - exp α - 1 F RT η (6) η diffusion = - RT 2 F ln 1- i d i l With the Butler-Volmer formula (Equation (5)) (in which α and η are the transfer coefficient and the overpotential; R, T, and F are the universal gas constant, 8.314 J mol−1 K−1, the Kelvin temperature, the Faraday constant, 96,485C mol−1, respectively), the HOR exchange current density i0 can be estimated by fitting of the current density with Butler-Volmer formula (Fig. 4 ). The exchange current density i0 is the most important parameter to characterize the catalytic activity of the catalyst. Besides the exchange current density i0 , the Nernstian diffusion overpotential (ηdiffusion) is another important parameter that can be obtained from equation (6), in which il is the hydrogen diffusion limited current density [60].Platinum group metals (PGM) including Ir, Pt, Pd and Rh show excellent catalytic performance for HOR [61-66]. Pt shows the highest catalytic activity for HOR. However, in alkaline electrolytes, the catalytic activity of PGM catalysts is greatly reduced, which is about two orders of magnitude lower than that in acidic media. For platinum group metal HOR catalysts used in alkaline environment, the current research focus is to develop various nanostructures to reduce the amount of precious metals [51,67-69]. On the other hand, researchers devote to optimizing the catalytic performance by introducing other components [41,42,70,71]. Researchers have shown the electronic structure of PGM active sites can be improved by alloying, which can promote the Volmer step and increase the HOR activity. One of the typical examples is the AuPt/C catalyst for HOR [62]. Compared with commercial PtRu/C and commercial Pt/C, the resulting AuPt/C catalyst showed superior HOR activity with a normalized current density up to 0.158 mA/cm2 in 0.1 M KOH solution. The researchers believed that the improved HOR catalytic activity is mainly due to the introduction of Au that reduces the interaction strength between Pt atoms and adsorbed hydrogen atoms. Yang et al. found Rh2P/C catalyst showed improved catalytic HOR performance than that of Rh [64]. They proposed that the presence of phosphorus atoms may help reduce the HBE of active sites on the catalyst surface and thus promote the improvement of catalytic performance. The exchange current density obtained with the Rh2P/C catalyst is 2.4 times of the Rh/C catalyst under the same conditions. Recently, P-Rh/C and P-Ru/C catalysts were also reported for HOR, further confirming the introduction of P can improve the catalytic activity [72,73]. The optimized P-Ru/C catalyst shows a normalize exchange current density of 0.72 mA cm−2 which is even 2 times higher than that of commercial Pt/C [73].Based on the electrodeposition method, Liu et al. selected a variety of metals (Mg, Cr, Mn, Fe, Co, Ni, Cu, Ru, La, and Ce) to modify the Pt planar electrode, indicating that both of the electronic effect and the oxophilic effect due to the presence of introduced metals play important roles for HOR, but the former can affect the catalytic performance more significantly [74]. To distinguish the influence of the electronic effect and the oxophilic effect on the catalytic activity, core–shell structure is an ideal model catalyst for the study. Considering the relatively lower price of Ir metal, Liu et al. synthesized IrNi@Ir core–shell nanoparticles. The exchange current density of IrNi@Ir core–shell particles reaches 1.22 mA cmIr -2 at overpotential of 0.05 V in 0.1 M KOH solution. In this core–shell catalyst, the IrNi core changes the electronic structure of the Ir shell and optimizes the HBE on Ir sites [61]. It seems that no oxophilic effect provide contribution for the improved catalytic process, since the introduced nickel atoms locate at the core position.Although PGM present good catalytic performance for HOR, the high price of PGM still hinder their wide application. Especially in alkaline media, a large amount of PGM catalysts must be loaded to overcome the challenge of two orders of magnitude decrease of catalytic activity for HOR. It prompts researchers to develop non-precious metal catalysts [75-79]. To date, the non-precious metal-based HOR catalysts are mainly Ni-based materials (Table 1 ) [77,80-83]. Recently, Oshchepkov et al. reviewed the electrochemical behavior of Ni catalysts for the oxidation of hydrogen-containing fuels [77]. Here, we focus on the HOR catalyzed by Ni-based catalysts.Studies have shown that with partial oxidation, the catalytic performance of Ni catalysts can significantly improve [84,85]. The Ni-NiO composite structure formed by partial oxidation has a bifunctional surface, in which the metallic Ni can adsorb hydrogen and the NiO phase can adsorb OH species. The thus bifunctional surface promotes its catalytic activity. In addition, the presence of NiO may also be beneficial to optimize the adsorption of hydrogen on the surface of nickel nanocrystals. For example, by partially oxidizing of Ni nanoparticles, the exchange current density can increase from 6.2 to 56 μA/cm2 (Fig. 5 ) [86]. Based on the Arrhenius formula, Oshchepkov et al. studied the effect of temperature on the kinetics of HOR/HER with Ni electrode in the range of 298–338 K [86]. It was found that partial overlaying of NiO on Ni electrode can lead to the reduction of HBE on Ni sites and promote an increase in the reaction rate of the Volmer step, thereby induce a significant enhancement of the HOR/HER kinetics [87].Oshchepkov et al. also explored how does the oxidation of Ni electrode effect its kinetics in catalyzing HOR/HER [88]. They show 10 times of enhancement of the catalytic activity for HOR and HER with Ni catalyst that was oxidized in air. The researchers also believed that the presence of NiO optimizes the adsorption of hydrogen atom on the surface of Ni.With this acknowledgement, researchers designed and prepared Ni-NiO catalysts to enhance its catalytic HOR activity. Yang et al. synthesized a variety of Ni/NiO/C catalysts at different temperature through the calcination of metal organic framework precursors. The catalytic performance is an order of magnitude higher than Ni/C. Both of the catalytic stability and CO tolerance ability of the Ni/NiO/C catalyst are better than those of Pt/C [89]. Pan et al. obtained Ni(OH)2-Ni/C catalyst by electrochemical oxidation of Ni/C catalyst and found that after electrochemical oxidation, the HOR exchange current increased by 6.8 times (Fig. 6 ) [90].Except for NiO, the introduction of CeO2 and MoO2 can also improve the performance of metallic Ni catalysts. The improving mechanism for catalytic activity is similar to that of NiO. The Ni active sites on Ni-CeO2 catalyst own a more thermo-neutral ΔGH* than that on pristine Ni catalyst, that is, the induction of CeO2 weakened the HBE. In addition, an enhanced adsorption behavior for OH* was also found on Ni-CeO2. These two effects synergistically accelerate the Volmer step so that improves the HOR activity [91]. Similarly, Deng et al. modified Ni nanocrystals with MoO2, where Ni nanocrystals provided hydrogen adsorption sites. MoO2 as a corrosion-resistant stable oxide can not only promote the dissociation of water, but also optimize the OH adsorption due to the positively charged surface, and thus accelerating the Volmer step, eventually improving the HOR activity [92].Another strategy to optimize the performance of the Ni-based HOR catalysts is to introduce a second metal component, which will optimize the HBE of the Ni active sites through the interactions between them [71,93-95]. Usually, the introduction of the second metal component will also help the improvement of the oxidation resistance for Ni nanocrystals. Based on density functional theory, Tang et al. predicted that Ni-Ag alloys have multiple adsorption centers, and some active centers possess the optimal HBE [48]. With a low temperature physical vapor deposition method, they synthesized a series of binary Ni-Ag alloys, which can catalyze both HER and HOR with catalytic performance much high than pristine metallic Ni and with improved stability. Besides Ni-Ag bimetals, theoretical calculation also indicates that the ternary CoNiMo alloys surface shows optimized HBE (Fig. 7 ) [96]. Compared with pure Ni materials, NiMo bimetals synthesized by electroplating method exhibit significantly increased HOR activity. Significantly, the performance of ternary CoNiMo product presents 20 times higher catalytic activity than pure Ni. The presence of Co in the catalyst would adjust the structure of d-orbitals of nickel, inducing the improvement of catalytic activity. Kabir et al. synthesized NiMo catalysts by thermally reducing the transition metal precursor with carbon carrier, and suggested that the Mo component does not directly participate in HOR process, but affects the HBE of Ni active centers [97].Another metal that can significantly improve the HOR performance of Ni catalytic centers is Cu [98-101]. The Ni-Cu alloy is thermodynamically unstable. Oshchepkov et al. discussed the effect of different amounts of Cu in Ni/C catalysts on the HOR activity, and found that the addition of Cu can enhance the oxidation resistance of Ni [98]. After that, Cherstiouk et al. also prepared NiCu/C catalysts [99]. The optimized mass activity of Ni0.95Cu0.05/C catalyst was 1.5 times of that of pure Ni sample. The exchange current density reached 14 μA cm−2. The catalytic activity of Cu-Ni alloys varies with the preparation methods. Ni0.6Cu0.4 catalyst prepared by magnetron co-sputtering method shows significantly improved activity with the exchange current density four times of pure Ni catalyst [100].Besides, recent researchers found that W, Pd, Ru, Mo, etc. can also improve the catalytic performance of nickel. The modification of W on Ni nanocrystals can not only adjust the electronic structure of the Ni surface, but also restrain the oxidation of Ni surface [29]. Baoks et al. deposited Pd on the surface of Ni film by electrodeposition technology. Within 17% of Pd coverage, the catalytic current density increases linearly with increasing Pd coverage [102]. Recently, single atomic Ni sites loaded on Ru nanosheets were firstly studied as HOR catalysts. The single atomic Ni sites loaded on Ru nanosheets possess optimized HBE, which results in improved catalytic activity for HOR. It is believed that both of Ni and Ru centers take part in the catalytic process [103]. Metallic of Mo can also enhance the catalytic process of Ni-based catalysts [104]. Duan et al. recently reported a tetragonal nickel-molybdenum nanoalloy, MoNi4, which shows a high apparent exchange current density of 3.41 mA cm−2. It is suggested that the high exchange current density is attributed to the improved filling antibonding state [104].The support highly affects the catalytic activity. There is strong interaction between metallic active species and the supports, which will cause electron transfer between them and finally turning the adsorption behavior for hydrogen, improving the catalytic performance. For nickel-based HOR catalysts, heteroatoms (N, S, P) doped carbon materials are excellent supports, which can significantly improve the performance. N atoms on the support can affect the d-orbitals of nickel and improve the HOR performance. In the Ni/N-doped carbon nanotubes catalyst (Ni/N-CNT), N-CNT was used as a support for Ni particles (Fig. 8 ) [105]. The N-CNT support makes the mass activity and exchange current density increase by 33 and 21 times; the exchange current density of Ni/N-CNT catalyst reached 28 μA cmNi -2 (Fig. 9 ). A series of heteroatoms-doped carbon supported Ni nanoparticles catalysts (the heteroatoms are S, B, and N) were prepared through pyrolysis of organic molecules followed by NaBH4 reduction. The results show that the catalytic activity for HOR is in the order of Ni/S-doped carbon > Ni/N-doped carbon > Ni/B-doped carbon≈Ni/carbon. The exchange current density of the Ni/S-doped carbon catalyst is up to 40.2 μA cm−2. The optimal performance can be attributed to the following two main factors. On one hand, the carbon support anchored Ni nanoparticles, which helps to prevent the aggregation of Ni nanoparticles and so to keep the Ni nanoparticles with smaller size. The smaller Ni nanoparticles make the catalyst a larger specific surface for electrochemical reaction. On the other hand, that is more important, the electronic interactions between Ni sites and the support optimized the HBE of Ni and eventually improves the catalyst activity [106]. However, systematical acknowledge about the interaction between supports and Ni centers in HOR catalysts is still lacking.The heteroatom content in carbon is obviously an important parameter for the catalyst. With SiO2 microspheres as a template and melamine as a raw material, Jiang et al. designed a N-doped carbon support with spherical shell microstructure, in which the nitrogen content can be tuned from 0 to 21.6 at%. Ni nanoparticles were loaded on the support by an impregnation-reduction method. The electrochemical measurements in alkaline media unveils the “volcano-like” regularity for the N-doping content and HOR activity. The catalyst with 8.7 at% nitrogen doping shows the optimal catalytic activity with calculated exchange current density of 30 μA cmNi -2 [107]. For the preparation of Ni-based catalyst with carbon support, Simonov et al. developed a simple method for Ni/C catalyst preparation by nitrate decomposition. The introduction of Ni(CH3COO)2 can prevent the gasification of the carbon support during the catalyst synthesis [108].Besides metallic Ni as HOR catalysts, some Ni-based compounds including Ni3N can also catalyze HOR. By calcining of NiO or Ni(OH)2 under a nitrogen atmosphere. Ni3N can be obtained. During the transition from Ni to Ni3N, the d band of Ni shifted downward, and the interface charge was transferred from Ni3N to the carbon carrier. It weakens the HBE on nickel sites and the resulting product shows significantly enhanced HOR activity and stability [109]. Yang et al. discussed the electron transfer effect in Ni3B/Ni heterostructure. The inter-regulated effect not only weaken H* adsorption, but also strengthen OH* adsorption, prompting the enhancement of HOR activity [110].The present studies on HOR catalysts mainly focus on the development of catalysts used in alkaline media. Although platinum group metals such as Pt, Ir, Rh, and Ru show high electrocatalytic activities in acidic media, their high cost limits the wide application. In addition, they also face a major challenge of two orders of magnitude decrease in catalytic kinetics under alkaline media. Therefore, designing of low-cost, high-efficiency, non-precious catalysts working at low pH medium highlights the research value. The common designing ideas for non-noble metal-based HOR catalysts that are mainly composed by Ni-based materials include the introduction of second component (such as oxides/hydroxides, alloying), the using of various heteroatoms doped supports, with the aim of adjusting the electronic structure of the catalytically active sites and making their electronic structure be close to the optimal state. Designing catalyst surface with bifunctions (adsorption of hydrogen and adsorption OH species at the same time) is also believed to be an effective strategy for the improvement of HOR activity. Generally, promoting the reduction of the hydrogen binding energy of Ni sites plays a vital role for the catalytic process, which would accelerate the kinetic process of the rate-determining step for HOR.On the other hand, for Ni-based HOR catalysts, some important issues should also get attention. Firstly, Can the metallic Ni keep the original crystal structure on the surface before and after the catalytic process? The possible surface re-construction on Ni surface during electrocatalytic process should be understood. Secondly, the hydrogen diffusion in the catalyst layer and the adsorption behavior of hydrogen on various Ni sites are also not fully clarified at this stage. In addition, the mechanisms of the Ni-based catalyst stability and the performance decay are also not clear. Detailed experimental studies with in-situ characterization techniques and systematically theoretical calculation would provide useful information on the above listed issues.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors are grateful for National Natural Science Foundation of China (No. 21776115). Six talent peaks project in Jiangsu Province (XCL-2018-017). Foundation from Marine Equipment and Technology Institute for Jiangsu University of Science and Technology, China (HZ20190004)

With the rapid development of anion exchange membranes, researchers began to shift their attention from proton exchange membrane fuel cells (PEMFCs) with acid electrolytes to anion exchange membrane fuel cells (AEMFCs) with alkaline electrolytes. Non-precious metal catalysts such as Fe-N-C catalysts are available for ORR at the cathode in alkaline electrolytes. But at the anode, the catalytic reaction kinetics of Pt catalysts in alkaline media is two orders of magnitude slower than that in acid media, which prompts researchers to develop new low-cost and high-efficient non-precious metal catalysts for HOR. Up to date, the typical non-precious metal catalysts for HOR are Ni-based materials. In this minireview, we firstly introduced the elementary steps of HOR and the important activity parameters for a HOR catalyst. Secondly, we briefly describe the performance of various Ni-based HOR catalysts reported in recent years.

Over the decades, the large consumption of fossil fuels due to anthropogenic activities has released heat and greenhouse gas into the atmosphere. The concentration of CO2, one of the major greenhouse gases, in the atmosphere exceeds 410 parts per million (ppm), which is much higher than 280 ppm in the pre-industrial period. 1 The thermal radiation from the sunlight and the surface of the earth can be trapped by CO2 molecules, augmenting the global temperature and aggravating the issue of global warming. 2 The over-emitted CO2 is truly one of the top concerns for human society. The CO2 reduction reactions (CO2RR) that convert CO2 molecules into value-added products such as CH3OH, CH4, CO, and HCOOH give an alternative strategy to consuming CO2 while mitigating the energy crises. In addition, realizing CO2RR by clean energies such as solar energy offers an approach to alleviate both global warming and energy crisis issues in a green concept. 3 To fulfill this, photocatalytic CO2 reduction reactions (PCO2RR) has gained much attention because it utilizes semiconductor materials as photocatalysts to absorb sunlight as the driving force to reduce CO2. Fig. 1 a depicts the fundamentals of PCO2RR on a semiconductor catalyst. Typically, the photocatalysis takes three main steps: i) the generation of the photo-generated carriers (e−/h+ paires) within the semiconductor photocatalyst through harvesting the incident light; ii) the transfer of the photo-generated electrons and holes to the surface of the photocatalyst; iii) the catalytic reactions (CO2RR and the oxidation half-reaction) on the surface of the photocatalyst. 4 The photo-generated carriers (e−/h+ paires) are essential to PCO2RR as they are the keys to the catalytic reactions. The photo-generated electrons in the conduction band (CB) participate in CO2RR to reduce CO2 molecules. The photo-generated holes in the valence band (VB) participate in the counter-reaction (oxidation half-reactions, such as water oxidation reaction). During the formation of the photo-generated carriers, the electrons and holes tend to recombine with each other through surface recombination and volume recombination, which prohibits catalytic efficiency. Besides, the potentials of the VB maximum and the CB minimum need to straddle the redox potentials of the oxidation half-reactions and CO2RR, respectively. 5 Equations (1)–(8) shows some typical reaction steps involved in CO2RR, along with the reactions for H2 and O2 productions (vs. Normal Hydrogen Electrode (NHE), in aqueous solution of pH = 7). 3 , 6 The semiconductor catalyst needs to meet not only the kinetic barriers but also the thermodynamic requirements to achieve a successful PCO2RR. Moreover, the bandgap between CB and VB determines the wavelength range that the semiconductor can absorb from sunlight. When the value of the bandgap is larger than ca. 3.1 eV, the semiconductor catalyst can solely harvest ultra-violet illumination, which is only a small fraction of the sunlight. Increasing the light-harvesting ability of the photocatalysts requires the bandgap of the semiconductor materials to be relatively narrow. The single pure semiconductor catalyst alone usually cannot afford sufficient catalytic efficiency for PCO2RR. Great efforts are made in the modification of the semiconductor catalysts (e.g. the refinement of crystallinity, defect engineering, heterojunction construction, etc.) to increase the catalytic performance (e.g. product selectivity and activity, light harvesting capability, stability) during PCO2RR. 5 The goal is to design rational photocatalysts with I) the matching CB and VB potentials for CO2RR and the oxidation half-reactions; II) the narrow bandgap to absorb sufficient sunlight; III) low recombination rate of the photo-generated carries; IV) sufficient and highly efficient active sites for CO2RR and the oxidation half-reactions. This review concentrates on the most recent advanced photocatalyst designs for PCO2RR, where the superiorities of semiconductor modification and integration are highlighted (Fig. 2 ). Hybridization strategies of photocatalysts such as surface engineering and band engineering are explained with some typical examples (e.g. co-catalyst designs, photosensitizer, heterojunction construction). Then, promising results from structural engineering and single-atom active site fabrications are exposed, along with the biohybrid catalyst designs. Finally, the perspectives on the remaining challenges and future focuses are presented.Currently, mainstream works focus on modifying the semiconductor photocatalysts by hybrid engineering such as surface modification and the integration of different semiconductors to form heterojunctions. Table 1 lists some typical hybrid photocatalysts for PCO2RR.As the catalytic reaction takes place on the surface of the catalyst, its modification with active sites or support co-catalysts is a promising way to enhance the catalytic activities of the catalyst materials. For example, An et al. modified the Fe tetraphenyl porphyrin (FeTPP) catalyst with an alkyne-functionalized supramolecular synthon to form an iron porphyrin box (PB) bearing 24 cationic groups (FePB-2(P)) that offered a synergy of porosity and charge effects. The modified FePB-2(P) exhibited a 41-times enhancement in catalytic performance, as compared to the original FeTPP, toward CO production in PCO2RR. 8 The modification of the metal co-catalysts on the semiconductor surface can serve as not only the active sites to capture CO2 molecules for the activation but also the electron trap to separate photo-generated carriers, demonstrating excellent PCO2RR performance (e.g, Au–Cu alloy modified on TiO2 substrates for CH4 and C2H4 productions 9 ; Ni cluster shell on NiO core for the CO generation 10 ).Surface engineering for light harvesting enhancement is another approach to increase photocatalytic efficiency. As photocatalysis utilizes solar energy to drive the catalytic reactions, the photocatalysts’ efficiency of harvesting sunlight is vital in the practical aspect. Currently, many good semiconductor catalysts such as metal oxides exhibit wide bandgap, which limits light absorption within the ultra-violet range, only a small fraction of the incident solar radiation (less than 5%). 11 It makes those wide-bandgap photocatalysts less attractive in industrial applications, as sunlight mostly contains visible and infrared lights. The integration of semiconductor catalysts with light-response-efficient materials allows for the improved light-harvesting ability of the photocatalysts. Modifying semiconductor catalyst surfaces with photosensitizers is one of the approaches to achieve this goal. Wang et al. constructed a number of photosensitizers of homoleptic Al (III) for PCO2RR with emission quantum yields from 10% to 40%. 12 Fig. 3 a shows the molecular structures of the homoleptic Al (III) complexes. The light absorption band center can be tuned by changing the ligands attached to the Al (III) photosensitizers (Fig. 3b). Fig. 3d illustrates the PCO2RR activities of different catalysts coupled with Al (III) photosensitizers. The catalyst of [Fe(qpy)(OH2)2](ClO4)2 (FeQPY; qpy = 2,2′:6′,2″:6″,2‴-quaterpyridine) affords the most durable and stable CO production among all the catalysts. Fig. 3c demonstrates the reaction scheme for the PCO2RR with Al (III) photosensitizers/FeQPY with 1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzo[d]-imidazole (BIH) as the sacrificial electron donor. This noble-metal-free system achieves a CO selectivity of 99% and a TON value of 10250 under 450 nm light illumination in PCO2RR (Fig. 3e). Das et al. developed a porous organic polymer (POP) as a light harvester for improving PCO2RR. The composite catalyst of In2.77S4 and POP was built by electrostatic interaction and exhibited a C2H4 product selectivity of 98.9% with a yield rate of 67.65 μmol g−1 h−1 under irradiation of visible light. 13 To maximize the utilization of sunlight, scientists have discovered some promising semiconductor photocatalysts with narrow bandgap to harvest visible light, such as carbon nitride, 14–16 bismuth oxyhalides, 17 , 18 CdS, 19 layered double hydroxides (LDHs), 4 , 20 , 21 etc. However, those visible light-response semiconductor materials suffer from severe recombination of the photo-generated carriers. The surface modification of the semiconductor catalyst with a metal co-catalyst such as Pd/TiO2 22 and Ag–TiO2, 23 which can trap the photo-generated electron, emerges as a powerful strategy to separate the photo-generated carries and contributes to for an enhanced PCO2RR performance. It is well applied to visible light-response photocatalysts as well. For instance, Yue et al. reported a Bi-MOF/BiOBr photocatalyst where Bi-MOF was in-situ mounted on the BiOBr for PCO2RR with a CO yield rate of 21.96 μmol g−1 h−1. 24 Another approach is to integrate the semiconductor catalysts with metal co-catalysts. In this strategy, using plasmonic metals offer extra benefits for light harvesting. The nanocatalysts of plasmonic metal interact strongly with incident light, generating a localized surface plasmon resonance (LSPR). The LSPR effect can easily tune the light absorption wavelength from ultra-violet to near-infrared through adjusting the geometry structure of the plasmonic metal catalyst. 25 , 26 For example, Cu/TiO2 enables absorbance at 500–600 nm from the LSPR of Cu particles to improve the CO production in PCO2RR. 27 The LSPR effect of Au particles helps achieving the PCO2RR under low-intensity irradiation at 420 nm. 26 Based on the promising results provided by the assistance of the plasmonic metal Au in light harvesting and catalytic performance, modifying Au catalyst with other metal alloys has proven to enhance the employment of photons with low energy for PCO2RR. Hu et al. fabricated plasmonic light harvesting Au rods and coupled them with a co-catalyst shell of CuPd alloy to achieve highly effective PCO2RR to CH4 production. 28 Fig. 4 a demonstrates the structure of the Au/CuPd core-shell composite. Plasmonic catalysis often happens close to the surface of the catalyst (within the range of plasmon-induced local field). Pure Au nanorods are unlikely to collide with CO2 molecules, resulting in its low CO2 conversion efficiency (Fig. 4a, left). For Au/CuPd, the CuPd shell can capture CO2 molecules to increase the CO2 concentration on the surface of the catalyst, promoting the probability of further conversion and activation (Fig. 4a, right). The thickness of the shell and the Cu/Pd ratio can be easily adjusted by controlling the number of metal precursors during the synthesis. The optimized Au/CuPd catalyst can achieve a CH4 yield rate of 15.6 μmol g−1 h−1, which is almost 40 times higher than the one achieved by pure Au rods (Fig. 4b). The apparent quantum efficiencies are up to 0.1% at 800 nm (Fig. 4d). The optimized thickness of the CuPd shell is believed to maximize the number of active sites on the catalyst surface and strengthen the electron-phonon scattering effect, contributing to the best CH4 production. Fig. 4c illustrates that C2H4 and C2H6 can also be detected after the long-term reaction test. This suggests the multiple proton-coupled electron transfer ability of the Au/CuPd core-shell catalyst. To avoid the scattering loss of incident light during PCO2RR, a spherical-structured gas-solid reaction system is further applied (Fig. 4e). The novel spherical-structured reaction system can realize the re-incidence of scattered photons, which benefits the catalytic reaction system with a plasmonic effect. With the optimization of the partial pressure of CO2 and the volume of H2O in the system, the production rate of CH4 can be further improved to 0.55 mmol g−1 h−1 (Fig. 4f). The spherical-structured reaction system affords a catalytic performance of around 35-folds to that of the conventional reaction system which only allows one single incident photon pass. Besides, the apparent quantum efficiency in the spherical-structured reaction system reaches 0.38% at 800 nm. It offers the opportunity to utilize the Au/CuPd core-shell catalyst in PCO2RR under low-energy near-infrared irradiation.When applied in photocatalysis, single pure semiconductors often suffer from the issues of wide bandgap that limits the visible light harvesting, unmatched CB or VB potential positions to drive the catalytic reactions, etc. Integrating two semiconductors with different CB and VB potential positions to form a heterojunction becomes a useful approach to overcome these difficulties.There are three types of heterojunctions when integrating two different semiconductors. Type I is “straddling” where the CB and VB of one semiconductor are entirely contained in those of another semiconductor (Fig. 5 a). In this type of heterojunction, both photo-generated electrons in CB and holes in VB of semiconductor I tend to transfer to the CB and VB of semiconductor II by the potential difference, respectively. It comes with two drawbacks. First, the CB and VB potentials of semiconductor II are less powerful to drive the catalytic redox reactions. Second, the accumulated photo-generated holes and electrons in semiconductor II exacerbate the recombination of the photo-generated carriers. Hence, the “straddling type” heterojunction is not ideal for photocatalysis.Type II is “staggered” where the CB and VB of one semiconductor overlap with those of another semiconductor (Fig. 5b), generating a favourable heterojunction. Once semiconductor I have more negative CB and VB potential positions than semiconductor II, the photo-generated electrons tend to transfer from the CB of semiconductor I to the CB of semiconductor II while photo-induced holes tend to transfer from the VB of semiconductor II to the VB of semiconductor I. This new possible transfer pathway created by the “staggered type” heterojunction can separate the photo-generated electrons and holes in the semiconductors and significantly reduce the recombination rate of the photo-induced carriers in both semiconductor I and semiconductor II, boosting photocatalytic performance. The only drawback of this type of heterojunction is that the CB and VB that participate in the catalytic reactions are the less effective ones. Namely, the CB potential of semiconductor II is less negative than that of semiconductor I, and the VB of semiconductor I is less positive than that of semiconductor II.Type III is a “broken gap” where the bandgaps of two semiconductors do not overlap (Fig. 5c). In this circumstance, the photo-generated electrons and holes between two semiconductors fail to transfer. 29 Therefore, the type II heterojunction is of interest in photocatalysis. Many works focus on the fabrication of type II heterojunction to boost catalytic performance. For example, a heterojunction between CsPbBr3 and graphitic carbon nitride (g-C3N4) is built based on the band position differences between g-C3N4 and CsPbBr3. It provides a stable CO yield rate of 975.57 μmol g−1 h−1 in PCO2RR for 76 hours. 30 Zhu et al. constructed the heterojunction in a heterostructure of ZnIn2S4–CdS for PCO2RR. 31 The ultrafast transient absorption spectroscopy proves the accelerated charge transport in the heterostructure (Fig. 6 a and b). With the assistance of Co(bpy)3 2+ as co-catalyst for CO2RR and TEOA as the sacrificial compound to consume photo-generated holes (Fig. 6e), it exhibits the CO yield of around 33 μmol in the first hour (Fig. 6c) under visible light irradiation and retains its initial catalytic activity after 5 cycles (Fig. 6d) of the PCO2RR measurements.As aforementioned, the type II heterojunction still shows limits concerning the CB and VB positions that participate in the catalytic reactions. Inspired by natural photosynthesis, an electron transfer pathway where the photo-induced electron in the CB of semiconductor II travels to the VB of semiconductor I to perform the recombination is proposed (Fig. 5d). It creates an electron transfer pathway with the shape of the letter Z, which is named “Z-scheme” in the heterojunction construction. The Z-scheme electron transfer pathway allows photo-generated carriers from the stronger CB and VB in the two semiconductors to participate in the catalytic reactions, which overcomes the difficulty raised in the type II heterojunction catalysts.In another approach, the Z-scheme is constructed without building any additional bridge between two semiconductors, this is called the direct Z-scheme system. 6 The construction of the internal electric field in the heterojunction between two semiconductors proves to be an effective strategy to achieve the direct Z-scheme. Wang et al. built a 2D/2D Z-scheme heterostructure of Ni–CsPbBr3/Bi3O4Br for PCO2RR. 32 Driven by the difference in the Fermi levels of the two semiconductors, the photo-generated electrons transfer from Ni–CsPbBr3 to Bi3O4Br. It leads to increased charge densities at the interface, positive on the Ni–CsPbBr3 side and negative on the Bi3O4Br side, respectively. The internal electric field formed in this space charge region creates the band-bending effect. Fig. 7 a depicts the Z-scheme electron transfer pathway where the electrons are directed from the CB of Bi3O4Br to the VB of Ni–CsPbBr3. Because of the bending effect, the shape of the electron pathway looks more like the letter “S” rather than “Z”. Under this circumstance, the term “S-scheme” is preferred over “Z-scheme” to highlight the band-bending effect. The Ni–CsPbBr3/Bi3O4Br with the S-scheme heterostructure demonstrates an excellent 98.2% CO selectivity with a CO yield of 387.57 μmol g−1, which is more than 10 times higher than that of CsPbBr3 (Fig. 7b). Long et al. combined Fe2O3 and CdS to achieve the direct Z-scheme charge transfer pathway by a built-in internal electric field as well. The Z-scheme heterojunction of Fe2O3/CdS enables a CO yield of 9.3 μmol g−1 in the first hour during PCO2RR, which is much better than the sum of those of pure Fe2O3 and CdS. 33 The unique Z-scheme favors not only the separation of photo-generated electron/hole pairs but also the redox capacity of the photo-generated electrons and holes involved in catalytic reactions such as CO2RR.Alternatively, when the Z-scheme is constructed with the help of the additional bridge between two semiconductors, it is called the indirect Z-scheme system. For example, Wang et al. delicately fabricated lanthanum (La)- and rhodium (Rh)-doped SrTiO3 (SrTiO3:La, Rh) and the light absorber of Co(II) bis(terpyridine) modified molybdenum (Mo)-doped BiVO4 (BiVO4:Mo) and the RuO2 catalysts on a gold layer (SrTiO3:La, Rh|Au|RuO2–BiVO4:Mo) for the PCO2RR (Fig. 7c). The Au layer in SrTiO3:La, Rh|Au|RuO2–BiVO4:Mo is believed to bridge the CB of RuO2–BiVO4:Mo and the VB of SrTiO3:La, Rh, which constructs an effective indirect Z-scheme for the PCO2RR to HCOO− and the H2O oxidation reaction. It affords an HCOO− selectivity of 97% with a production rate of around 20 μmol g−1 h−1 (Fig. 7d). 34 The catalyst structure plays an essential role during photocatalysis, benefiting the spatial separation of the photo-generated carriers for improved catalytic performance. Taking 2D catalysts as an example, the ultra-thin 2D nanosheets facilitate the transfer of the photo-generated electron/hole pairs to the surface of the catalyst, promoting catalytic activity. For instance, ultrathin 2D NiMgV-layered double hydroxide nanosheets afford excellent CO and CH4 productions in PCO2RR. 35 Liang et al. constructed a 2D dislocated bilayer MOF that allowed 100% product selectivity of CO in PCO2RR. 36 However, 2D structured materials often suffer from agglomeration, leading to the prolongation of the carriers’ transfer and a decreased active area of the catalysts, which jeopardizes the catalytic performance. Wang et al. morphologically modified CuInZnS by introducing negatively charged Ti3C2T x to interfere with the nucleation and growth processes of CuInZnS. It creates a defect regulation in CuInZnS and results in thinner 2D nanosheets of CuInZnS with a bigger specific surface area and larger pore size than those of the pristine CuInZnS. The hybrid 2D Ti3C2T x -CuInZnS exhibits a CO production rate of 42.8 μmol g−1 h−1 in PCO2RR. 37 With the proper synthetic routes, escalated hierarchical structures such as leaf/flower/litchi-like nanostructures can be fabricated through 2D nanostructures. 38 As aforementioned, Zhu et al. constructed the heterojunction in a heterostructure of ZnIn2S4–CdS for PCO2RR. 31 The authors also took advantage of the hierarchical structure construction by anchoring 0D CdS quantum dots on the 3D ZnIn2S4 nanoflowers, which contributes to better charge transfer and separation during PCO2RR. Other than the stacked 2D nanosheets/nanofibers to construct hierarchical structures, the 3D core-shell structure is also a hot topic in this field. Li et al. reported a 3D core-shell heterostructure made of c-TiO2@aTiO2-x (OH) y with HO–Ti-[O]-Ti surface frustrated Lewis pairs (SFLPs) on the shell in PCO2RR. 39 Fig. 8 a illustrates that the SFLPs dissociate dihydrogen, forming hydrides and charge-balancing protonated hydroxyl groups at unsaturated Ti sites on the surface of the catalyst to promote the PCO2RR performance. The crystalline-amorphous heterostructure prolongs the lifetime of the electron/hole carriers (Fig. 8b). It allows a CO production rate of 5.3 mmol g−1 h−1 which is 350 times of the original c-TiO2 catalyst.Wei et al. developed a hollow multi-shelled structure (HoMS) of CeO2@CeO2/TiO2 for CH4 production in PCO2RR. The high-angle annular dark field scanning TEM (HAADF-STEM) and X-ray energy dispersive spectral (EDS) mapping images in Fig. 8 e and f present the multi-shelled structure of CeO2@CeO2/TiO2. The quadruple inner shells are made of CeO2, which reduces CO2 into CO accumulated within the multi-shelled structure. The amorphous outer shell is made of TiO2, which further converts the accumulated CO into CH4 (Fig. 8g). The tandem CO2–CO–CH4 reaction of the HoMS CeO2@CeO2/TiO2 affords the CO and CH4 production rates of 97.6 μmol g−1 h−1 and 15 μmol g−1 h−1, respectively (Fig. 8 c and d). The HoMS structure can be destroyed by grinding CeO2@CeO2/TiO2 into debris, causing a dramatic catalytic performance decrease (CH4 production rate of 3.4 μmol g−1 h−1). The control experiment highlights the significance of the hierarchical structure for the tandem reaction in PCO2RR. 40 It reveals that the rational structure construction strategy of hybrid catalysts can further promote catalytic performance by combining different functional semiconductors to achieve a complex reaction of PCO2RR for high-value-added products.The production of C2+ is difficult for CO2RR. Cu-based materials are the most known metal catalysts to facilitate the C–C coupling for the formation of C2+ products. 41 Many Cu-based photocatalysts show catalytic activities towards the production of C2+ products, such as In–Cu SA/PCN for CH3CH2OH production, 42 P/Cu SAs@CN for C2H6 production 43 and NiCo SA-TiO2 for the CH3COOH production. 44 With the assistance of the hierarchical structure engineering strategy, Cu-based photocatalysts demonstrate promising activity toward C2+ production in PCO2RR. Chakraborty et al. applied the operando surface reconstruction of Wurtzite phase CuGaS2 to form a 2D CuO layer-modified CuGaS2, which led to a C2H4 production of 20.6 μmol g−1 h−1 with a selectivity of 75.1% in PCO2RR. 45 Jia et al. grew p-type Cu2O selectively on an Au bipyramid with the assistance of CTAB, fabricating a hetero-structure of dumbbell top which spatially separate Au and CuO active catalysts in an hybrid photocatalyst Au/CuO. Taking advantage of both the structural engineering and the LSPR effect, dumbbell-shaped Au/CuO affords much higher C2+ productions under near-infrared irradiation than under visible light irradiation. 46 Nowadays, atomically dispersed catalysts and single-atom catalysts (SACs) have gained much attention for their maximum active site utilization and superior catalytic performance. 47 , 48 SACs have commonly supported metal catalysts that consist of isolated monometallic moieties (single metal atoms surrounded by neighboring atoms within the support), which offer well-defined active sites in the catalysts. 49 , 50 Integrating the state-of-art highly active SACs with semiconductor substrates holds great promise in the catalytic performance enhancement of PCO2RR. Anchoring the metal atoms on the substrate catalysts is currently one of the most common strategies to construct SAs such as Au, Ag, and Pt SAs on graphene. 51 Ou et al. fabricated Au SAs on red P support (Au1/RP) with low electronegativity to absorb CO2 for PCO2RR. 52 RP is a single-element constituent, which provides a uniform coordination environment for Au SAs. Fig. 9 a-e depicts that Au SAs are evenly distributed on the RP support. Au1/RP affords a C2H6 production rate of 1.32 μmol g−1 h−1 with a selectivity of 96% (Fig. 9f). The turnover frequency (TOF) also reaches 7.39 h−1, which is pretty high among many photocatalysts (Fig. 9g). The CO2 adsorption analysis (Fig. 9h) and the in-situ diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS, Fig. 9i) confirm that Au1/RP provides better CO2 adsorption and activation than those of pure RP. It is suggested that the P atoms around the Au SAs are electron rich and capable of serving as active sites for CO2 activation. In the meantime, the Au SAs can decrease the C–C coupling energy barrier for C2H6 production in PCO2RR (Fig. 9j).Metal organic frameworks (MOFs) are popular precursors and templates in the CO2RR catalyst fabrication, as well as in the SACs synthesis. 53–55 The metal node and organic ligand structure of MOFs facilitate the formation of SAs with a post-treatment of pyrolysis. 56 , 57 Using MOFs as the precursors is a solid strategy to fabricate SAC-combined semiconductor catalysts. Other than the MOF strategy, the facile thermal polymerization strategy is also successful in SACs fabrication. Shi et al. reported a Cu–In dual-metal SAC by the polymerization strategy for PCO2RR. 58 As demonstrated in Fig. 10 a, the metal ion salt, urea, and MIL-68 are used as precursors to fabricate the polymer. Followed by the calcination, Cu–In dual site SACs are prepared. TEM results in Fig. 10 b-d show that Cu and In are evenly distributed on the CN nanosheets with isolated atom spots. The X-ray absorption spectroscopy reveals that the coordination environments of Cu and In in CuInCN are dominated by Cu–N and In–N rather than Cu–Cu and In–In, which confirms the single Cu and In atoms in the CuInCN catalyst (Fig. 10 e-p). The Cu–In dual site SAC affords the superior CO yield rate of 1.2 mmol g−1 h−1 that is almost ten times of the one of the original CN catalyst under visible light irradiation while maintaining the catalytic activities after 6 runs of the tests (Fig. 10 q and r).Besides, many novel synthetic strategies were discovered over recent years. For instance, Wang et al. applied a co-dissolution strategy to dissolve [PtV9O28]7− into [V10O28]6− to obtain the Pt single-atom catalyst that allowed a CH4 production rate of 247.6 μmol g−1 h−1, much higher than that of Pt particles. 59 The numerous possibilities for the SAC synthesis of different elements hold great potential for refining the PCO2RR catalysts for better catalytic activities.For abiotic photocatalysts, the effective transfer of the photo-generated electrons into chemical bonds for CO2RR is challenging. Biohybrid catalysts offer alternative means for CO2RR for the production of biofuels and biochemicals with higher product selectivity by integrating catalyst materials with biological cells. 60–66 Many successful biohybrid semiconductor photocatalysts have been developed for PCO2RR. For example, moorella thermoacetica-based biohybrid photocatalysts are efficient for the production of acetic acid from CO2, such as the self-photosensitization of moorella thermoacetica/CdS nanoparticles (Fig. 11 a and b) 67 and moorella thermoacetica/Au nanoclusters (Fig. 11 c and d). 68 Moreover, light-harvesting artificial cells containing cyanobacteria afford to fix CO2 into glucose. 69 Other than bacteria, protein can also be used in biohybrid semiconductor systems for PCO2RR. Saif et al. designed a –NH2 group functionalized 1D protein-encapsulated CeO2 nanorods (PCNRs) for CO and CH4 productions in PCO2RR. 70 As depicted in Fig. 12 a, the bovine serum albumin (BSA) is applied as an efficient biotemplate to synthesize PCNRs. With TEOA as the electron donor to consume the holes and the assistance of RhB, PCNRs demonstrate excellent activity toward H2 production (Fig. 12b). When carried out in a CO2 gas environment, PCNRs show catalytic activities towards CH4 and CO productions under light irradiation of 400 nm < λ < 780 nm (Fig. 12c) with great suppression of H2 production. Fig. 12d and e reveal that PCNRs exhibit CO and CH4 production rates of 5.0 and 3.3 μmol h−1 g−1, which are 50 and 83 times higher than those of non-biohybrid CeO2, respectively. The authors believe that the protein hybrid PCNRs significantly enhance the material stability and facilitate the transfer of photo-generated holes to promote the separation process of photo-generated carriers.Researchers should still pay attention to some key points, concerning the future design of biohybrid photocatalysts. First, semiconductor materials that are compatible with bio cells need to be rationally constructed to protect microbial cells from deactivation while sufficiently harvesting visible light to provide enough electrons for CO2RR. Second, the interface of the biotic–abiotic should be tailored for quick charge transfer, accelerating the separation rate of electron/hole pairs for better catalytic performance. Finally, the integrated bio-material should be able to efficiently utilize photo-generated electrons to produce fuel chemicals from PCO2RR. 60 After the design of highly efficient catalysts, many characterization techniques are required to understand the properties and uniqueness of the photocatalyst material for excellent catalytic efficiency. X-ray photoelectron spectroscopy (XPS), X-ray powder diffraction (XRD), UV–Vis spectroscopy, and X-ray absorption spectroscopy (XAS) are common techniques for exploring the element valence, crystallization, bandgap value, and coordination environment information of the materials. Scanning electron microscopes (SEM), transmission electron microscopy (TEM), and energy dispersive spectroscopy (EDS) can directly provide visual images of morphologies and element distributions of catalyst materials. The combination of different characterization techniques is important to confirm one's assumption. For example, the analysis of XAS and XPS helps to confirm the N/O ratio in N–Ti–O/V[O] for the production of C2+ products on TiO2 catalysts. 82 For revealing the reaction mechanism of CO2RR, the density functional theory (DFT) simulations emerge as one of the most useful tools to demystify the structure-activity relationship and catalytic mechanism for complex catalytic systems such as hybrid catalysts. 83–86 In the meantime, in-situ characterizations, such as in-situ XRD, in-situ XAS, and in-situ Raman spectroscopy, are powerful experimental approaches to trace the evolution of the catalyst structures and reaction intermediates during the catalytic reaction of CO2RR. 87–90 Apart from Raman spectroscopy, in-situ Fourier transform infrared absorption spectroscopy (FTIR) can also provide evidence of reaction intermediates, uncovering the reaction pathway of CO2RR. 91 , 92 Wu et al. prepared an oxygen vacancy (Vo)-rich MoO2-x for PCO2RR. 93 The Vo-rich MoO2-x exhibits a CH4 production rate of 5.8 and 12.2 μmol g−1 h−1 under NIR and full light irradiation in PCO2RR, which is around 10- and 7-fold to those of the Vo-poor MoO2-x , respectively (Fig. 13 a). Besides, the Vo-rich MoO2-x performs PCO2RR directly under an air atmosphere with a CO production rate of 6.5 μmol g−1 h−1 under NIR irradiation (Fig. 13b). Vo-rich MoO2-x also demonstrates good stability in PCO2RR activity under NIR irradiation in concentrated CO2 after 4 runs (Fig. 13c). In-situ FTIR is applied to reveal the reaction mechanism of PCO2RR over the MoO2-x catalysts. Fig. 13d shows that carbonate species and *CO2 − appear in the dark, suggesting the absorption and activation of CO2 on the Vo-rich MoO2-x surface. Additional peaks of *COOH (1593 cm−1), *CH3O (1170 and 1100 cm−1), and *CHO (1082 cm−1) appear and are gradually strengthened with the increase of illumination time under NIR (Fig. 13e). These intermediates are essential to the production of CH4 in CO2RR. When illuminated under full spectrum light, the IR peak intensities of the intermediates further increased (Fig. 13f). It reveals the efficient light response of Vo-rich MoO2-x and the possible reaction pathway (CO2→ *CO2→*COOH→ *CO→CO or *CHO→*CH2O→*CH3O→CH4) for PCO2RR.With the advancing of characterization techniques, powerful characteristic techniques are exploited for revealing the compositional effects in hybrid catalysts for PCO2RR. Chen et al. recently applied spatiotemporally resolved surface photovoltage measurements (SPVM) on the facet and defect-engineered Cu2O catalysts (Fig. 14 a) to map the holistic charge transfer processes at the single-particle level on the femtosecond timescale. 94 Fig. 14b depicts that the {001} facet has more accumulated photo-generated electrons than the {111} facet of the Cu2O octahedron, owing to the high Cu vacancies (VCu) on the {001} facet. Fig. 14c demonstrates that the anisotropic charge transfer is optimized with a truncated octahedral configuration, suggesting the contribution of the inter-facet built-in electric field to the anisotropic charge transfer. SPVM in Fig. 14d further illustrates that the moderate hydrogen-compensated VCu (H–VCu) results in an efficient spatial separation of the photo-generated carriers on {111} and {001} facets. On the contrary, the extreme incorporation of H–VCu leads to the quench of the photo-generated electron/hole pairs (Fig. 14e). The photoemission electron images in Fig. 14f visualize the dynamics of anisotropic electron transfer for single Cu2O particle, indicating that the ultrafast inter-facet electron transfer contributes significantly to the anisotropic electron distribution. Au being selectively deposited on the {001} facet of Cu2O can also be successfully probed by SPVM (Fig. 14 g-i). It is confirmed that the H2 evolution performance is associated with the anisotropic charge transfer of Cu2O (Fig. 14j). This powerful SPVM technique brings meaningful insights into the photo-carrier transfer dynamics, which can be transplanted to PCO2RR. With more advanced characterization techniques developed and applied to the PCO2RR catalysts, the rational design of the next-generation photocatalyst for excellent catalytic performance can be precisely and systematically guided.To sum up, semiconductor photocatalysts often suffer from unsatisfying catalytic performance (e.g. with production rate at the level of μmol g−1 h−1) in PCO2RR, owing to the poor light harvesting ability, the low separation rate of the photo-induced carriers, and stability issues. Designing novel semiconductor materials with highly efficient catalytic performance for PCO2RR that address these issues is a priority. The hybridization of semiconductor catalysts through different approaches such as surface modification and band engineering strategies can integrate the advantages of the different semiconductor catalysts and co-catalysts to prohibit the recombination of the photo-generated electron/hole pairs and promote the light response of the semiconductor catalysts for PCO2RR. The hierarchical structure construction of semiconductor catalysts also contributes to the separation of the photo-generated electron/hole pairs and sometimes can even achieve a spatially coordinated tandem reaction to produce C2+ products from PCO2RR. Active sites are essential to catalysis. Anchoring highly catalytically effective SACs on semiconductor catalysts holds great promise for the augmentation of catalytic activities in PCO2RR. To improve product selectivity, integrating biological materials that are highly selective in photosynthesis with semiconductor catalysts has been proven to be an effective solution. Moreover, by using advanced characterization techniques for in-situ probing, the underlying mechanism of the reaction pathway and catalyst structure evolution can be demystified for next-generation PCO2RR catalyst design. Other than the proposed solutions discussed above, current research on PCO2RR still needs to focus on the following perspectives: • The duration of the catalytic performance is always a big issue in PCO2RR. Most photocatalysts present a catalytic activity duration of dozens of hours, which is excessively low for industrial requirements (more than thousands of hours). 95 , 96 The fabrication of highly active semiconductor materials for CO2RR without accumulating residues of poisoning intermediates is forever a priority. Besides, the addition of the protective layer on the surface of photocatalysts to mitigate the decay of the catalyst structure during PCO2RR could be a possible solution to enhance the catalyst stability. 7 • In addition, the combination of photocatalysis with other catalytic methods such as electrochemical catalysis and thermal catalysis (e.g. photo-electrocatalysis) holds great potential to improve both the catalytic activities and product selectivity in CO2RR. 97 , 98 CO2 capture and storage (CCS) takes a vital part in the mitigation of over-emitted CO2 because it is energy-consuming. 99 Sorbent porous materials such as metal oxides, 100–102 zeolites and amine-functionalized silicas, 103 covalent organic frameworks (COFs), 104–106 MOFs, 107–109 and porous carbons 110 are effective toward CO2 captures. Many of these sorbent materials also demonstrate photocatalytic activity towards CO2RR. Hence, coupling CO2-capture with PCO2RR can be a practical means in the further market for better efficiency. 111–115 • To finally achieve a practical application, a rational catalysis setup needs to be designed for high catalytic efficiency and scale-up production. Many promising CO2RR systems have been developed for electrolysis, such as gas phase flow cells, solid oxide electrolysis cells (SOECs), etc. 116–118 For the photocatalysis-related setup upgrade, there are also some interesting discoveries, e.g., the back-illuminated photoelectrochemical flow cell for the increased solar-to-fuel conversion efficiency, 119 the aerobic environment for the improved PCO2RR in a less restricted reaction condition. 120 The duration of the catalytic performance is always a big issue in PCO2RR. Most photocatalysts present a catalytic activity duration of dozens of hours, which is excessively low for industrial requirements (more than thousands of hours). 95 , 96 The fabrication of highly active semiconductor materials for CO2RR without accumulating residues of poisoning intermediates is forever a priority. Besides, the addition of the protective layer on the surface of photocatalysts to mitigate the decay of the catalyst structure during PCO2RR could be a possible solution to enhance the catalyst stability. 7 In addition, the combination of photocatalysis with other catalytic methods such as electrochemical catalysis and thermal catalysis (e.g. photo-electrocatalysis) holds great potential to improve both the catalytic activities and product selectivity in CO2RR. 97 , 98 CO2 capture and storage (CCS) takes a vital part in the mitigation of over-emitted CO2 because it is energy-consuming. 99 Sorbent porous materials such as metal oxides, 100–102 zeolites and amine-functionalized silicas, 103 covalent organic frameworks (COFs), 104–106 MOFs, 107–109 and porous carbons 110 are effective toward CO2 captures. Many of these sorbent materials also demonstrate photocatalytic activity towards CO2RR. Hence, coupling CO2-capture with PCO2RR can be a practical means in the further market for better efficiency. 111–115 To finally achieve a practical application, a rational catalysis setup needs to be designed for high catalytic efficiency and scale-up production. Many promising CO2RR systems have been developed for electrolysis, such as gas phase flow cells, solid oxide electrolysis cells (SOECs), etc. 116–118 For the photocatalysis-related setup upgrade, there are also some interesting discoveries, e.g., the back-illuminated photoelectrochemical flow cell for the increased solar-to-fuel conversion efficiency, 119 the aerobic environment for the improved PCO2RR in a less restricted reaction condition. 120 There is no conflict of interest to declare.This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), the Fonds de Recherche du Québec-Nature et Technologies (FRQNT), Centre Québécois sur les Materiaux Fonctionnels (CQMF), Institut National de la Recherche Scientifique (INRS), and École de Technologie Supérieure (ÉTS). Dr. G. Zhang thanks for the support from the Marcelle-Gauvreau Engineering Research Chair program.

Using clean solar energy to reduce CO2 into value-added products not only consumes the over-emitted CO2 that causes environmental problems, but also generates fuel chemicals to alleviate energy crises. The photocatalytic CO2 reduction reaction (PCO2RR) relies on the semiconductor photocatalysts that suffer from high recombination rate of the photo-generated carriers, low light harvesting capability, and low stability. This review explores the recent discoveries on the novel semiconductors for PCO2RR, focusing on the rational catalyst design strategies (such as surface engineering, band engineering, hierarchical structure construction, single-atom catalysts, and biohybrid catalysts) that promote the catalytic performance of semiconductor catalysts on PCO2RR. The advanced characterization techniques that contribute to understanding the intrinsic properties of the photocatalysts are also discussed. Lastly, the perspectives on future challenges and possible solutions for PCO2RR are presented.

In recent decades, much attention has been paid to various methods of water remediation, including sorption (Abdel Maksoud et al., 2020; Ahsan et al., 2020; Bartczak et al., 2018; El-Sayed, 2020; Fang et al., 2020; Żółtowska-Aksamitowska et al., 2018), ozonation (Schmitt et al., 2020), photocatalytic and catalytic approaches (Acharya and Parida, 2020; Fang et al., 2020; Kubiak et al., 2020; Kumari et al., 2020; X. Liu et al., 2020b; Lu and Astruc, 2020; Siwińska-Ciesielczyk et al., 2020), membrane separation and sedimentation (Chen et al., 2020). However, each method has drawbacks and may lead to the production of problematic wastes requiring safe and efficient disposal. Although there is no universal solution to these problems, catalysis currently seems to have a crucial role in the development of effective processes and methods that can maximize effectiveness and minimize waste generation and energy demand (De et al., 2016).Particular interest is focused on metal-based catalysts, which have been extensively utilized in a wide range of applications (Finiels et al., 2014; Lee and Lee, 2020; H. Liu et al., 2020a; Ma et al., 2019; Yan et al., 2016; Yang et al., 2019; Zhang et al., 2007; Zheng et al., 2017; Zuo et al., 2016), including biotechnological treatment processes (El-Sayed, 2020; Jankowska et al., 2019; Zdarta et al., 2019). Such materials have been investigated in particular with regard to their use in oxidation–reduction reactions of organic compounds (Ambursa et al., 2021; Parmeggiani and Cardona, 2012). For comparison, Table 1 presents a set of studies on the catalytic oxidation of phenol and its derivatives and the reduction of 4-nitrophenol using various metal-based catalysts.As shown above, heterogeneous catalysis offers endless possibilities in the degradation of organic compounds via either oxidation or reduction reactions. Recent results have proved the usefulness of a wide range of materials in the removal of phenolic compounds from water and wastewater. It is notable that composites based on transition metals are most commonly used in the degradation of organic compounds (Verdine, 2019).Among the metals commonly applied in catalysis, nickel is one of the best known (Finiels et al., 2014; Lipshutz et al., 2003; Plumejeau et al., 2015). It is one of the most abundant elements in the Earth's crust and is approximately 5000 times cheaper than gold (De et al., 2016). Due to the long history of the use of nickel in catalysis, the literature on nickel catalysts is vast and covers an enormous number of reactions, including oxidation, reduction, hydrogenation and reforming reactions (Lipshutz, 2001; Lipshutz et al., 2003). Despite the significant success of applications of this element in industry, purely nickel-based catalysts still cannot be used for environmental applications, due to problems with selectivity, stability and activity (Kour et al., 2020; Qin et al., 2020). To overcome these obstacles, nickel–carbon composites have been developed. The use of carbon as a support enables the good dispersion of the metal-containing phase on the support surface. Moreover, the properties of carbonized materials, including high chemical and thermal stability, high porosity, low density and weight, may help to fulfil the strict requirements for environmental applications (De et al., 2016).The main aim of this study was to prepare novel, 3D fibrous-like nickel-based bio‑carbons and test their potential use as catalysts in model reactions. As a source of carbonaceous material, spongin-based scaffolds derived from the marine demosponge Hippospongia communis were used. This biopolymer creates unique systems of channels built by interwoven fibres. Spongin chemistry is considered complex. Despite some similarities to collagen and keratin, spongin is distinguished by the presence of halogens (such as I and Br) and xylose. The physicochemical properties of spongin-based scaffolds, including thermal stability, persistence in acidic media and the presence of various heteroatoms, suggest that they can be used as an innovative precursor of bio‑carbons, including metallized 3D carbon materials (Petrenko et al., 2019; Szatkowski et al., 2018). In this study, the low-temperature carbonization of spongin-based scaffolds was used to generate hierarchical 3D carbonaceous structures preserving the original morphology of the spongin-based skeleton. The scaffolds underwent modification with nickel compounds via the simple and fast sorption reduction method, to obtain novel catalysts. The resulting nickel–carbon composites were effective catalysts in the reduction and oxidation reactions of various phenolic compounds. The reaction kinetics and the reusability of the prepared catalysts were also investigated, and possible mechanisms of reduction and oxidation were proposed.Spongin-based skeletons of the marine sponge Hippospongia communis (Porifera: Demospongiae), purchased from INTIB GmbH (Germany), were used as a precursor material. The samples were cleaned in distilled water for 1 h, then moved to an ultrasonic bath for 40 min. The sponge skeletons were then immersed in 3 M HCl in a purification process. This process was carried out in three stages; after each stage the solution of HCl was exchanged for a fresh one with a concentration of 3 M. The first and second stages of purification were conducted for 6 h, and the third for 3 h. After the acid purification process, samples were cleaned with distilled water to neutral pH, dried, and cut into smaller pieces. The prepared material was subjected to a carbonization process. Carbonization of spongin-based samples was conducted in an R 50/250/13 tube furnace (Nabertherm, Germany) in a nitrogen atmosphere. The process was carried out in a temperature range from 400 to 600 °C, with a 1 h plateau and a heating rate of 10 °C/min, and cooling by thermal inertia to 50 °C. Before the carbonization process, samples were conditioned for 2 h in a nitrogen atmosphere at a temperature of 20 °C.The method was based on the treatment of carbon materials with a solution of nickel nitrate in a concentration of 5 mg/L. Each sample was placed in a three-neck round-bottom flask filled with 50 mL of nickel nitrate salt solution. The first stage, including sorption, was carried out for 1 h with continuous stirring (800 rpm). Next, the reduction was carried out by dropping into the former solution 50 mL of 0.5 mol/L sodium borohydride at a rate of 5 mL/min. After dosing, the reduction process was continued for an additional 30 min, again with continuous stirring (800 rpm). The sorption and reduction procedures were repeated three times. Finally, the metallized materials were dried at 60 °C.The crystalline structure of the prepared materials was evaluated by the X-ray diffraction method, using a Rigaku Miniflex 600 analyser (Rigaku, Japan) operating with Cu Kα radiation (α = 1.5418 Å). Patterns were obtained over an angular range of 10–80°. Parameters of the crystalline structure of the samples were calculated using PDXL: Integrated X-Ray Powder Diffraction Software (Rigaku, Japan). The analysis was based on the International Centre for Diffraction Data (ICDD) database.SEM analysis was performed using an EVO-40 scanning electron microscope (Zeiss, Germany). Transmission electron microscopy (TEM) investigations were carried out using a Hitachi HT7700 microscope (Hitachi, Tokyo, Japan) operating at an accelerating voltage of 120 kV. Materials were prepared in epoxide resin and cut into thin layers using a microtome to prepare specimens. EDS X-ray microanalysis was prepared using a Tescan apparatus (Czech Republic) with Gamma-Tec instrumentation from Princeton Inc. (USA). Energy-dispersive X-ray fluorescence spectrometry (XRF) was carried out using an Epsilon 4 spectrometer equipped with a high-resolution silicon drift detector (SDD), typically 135 eV@ Mn-Kα (Malvern Panalytical, UK).XPS analysis was performed using a Prevac spectrometer (Prevac Ltd.) with a hemispherical Scienta R4000 electron analyser with a Scienta SAX-100 X-ray source (Al Kα, 1486.6 eV, 0.8 eV band) and an XM 650 X-ray monochromator (0.2 eV band). The pass energy of the analyser was set to 50 eV for the regions (high-resolution spectra) Ni 2p, O 1s and C 1s (with a 50 meV step). The base pressure in the analysis chamber was 5·10− 9 mbar, and the pressure during the collection of spectra was not higher than 3·10− 8 mbar. The porosity characteristics of the obtained materials were determined by the multipoint BET (Brunauer–Emmett–Teller) method using data for adsorption under relative pressure (p/p o ) obtained with an ASAP 2020 instrument (Micromeritics Instrument Co., USA). FTIR analysis was performed with a Vertex 70 apparatus (Bruker, Germany) using the attenuated total reflection (ATR) method. Electrophoretic mobility was measured using a Zetasizer Nano ZS instrument equipped with an autotitrator (Malvern Instruments Ltd., UK) by analysing 0.01 g of catalyst in 25 mL of 0.001 mol/L sodium chloride solution at 25 °C. Changes in the conductivity and pH values of the suspension were observed during the measurement. The pH of the suspensions was adjusted by an automatic titrator using hydrochloric acid (0.2 mol/L) or sodium hydroxide (0.2 mol/L). The zeta potential was obtained from the electrophoretic mobility by the Smoluchowski equation (Sze et al., 2003).To evaluate the possible application of the prepared metallized materials, they were used as catalysts in the reduction reaction of 4-nitrophenol (4-NP) to 4-aminophenol (4-AP). This reaction was carried out in a quartz cuvette containing 2.5 mL of 4-nitrophenol solution in water (concentration 10 mmol/L). After adding a water solution of sodium borohydride (0.5 mL with concentration 100 mmol/L) and 5 mg of catalyst, the reaction was started. The reduction progress was measured using a UV–Vis spectrophotometer (Jasco V700, Japan) based on spectra obtained after every 60 s of the reaction. In addition, the kinetics of the reaction were calculated based on the pseudo-first-order kinetic model. The choice of this model is a consequence of the excess quantity of sodium borohydride used during the reaction, which means that its concentration can be assumed constant. The proposed model can be described with Eq. (1): (1) ln C t / C 0 = ln C 0 – kt where C 0 and C t denote the initial concentration of 4-NP and the concentration at time t (min), and k (min− 1) denotes the rate constant.The oxidation reaction was carried out in a three-neck round-bottom flask placed in a water bath. First, the catalyst (50 mg), phenol (50 mL of water solution at concentration 0.5 mmol) and 1 mL of 31 wt% hydrogen peroxide (H2O2) were loaded into the reactor. This mixture was stirred (800 rpm) at 60 °C for 4 h under a reflux condenser and then immediately cooled in an ice bath to stop the reaction. Then methanol was added to the mixture to quench the reaction and remove excess hydrogen peroxide. The HPLC-MS system was used to analyse the efficiency of oxidation. The same starting conditions were used for the oxidation of 4-chlorophenoxyacetic acid and methylchlorophenoxypropionic acid. Due to the exothermic nature of this reaction, the concentration of reagents was kept low. Moreover, the agitation speed, reaction volume and amount of catalyst were chosen to avoid external diffusion problems.LC analysis was performed using the UltiMate 3000 RSLC chromatographic system from Dionex (Sunnyvale, CA, USA). 5 μL samples were injected into a Hypersil Gold C18 RP analytical column (100 × 2.1 mm × 1.9 μm) (Thermo Scientific, USA). In the typical procedure, the column was kept at 35 °C, and the mobile phase was 5 mM ammonium acetate in water (A) and methanol (B), at a flow rate of 0.2 mL/min. The following gradients were used: 0 min 80% B, 5 min 100% B for phenol, and 0 min 50% B, 5 min 100% B for 4-CPA and MCPP. The LC system was connected to an API 4000 QTRAP triple quadrupole mass spectrometer (AB Sciex, Foster City, CA, USA). Compounds were determined by electrospray ionization mass spectrometry (ESI-MS) in negative ion mode. The analyte was detected using the following settings for the ion source and mass spectrometer: curtain gas 10 psi, nebulizer gas 40 psi, auxiliary gas 40 psi, temperature 200 °C, ion spray voltage −4500 V, collision gas set to medium. For more details, see Supplementary Note 1.The evaluation of physicochemical properties is an important step in describing a new catalytically active material. The XRD patterns of the functionalized catalysts and of the carbonized spongin-based supports before functionalization are shown in Fig. 1 .As shown in Fig. 1, the XRD patterns exhibit diffractions characteristic for different phases of nickel, carbon, and silicon dioxide. The XRD patterns of the catalysts (Fig. 1A) do not show significant differences. The nickel phase is represented by the three broad diffraction peaks at 12°, 34° and 60°, which can be indexed respectively to Ni(OH)2 (001), Ni(OH)2 (110) and Ni(OH)2 (003) (JCPDS no. 1011134). The formation of a nickel oxide phase is confirmed by the presence of diffraction peaks derived from the NiO (220) phase at 46° (JCPDS no. 1010095). A small but well visible peak, characteristic for the Ni (111) phase, appeared at 43°, and a peak for Ni (220) at 72° (JCPDS no. 9008509). The diffractogram of the catalyst also contains a broad peak characteristic for C(002) (JCPDS no. 9011577). However, the intensity of this peak is significantly lower than for the corresponding phase in the XRD spectra of the pure supports (Fig. 1B). Also, features derived from various forms of silica (Fig. 1B) are not visible in the XRD pattern of the catalysts. This seems to be a result of the modification process, where the carbonized fibres are covered with a metal-containing phase. The strong effect of nickel atoms and the low content of silica compounds results in a lack of silica features in the XRD diffractogram of the catalysts.The chemical reduction of nickel ions adsorbed onto the surface of carbonaceous scaffolds is an efficient method of functionalization and is well described in the literature (Chen et al., 2017; Kuang et al., 2001; Sahiner et al., 2010). The SEM + EDS analysis provides information about the surface morphology and chemistry of the examined samples (results are shown in Fig. 2 ).From the Ni mapping presented in Fig. 3 , it is apparent that the metal-containing phase evenly covers all of the prepared catalysts. The highest efficiency of nickel functionalization was achieved for the NiO/Ni(OH)2/Ni_600 catalyst (26.01% mass of Ni), and the lowest for the sample carbonized at 500 °C (15.19% mass of Ni). The variation in the amount of nickel on the surface of the carbonized materials may be explained by the different ability of the carbonized scaffolds to bind nickel ions during the modification process. As has been described elsewhere (Petrenko et al., 2019; Żółtowska et al., 2021), the increase in carbonization temperature results in the formation of bio‑carbons with higher carbon content and decreasing content of other heteroatoms, including sulphur and nitrogen, among others. Consequently, the use as supports of bio‑carbons obtained at different carbonization temperatures leads to the formation of composites with different nickel loading. Owing to this fact, and referring to the results obtained in previous studies (Petrenko et al., 2019; Żółtowska et al., 2021), a broad statement can be made that the affinity of nickel ions towards the surface of bio‑carbons increases with the content of carbon and the level of graphitization of the carbonized material.Moreover, XRF analysis was performed to evaluate the elemental composition. The results provide evidence of traces of bromine and iron within the structure of the prepared materials; however, the content of these elements is low. Thus, it is assumed that their presence has no significant effect on the catalytic ability; rather, these elements participate in increasing the diversity of surface functional groups. A detailed discussion is given in Supplementary Note 2, and information regarding the surface functional groups is presented in Supplementary Note 5.It should be noted that the presence of iron (Fe) and aluminium (Al) is linked to the natural origin of the spongin-based scaffolds (Szatkowski et al., 2017; Jesionowski et al., 2018), while the silicon and calcium, according to previous research (Petrenko et al., 2019), are internal elements of the spongin skeleton. As is shown, the applied purification treatment with HCl acid results in the complete removal of calcium; however, it does not lead to the total removal of silicon species. Thus, carbonized spongin-based scaffolds appear to consist of a naturally occurring composite containing carbon, oxygen, nitrogen, sulphur and silicon traces, aluminium, and iron. For catalytic purposes, the existence of small amounts of silicon dioxide or alumina (less than 1 wt%) does not exclude the use of the bio‑carbons as supports. Moreover, the presence of heteroatoms such as sulphur, nitrogen, and iron in the catalyst structure may enhance its catalytic properties (Moosapour Siahkalroudi et al., 2021; Wang et al., 2021).The results of higher-resolution microscopy analysis (SEM and TEM) of the obtained composites are shown in Fig. 3. SEM images of the spongin-based scaffolds before carbonization and functionalization are compared in Supplementary Note 3.The results of scanning electron microscopy analysis provide evidence that the functionalization of the carbonaceous supports through reduction of adsorbed nickel particles is an effective method. Consequently, the naturally prefabricated, three-dimensional scaffolds are tightly covered with the metal-containing phase, which forms particular structures. Interestingly, the morphology of the metal-containing phase varies with the support used. For the NiO/Ni(OH)2/Ni_400 material, the metal-containing phase creates needle-like structures with a length of around 3–5 μm (Fig. 3A). For NiO/Ni(OH)2/Ni_500, spherical agglomerates of the metal-containing phase are visible (Fig. 3C), while the surface of NiO/Ni(OH)2/Ni_600 is again characterized by the presence of needle-like structures (Fig. 3E), although these structures are thicker and longer than those observed on the fibres of NiO/Ni(OH)2/Ni_400. These interesting differences in the morphological structure of the catalysts cannot be related to the method of synthesis, because all supports were treated in the same way. A possible explanation of these variations may be differences in the loading of the nickel-containing phase; as was mentioned previously, the material prepared using the carbonized support obtained at the highest carbonization temperature seems to have a superior ability to bind the nickel species, and thus the agglomerates formed are larger than in the case of the other materials. Consequently, differences in surface chemistry, such as graphitization level and carbon content, may affect the efficiency of the functionalization process and indirectly the final structure of the metal-containing phase (Petrenko et al., 2019). The influence of surface properties on the loading of the metal-containing phase, as observed in this study, is a well-known phenomenon sufficiently described in other works (Alijani et al., 2021; Peng et al., 2021).The most promising aspect of the application of carbonized spongin-based scaffolds as a support for the metal-containing phase is their three-dimensional fibrous architecture. The SEM images show a unique structure of interlaced fibres, which usually form channels with diverse shapes: triangle-like, rectangular-like, pentagonal-like and hexagonal-like, with sizes ranging from 1 to 300 μm. These structures are well visible on the SEM images (for comparison, see Fig. 2 and Supplementary Note 3). Therefore, it should be emphasized that the approach proposed in this paper, where the spongin-based scaffold was used as a bio-template together with a simple method of functionalization, led to the obtaining of desirable 3D structures.The TEM images in Fig. 3B, D, F show the metal-containing phase deposited on the bio‑carbon fibres. It can be assumed that in nanoscale, the structure of the metal-containing phase is similar and consists of thin sheets. Moreover, it seems that the modification process results in thick aggregates with irregular structure, forming the metal-containing layer on the surface of the carbonized bio‑carbon (Huang et al., 2013). Additionally, in Fig. 3D, some contaminants – derived from silicon dioxide – are marked with arrows.XPS analysis was carried out to evaluate in detail the surface composition of the prepared catalysts. The spectra obtained are shown in Fig. 4 .The Gaussian fitting method was used for a comprehensive analysis of the oxidation state of the nickel as well as the contributions of oxygen and carbon. In the case of all catalysts, the Ni 2p core-level spectra show two intense peaks at around 855.5 and 871 eV, attributed to Ni 2p3/2 and Ni 2p1/2 respectively, with corresponding satellites at around 861.3 and 879.7 eV, characteristic for Ni2+ (An et al., 2014; Cheng et al., 2017; Zhou et al., 2017). The core–shell peaks are attributed to NiO bonds, in this case associated with the hydroxide, as was confirmed in XRD analysis. The XPS spectra do not show any features corresponding to the NiO phase because its characteristic peaks appear at lower binding energies. These results might be related to additional atmospheric moisture adsorption. The O 1s spectra exhibit three oxygen contributions, labelled O1, O2 and O3. The O1 peak, derived from O–O–C bonds, is located at 534.1 eV (peak area 27.74%) for the NiO/Ni(OH)2/Ni_400 catalyst, 534.0 eV (peak area 10.35%) for NiO/Ni(OH)2/Ni_500 and 534.2 eV (peak area 8.36%) for NiO/Ni(OH)2/Ni_600. Its existence indicates that CO2 molecules were adsorbed on the surface of each catalyst. The O2 peak is commonly ascribed to physi-/chemisorbed water within the material's interface. The O3 feature, at 531.9 eV (peak area 35.04%) for NiO/Ni(OH)2/Ni_400, 531.3 eV (peak area 44.87%) for NiO/Ni(OH)2/Ni_500 and 534.5 eV (peak area 51.48%) for NiO/Ni(OH)2/Ni_600, is characteristic for metal–oxygen bonds (Payne et al., 2012; Weidler et al., 2017). A comparison of the C 1s spectra is given in Supplementary Note 4.The presented spectra do not differ significantly in terms of the surface composition, but show differences in the contents of various elements. Such results are not surprising considering the catalyst preparation method. Nevertheless, the XPS results show that NiO/Ni(OH)2/Ni_400 has the highest amount of CO2 adsorbed on the surface and produces more intense satellite peaks than the NiO/Ni(OH)2/Ni_500 and NiO/Ni(OH)2/Ni_600 catalysts.Further determination of surface functional groups was performed using FTIR spectroscopy. It was proven that the prepared materials contain various functional groups, including hydroxyl, amino and sulfoxide groups, among others. Besides, the formation of NiO groups is indicated (for detailed investigation, see Supplementary Note 5). The effect of pH on the zeta potential was also evaluated to investigate the electrokinetic behaviour of the prepared composites. The results indicate that the contents of nickel species and the electron releasing groups impact the value of the isoelectric point. Consequently, the NiO/Ni(OH)2/Ni_600 material, which has the highest content of nickel moieties, also has the highest IEP. A detailed discussion of the electrokinetic behaviour of the prepared materials is given in Supplementary Note 6. Fig. 5 presents the porous structure parameters, examined using the low-temperature nitrogen sorption technique.The nitrogen sorption capacity is highest for the NiO/Ni(OH)2/Ni_500 catalyst. It is slightly lower for NiO/Ni(OH)2/Ni_400 and approximately three times lower for NiO/Ni(OH)2/Ni_600. These results correspond to the decrease in the calculated BET surface area. The sorption isotherms of NiO/Ni(OH)2/Ni_400 and NiO/Ni(OH)2/Ni_500 can be classified as type IV isotherms. The well-visible hysteresis loop suggests that these materials have a mesoporous structure, with pore condensation at high pressure. The NiO/Ni(OH)2/Ni_400 catalyst has the largest pore size among the prepared materials, and NiO/Ni(OH)2/Ni_500 has the highest BET surface area. The isotherms of the NiO/Ni(OH)2/Ni_600 catalyst can be classified as type II, typical for low-porous materials containing both macropores and mesopores but no micropores. This result agrees with the finding that this sample has the lowest surface area.However, it must be recalled that the method of catalyst preparation results in coverage of the fibres of the carbonized spongin-based scaffolds with the metal-containing phase. For this reason, the surface area of the prepared materials does not exceed 10 m2/g. As a result, the three-dimensional fibrous support structure with channels of diverse shape may provide good accessibility to the catalyst surface.The catalytic properties of the prepared materials were first tested in the reduction reaction of 4-nitrophenol. This reaction is widely used as a determinant of the catalytic activity of heterogeneous materials, whether or not involving a support (Emam et al., 2017; Grzeschik et al., 2020; Hu et al., 2015; Strachan et al., 2020). Moreover, 4-nitrophenol is an important but toxic substrate used in the production of various drugs, dyes, and pesticides. Therefore, the evaluation of a fast and straightforward method of converting this compound to a functional product is potentially of great benefit from the engineering and environmental point of view (Blaser, 2006).The UV–Vis spectra (Fig. 6 ) measured during the reaction in the presence of the prepared catalysts show that all of the tested materials exhibit catalytic ability in the reduction of 4-nitrophenol. The reaction time varies between 4 and 6 min, whereas without the catalyst this reaction does not occur, as explained in Supplementary Note 7. The addition of sodium borohydride led to an increase in the reaction mixture pH from 7 to 10; in such conditions, the functional groups on the surface of the catalysts are deprotonated (see zeta potential measurements in Supplementary Note 4) and are thus negatively charged. Despite the negatively charged surface of the catalyst, only for the NiO/Ni(OH)2/Ni_400 material was an induction period observed; this is apparently related to charging transformation of the surface of the catalyst before reaction (Khalavka et al., 2009; Mahmoud and El-Sayed, 2011; Sarkar et al., 2011; Wu et al., 2011). In the case of the NiO/Ni(OH)2/Ni_500 and NiO/Ni(OH)2/Ni_600 catalysts, different behaviour was observed in the reduction of 4-nitrophenol. For the first-mentioned material, a slow reduction of the peak intensity assigned to the 4-nitrophenolane anion is visible, while for NiO/Ni(OH)2/Ni_600, the peak intensity decreases rapidly after 60 s of the reaction, and the reaction is completed after 4 min. This behaviour may be related to a different path of reduction. For these materials, surface-mediated hydrogen transfer seems to play a leading role during the reduction of 4-nitrophenol.Because the reducer, sodium borohydride, was applied in significant excess, the kinetics of the reaction were calculated based on the pseudo-first-order kinetic model (Table 2 ) (Jiang et al., 2012; Sahiner et al., 2010; Yang et al., 2019; Zhu et al., 2011).The highest reaction rate constant was calculated for the NiO/Ni(OH)2/Ni_400 catalyst. However, the time of reaction is similar for each catalyst used, although the value of the rate constant varies significantly. Even though the highest reaction rate was obtained for the NiO/Ni(OH)2/Ni_400 catalyst, the correlation coefficient took the lowest value. The variations in catalytic activity seem to be related to the structure of the prepared catalyst. The NiO/Ni(OH)2/Ni_400 material has a lower content of nickel phases than NiO/Ni(OH)2/Ni_600, although the needle-like structures observed on the surface of both catalysts are thinner and shorter in the case of NiO/Ni(OH)2/Ni_400 (see Fig. 4A). This may increase the contact area between the reagents and lead to an increase in the diffusion of substrates (Pushkarev et al., 2012; Wang et al., 2014).Additionally, the NiO/Ni(OH)2/Ni_400 material is characterized by the presence of iron and aluminium, among others, which may create additional active centres of the catalyst. The fact that NiO/Ni(OH)2/Ni_500 has the lowest catalytic activity may therefore be explained by the different morphological structure of the metal-containing phase, together with the lowest content of nickel phases. Since it had the highest rate constant (k), only NiO/Ni(OH)2/Ni_400 was considered for further evaluation of reusability. This material was repeatedly used five times in the catalytic reduction reaction to assess its stability. After each cycle, the catalyst was recovered by filtration, washed several times with deionized water, and dried in a dryer at temperature 60 °C. The calculated rate constants (from the pseudo-first-order model) are compared in Table 2.Comparison of these data shows a decrease in the rate constant and an increase in the reaction time with each catalytic run, probably because of loss of activity due to the blocking of active sites of the catalyst. On the other hand, it should be noted that after the fifth cycle, the rate constant is still high – comparable to the rate constant obtained for the NiO/Ni(OH)2/Ni_500 catalyst in its first cycle. Such results prove that the formation of a metal-containing phase on the carbonaceous fibrous support provides better stability for the catalytically active phase (Dhokale et al., 2014; Gu et al., 2014; Kongarapu et al., 2017).The mechanism of 4-nitrophenol reduction is exhaustively described in the literature (Khalavka et al., 2009; Mahmoud and El-Sayed, 2011; Sahiner et al., 2010; Sarkar et al., 2011; Wu et al., 2011). In the description of new catalysts, it is essential to note which component can be assumed as the active site of the catalyst. Based on the available literature (Wunder et al., 2010, 2011) and the present results, it can be assumed that the reduction occurs by way of sodium borohydride decomposition on the crystallites of nickel moieties. In the next stage, electron transfer occurs from BH4 − molecules towards 4-nitrophenoloane anions via Ni(OH)2/NiO/Ni grains, which play the role of electron carriers. In view of the previously described activity of carbonized spongin-based scaffolds in the reduction of 4-nitrophenol, this study provides evidence that modification with nickel hydroxide, nickel oxide and nickel leads to enhanced catalytic activity, while the carbonized spongin-based scaffolds and the metal-containing phase act synergistically during the reduction reaction. Besides, the three-dimensional, fibrous-like morphology with open channels enhances the diffusion of substrates towards the surface of the catalyst. The promising results concerning the catalytic activity of the prepared materials further encouraged us to evaluate their effectiveness as catalysts in oxidation reactions.Phenol, methylchlorophenoxypropionic acid (MCPP) and 4-chlorophenoxyacetic acid (4-CPA) were used as substrates for the catalytic oxidation reaction. These compounds are commonly used in the production of drugs (phenol) and pesticides (MCPP and 4-CPA). Their presence in water streams has been proved in several studies (Abdel Rahman and Hung, 2020; Somasundaram et al., 2018; Wang et al., 2020; Wang et al., 2016a), as have their toxic and bioaccumulation effects on the environment (Benny and Chakraborty, 2020; Piotrowska et al., 2017). The oxidation was carried out in a water solution at temperature 60 °C in the presence of hydrogen peroxide as an oxidizing agent, for a time of 4 h. The oxidation efficiency was calculated based on measurement of the concentration of the substrate after the reaction, using the calibration curve method. The obtained oxidation efficiencies are shown in Fig. 7 .The above results show the good catalytic ability of NiO/Ni(OH)2/Ni_400. Application of this catalyst at acidic pH leads to full oxidation of MCPP and 4-CPA (oxidation yield more than 99%) and sufficient oxidation of phenol (oxidation yield 80%) (Fig. 7A). Without a catalyst, the yield of oxidation is not higher than 10%. This proves that the presence of a catalyst is essential for the oxidation of phenolic compounds. Interestingly, the NiO/Ni(OH)2/Ni_500 and NiO/Ni(OH)2/Ni_600 catalysts presented significantly weaker catalytic properties. The yield of oxidation of any substrate was not higher than 33%. In contrast to the results for the reduction of 4-nitrophenol, where the difference in catalytic activity was relatively small, the same materials used in oxidation reactions differed significantly in catalytic activity. The favourable catalytic activity of NiO/Ni(OH)2/Ni_400 may be a consequence of the morphology of the metal-containing phase forming the well-dispersed needle-like structure, the fact that it has the largest pore size among the prepared materials, as well as the chemical composition of the bio‑carbon. The material obtained at the lowest carbonization temperature exhibits the presence of various heteroatoms, as described in other studies (Petrenko et al., 2019; Żółtowska et al., 2021). Owing to the chemical composition of the prepared composites, the NiO/Ni(OH)2/Ni_400 material has the most diverse elemental composition, with the highest percentage of oxygen, iron, bromine, and iodine. Thus, it can be concluded that this material, thanks to the diversification of its surface functional groups, can create various active centres, enhancing its catalytic activity. However, the nickel-containing phase seems to be the major player in the tailoring of the catalytic properties. Thus, it can be assumed that the morphology of the prepared materials and the nickel content may be the crucial factors impacting the activity of the catalyst. Consequently, as can be seen in Fig. 3, an excessive amount of nickel leads to the formation of larger nickel-containing clusters, but with lower surface area. Related to this assumption, BET data showed a higher surface area for the NiO/Ni(OH)2/Ni_500 material. For this reason, the NiO/Ni(OH)2/Ni_400 surface morphology seems to be the best suited for catalytic purposes.In the next step, the phenolic compounds were oxidized at alkaline pH (Fig. 7B). The results show that increasing the pH negatively affects the efficiency of oxidation of phenol, MCPP, and 4-CPA. (NiO/Ni(OH)2/Ni_500 was excluded from testing because it had the lowest catalytic activity.) These results may be explained by the fact that in a water solution with pH in the range 2–3, decomposition of hydrogen peroxide produces hydrogen radicals (OH ), which play the leading role in the oxidation of organic compounds (Xing et al., 2018; Zhang et al., 2019). At higher pH, the main product of hydrogen peroxide decomposition is hydroxyl ions. Therefore the concentration of radicals is significantly lower, which results in lower reaction efficiency. The significantly higher catalytic ability of NiO/Ni(OH)2/Ni_400 catalysts may be a result of their having the largest pore size among the materials, and the highest content of surface oxygen and the Ni2+ phase together with the presence of other heteroatoms (Fe, Al, S, among others). In consequence, the synergistic action of several factors impacts the final catalytic activity in oxidation reactions.The results of zeta potential measurements can provide further important information on the influence of pH. The pKa value was 9.94 for phenol and 3.56 and 3.75 for 4-CPA and MCPP respectively. It can be concluded that when the pH of the reaction is 3 the compounds are in a protonated state, with the surface groups of each catalyst positively charged. For reactions carried out at pH 8 the 4-CPA and MCPP molecules are deprotonated, while phenol is still in the protonated form, and the catalysts have negatively charged surface functional groups. It appears, therefore, that at acidic pH, the repulsive electrostatic interaction between the catalyst surface and the substrate molecules does not hamper the oxidation process significantly. Moreover, the prepared catalysts are involved in the formation of hydroxyl radicals (at pH 3) or hydroxyl ions (at pH 8), which attract molecules of the phenolic compounds. An increase in the pH of the reaction mixture, therefore, has a significant negative effect on the oxidation efficiency.As regards possible interactions between the substrates and catalysts, π-π interactions may be considered the most important. However, in the oxidation reactions, the catalyst acts via electron transfer. A more detailed examination of interactions between the substrates and catalysts will not be made here.To obtain a useful catalyst, not only high catalytic activity is essential. Such a material should also be stable over catalytic cycles. Thus, reusability experiments were carried out to evaluate the possibility of multiple application of the NiO/Ni(OH)2/Ni_400 material in repeated oxidation of phenol, MCPP, and 4-CPA.As the results (Fig. 8 ) show, an apparent decrease in oxidation efficiency is visible in the case of each compound. In the oxidation of MCPP and 4-CPA, the reduction of catalytic ability after the fifth run reaches 80%. The lowest decrease in catalytic activity is observed for the oxidation of phenol, where the catalyst retains 40% of its activity after the fifth catalytic run. Interestingly, the reaction with this compound gives the lowest yield in the first catalytic cycle. However, after the second, third, fourth, and fifth reaction runs, the calculated activity of NiO/Ni(OH)2/Ni_400 in the oxidation of phenol is higher than for both 4-CPA and MCPP after the same number of runs. The observed reduction in catalytic activity may be related to loss of catalyst mass during the recovery and washing process. However, deactivation of active sites by poisoning could not be excluded. This is the case especially for 4-CPA and MCPP, as the intermediate products of their oxidation contain chlorine, which can act as a poison on the catalyst surface. This assumption may explain the significant decrease in oxidation efficiency on repeated reuse (Mork and Norgard, 1976; Paquin et al., 2015).Because the mechanism of phenol oxidation is extensively discussed in the literature, it will not be evaluated in detail here. However, to investigate the mechanisms of 4-CPA and MCPP oxidation, MS spectra before and after oxidation were recorded (see Supplementary Note 8).Evolution in materials science and catalysis has led to work that merges two different topics: metals and biopolymers. There is an increasing number of reports on carbon materials derived from biomass. Most biopolymers (lignin, collagen, silk, cellulose, starch, chitin, chitosan, dextran, pectin, alginate, carrageenan) can be successfully used as precursors of carbon materials (Boury and Plumejeau, 2014; Lee et al., 2011, 2012; Qu et al., 2016; Q. Wang et al., 2016b; Zhao et al., 2016). Biopolymers of biological origin are of particular importance for obtaining carbon materials (Zhang et al., 2010). This is related to their chemical structure, which is rich in heteroatoms such as nitrogen or sulphur; the presence of these increases the diversity of surface functional groups. Therefore, these carbon materials are well suited as scaffolds for metal–carbon composites (Lee et al., 2011, 2012; Zhang et al., 2010).Among the many biopolymers commonly applied in the preparation of bio‑carbons, spongin-based scaffolds seem to be the most promising choice. Spongin belongs to the “collagen suprafamily” (Petrenko et al., 2019), and it is the main skeletal protein building the skeletons of sponges in the class Demospongiae, which can grow up to 70 cm. These spongin-based skeletons have been known since ancient times as natural (bath) or commercial sponges, and have been used for cosmetic or medical purposes. Their use for the preparation of bio‑carbons is economically feasible, because they are cultivated under marine farming conditions on a large scale worldwide, creating a sponge market worth more than US$20 million (Żółtowska et al., 2021). Moreover, in contrast to other biomass materials used to prepare bio‑carbons, which are fragile and can be applied only in the form of powder, these ready-to-use scaffolds preserve the unique structure of Demospongiae sponge skeletons even after carbonization at temperatures as high as 1200 °C. Carbonized spongin-based scaffolds are mechanically robust and can be used to prepare bio‑carbons with a strictly designed shape.Nevertheless, this study and other recently published work (Petrenko et al., 2019; Żółtowska et al., 2021) are just the beginning of efforts to discover the full potential of bio‑carbons derived from spongin-based scaffolds. Further research may focus on the atomistic simulation of bio‑carbons to provide an additional overview regarding the optimization of properties. Therefore, significant effort should be put into understanding the mechanism of transformation from organic precursor to inorganic carbon and the effect of additives on the structural and chemical properties of the resulting bio‑carbon. Moreover, knowledge of the mechanism of functionalization and effect of the bio‑carbon structure on the properties of the resulting composite may provide additional insight and represent a milestone in the development of efficient bioinspired materials. Besides, the detailed characterization of spent catalysts may be considered as a subject of future evaluation, encompassing changes in composite morphology, elemental content, BET surface area, leaching of the metal-containing phase, and deeper catalytic study, including the mechanistic aspects. Likewise, the heat conduction of modified bio‑carbon-based composites in catalytic applications is another potential area of research.Spongin-based scaffolds are indisputably promising materials for use for catalytic purposes. Their carbonization has been shown to lead to bio‑carbons with exceptional stability, robustness and physicochemical properties. The relatively simple functionalization method demonstrated in this study, which does not require the use of extreme synthesis conditions or toxic and expensive reagents, opens up new possibilities for the preparation of bio-inspired materials, which is in line with the philosophy of sustainable development.A spongin-based fibrous scaffold isolated from the marine demosponge Hippospongia communis was utilized as a precursor of a carbon support and subjected to carbonization (at various temperatures) followed by modification with nickel and nickel oxide. For the first time, a relatively simple method of functionalization of carbonized scaffolds was applied to obtain nickel-based bio‑carbon composites. Morphological and physicochemical analysis revealed moderate differences in the chemical nature of the prepared materials. It was shown that the temperature of carbonization influences the effectiveness of the modification process. However, this study mainly focused on the characterization of the prepared materials and their evaluation as potential catalysts for oxidation or reduction reactions of various phenolic compounds. The results confirm the promising activity of all of the materials in the reduction of 4-nitrophenol; the slight differences in the calculated rate constants were ascribed to the different nickel oxide contents in the tested catalysts. The mechanism of reduction was also predicted. In oxidation tests, the NiO/Ni(OH)2/Ni_400 catalyst showed excellent catalytic ability in the oxidation of MCPP and 4-CPA (99% oxidation efficiency at pH 3) and good activity in the oxidation of phenol (80% yield at pH 3). The use of hydrogen peroxide as an oxidizing agent eliminates the formation of additional toxic products of oxidant decomposition.This remarkable catalytic activity may be ascribed to (i) the diverse chemical composition of the nickel phase, consisting of nickel hydroxide Ni(OH)2, nickel oxide NiO and nickel(0), which creates various active centres of the catalysts; (ii) the even dispersion of the metal-containing phase on the fibres of the carbonized supports; and (iii) the unique needle-like morphology, which acts synergistically with the 3D structure of the supports to provide good diffusion and high contact area between the catalysts and reagents. This study provides evidence that spongin-based scaffolds can be utilized to produce a structured carbonaceous material that can be successfully functionalized with nickel moieties using a simple sorption–reduction approach. As a result, it becomes possible to produce unique composites based on bio‑carbon functionalized with nickel species, with impressive catalytic performance in the removal of emerging contaminants. Sonia Żółtowska: Conceptualization, Investigation, Writing – original draft, Writing – review & editing, Visualization. Zuzanna Bielan: Investigation. Joanna Zembrzuska: Investigation, Writing – review & editing. Katarzyna Siwińska-Ciesielczyk: Investigation. Adam Piasecki: Investigation, Resources. Anna Zielińska-Jurek: Investigation. Teofil Jesionowski: Supervision, Writing – review & editing, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The work was supported by the National Science Centre, Poland, project Etiuda no. 2019/32/T/ST8/00414 (S.Ż.), and by the Ministry of Education and Science, Poland (K.S.-C., T.J.). Sonia Żółtowska and Teofil Jesionowski would like to thank Professor Monika Mazik of TU Bergakademie Freiberg for assistance during the catalytic tests. Supplementary material Image 1 Supplementary data to this article can be found online at https://doi.org/10.1016/j.scitotenv.2021.148692.

Three different 3D fibrous-like NiO/Ni(OH)2/Ni‑carbonized spongin-based materials were prepared via a simple sorption–reduction method. Depending on the support used, the catalysts were composed of carbon, nickel oxide, nickel hydroxide and zero-valent nickel, with the surface content of the nickel-containing phase in the range 15.2–26.0 wt%. Catalytic studies showed promising activity in the oxidation of phenolic compounds in water and in the reduction of 4-nitrophenol. The oxidation efficiency depends on the substrate used and ranges from 80% for phenol at pH 2 to 99% for 4-chlorophenoxyacetic acid (4-CPA) and methylchlorophenoxypropionic acid (MCPP). In the reduction reaction, all catalysts exhibited superior activity, with rate constants in the range 0.648–1.022 min- 1. The work also includes a detailed investigation of reusability and kinetic studies.

The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study.PE-based polymers represent one of the most important materials class in our daily life due to their superb mechanical properties, chemical stability and low production costs [1]. In this field, the coordination-insertion polymerization via transition-metal complexes plays a crucial role in synthesis of polyethylene materials. Compared to the early-transition metal complexes, the late-transition metal-based analogues exhibit superior performance in such catalyzed polymerization [2–4]. These complexes are tolerant to moisture and air, enabling a long-term storage and delivery without inert atmosphere protection. Furthermore, due to the low oxophilicity, they are even able to perform the ethylene copolymerization with polar monomers, yielding PE copolymers containing polar groups [5–8]. Compared to the conventional multi-site heterogeneous catalysts, the single-site catalytic metal allows for the production of PE with narrow molecular weight distributions (PDI). The polymeric microstructures can be modulated via a modification of the complexes structures, leading to a variation in the molecular weight (Mw), PDI, melting transition, density, crystallinity, and branching density. The macroscopic characteristics such as mechanics, surface wettability, chemical and thermal stability, optics, and viscosity of PE can indirectly be altered by suitable modifications of the complex structures [9,10]. Thus, late-transition metal-catalyzed polymerization can be considered as one of the most promising approaches for future industrial olefin polymerization [2,10–13].The a-diimine Ni/Pd complexes are one of the best-studied and long-examined late-transition metal-based catalysts for ethylene polymerization. The initial work (A in Fig. 1 ) of Brookhart et al. demonstrated the high activity (107 g of PE (mol of Ni)−1 h−1) and great potential of such catalysts [14,15]. One of the major advantages of these complexes is the ability to produce thermoplastic-elastomers (TPEs) like branched PE via chain-walking mechanism, by use of ethylene as the only reactive monomer [16–18]. The chain walking process involves a competition between the chain growth and chain walking during the monomer insertion process. Cationic alkyl-metal active species migrate along the polyethylene backbone via rapid β-H elimination and reinsertion, leading to the formation of branched structures [19,20]. As reported previously, the remote substitutions, N-aryl groups and backbones are considered as the main factors that control the catalytic behavior of the a-diimine metal complexes. For example, the steric accumulation on the N-aryl group (e.g. B in Fig. 1) significantly changes the catalytic performance of the Ni center [21–23]. These bulky substituents (especially in ortho position of the N-aryl) act as a shield and offer confined space for monomer insertion, which slows down both the chain-transfer and the chain-walking process. A limited coordination space around the metal center favors a chain-growth mechanism, leading to a more linear and high molecular weight PE [21,24–31]. Additionally, electron-withdrawing groups could considerably enhance the catalytic performance of the cationic alkyl-metal center – the fluorine effect is of great importance in the application of the olefin polymerization, as the fluorine atom has an electron-withdrawing character due to its inductive effects [32–35]. In contrast to saturated electron-withdrawing groups, the fluorine exhibits π interactions with the aromatic rings. The interaction between the catalytic metal centers with fluorine atoms has been described in previous work, where it was shown to have great influences on the catalytic behavior in polymerization process [26,36–39]. The side-arm (i.e. steric and electronic) effects of the ligand moiety create distinctive coordination environments around the catalytic metal center, leading to important variety of polymer properties [2,7,40–44].Currently, numerous studies related to the synthesis of the new a-diimine Ni (II) complexes have been reported, where the symmetric arrangement of bulky N-aryls has been the focus [20,45]. However, the catalytic activity is decreased by the limited monomer access into the confined coordination space of the active metal center. The ligands containing bulky groups on both N-aryls tend to exhibit lower (1 or 2 order of magnitudes) catalytic activity than A in Fig. 1 [21,46]. As reported previously, it is difficult to simultaneously achieve both the high catalytic activity and high molecular weights. The unsymmetrical a-diimine Ni complexes bearing different anilines have been reported since the initial study of benzhydryl-substituted ligands [47,48]. The dibenzhydryl substitutions as the steric groups were incorporated on the first N-aryl, while the second N-aryl moiety was maintained as the less bulky (methyl group) substituent. The catalytic properties of such unsymmetrical a-diimine Ni complexes can be modified by the structural tunes of the solo N-aryl substitution with typical electronic and steric features [20,30,48,49].In this work, a simplified synthetic methodology was followed to synthesize a series of new unsymmetrical a-diimine Ni complexes (Ni-OH, Ni-FOH, Ni-PhOH and Ni-PhFOH in Fig. 1). The aim of our catalyst design was to realize the typical modulation of ethylene monomer insertion and chain-walking process while still maintaining high activity during the ethylene polymerization. It can be considered as a balancing combination of high activity and suppressed β-H elimination. Different steric enhancements (dibenzhydryl substitutions) on the para/ortho-N-aryls offer various environments to the coordination-insertion process. In order to systematically check the fluorine effect on the catalytic performance during the ethylene polymerization, the fluorine moiety on the N-aryls was synthesized and incorporated on these Ni complexes. The Cl atoms are selected as the affiliation of metal center rather than Br, leading to the formation of the a-diimine Ni dichlorides. The various polymerization conditions were optimized in detail, including co-catalyst, Al/Ni ratio, polymerization temperature, lifetime, and ethylene pressure. This facile strategy is expected to realize the further modifications of the catalytic behaviors of the proposed a-diimine Ni complexes and polymeric microstructure, such as the branches, Mw, PDI and melting transitions. Additionally, the design of ligands involved the incorporation of hydroxyl group, which could then be used to tether the Ni complexes to solid surfaces for gas- and slurry-phase polymerization. Although the hydroxyl group is very electron donating and active to the activated metal center, no reduction in the catalytic performance of the a-diimine Ni complexes was observed.The synthesis of the a-diimine ligands and Ni (II) complexes was carried out under air. Contrary to this, ethylene polymerizations were performed under Schlenk techniques and inert argon atmosphere protection. All the solvents and starting materials were purchased from Sigma-Aldrich and Chemie-Brunschwig. The 1H, 13C and 19F NMR spectrum of a-diimine ligands and Ni complexes were recorded on the Bruker Avance III 400 NMR spectrometer. Elemental analysis was carried out using Elementar-UNICUBE analyzer. Mw and PDI, branching density were determined by a 1260 infinity ii HT-GPC at 160 °C with 1,2,4-trichlorobenzene as the solvent against PS standards. The ESI-HRMS results were analyzed by an UHR-TOF BRUCKER Doltonik (Bremen, Germany) maXis with an ESI-quadrupole time-of flight (qToF) mass spectrometer. The melting points were determined by the differential scanning calorimeter (DSC, 214-Polyma) with second-heating-scan curves. 1H and 13C NMR spectra of the polyethylene were measured by using an ARX-300 spectrometer at 140 °C in bromobenzene- d 6. L-OH A solution of 3-((2,6-dibenzhydryl-4-methylphenyl)imino)butan-2-one (1.01 g, 2.0 mmol) and 4-amino-3,5-dimethylphenol (0.33 g, 2.4 mmol) in toluene (50 mL) with a catalytic amount of para-toluenesulfonic acid (0.035 g, 0.2 mmol) was stirred under refluxed over 36 h. Afterwards, the mixture was cooled down to the room temperature and concentrated under reduced pressure by rotary evaporator. The rest solid was purified by silica column chromatography (Heptane 3/1 Ethyl acetate) to afford L-OH as yellow crystalline solid (0.38 g, 30.0 %). 1H NMR (400.2 MHz, DMSO‑d 6): δ(ppm) 0.67 (s, 3H), 1.78 (s, 3H), 1.86 (s, 6H), 2.11 (s, 3H), 5.18(s, 2H), 6.49 (s, 2H), 6.61 (s, 2H), 7.00 (d, 4H, J = 8 Hz), 7.07 (d, 4H, J = 8 Hz), 7.17–7.30 (m, 12H), 8.86 (s, 1H). 13C NMR (100.6 MHz, DMSO‑d 6): δ(ppm) 13.9, 15.4, 15.5, 17.6, 21.0, 22.1, 28.3, 31.2, 51.7, 114.4, 125.0, 126.2, 126.2, 128.0, 128.2, 128.5, 129.0, 129.3, 130.9, 131.1, 140.2, 142.1, 143.1, 145.2, 152.8, 167.6, 169.6. Anal. Calcd for C45H42N2O (626.84): C, 86.22; H, 6.75; N, 4.47. Found: C, 86.22; H, 7.01 N, 4.17. L-FOH Based on the similar procedure described for L-OH, L-FOH was synthesized via the reaction of 3-((2,6-bis(bis(4-fluorophenyl)methyl)-4-methylphenyl)imino)butan-2-one (3.48 g, 6 mmol) and 4-amino-3,5-dimethylphenol (1.28 g, 6 mmol) with a catalytic amount of para-toluenesulfonic acid (0.105 g, 0.6 mmol) in toluene. L-FOH was isolated as yellow crystalline solid (1.85 g, 39.8 %). 1H NMR (400.2 MHz, DMSO‑d 6): 1H NMR (400.2 MHz, DMSO‑d 6): δ(ppm) 0.84 (s, 3H), 1.76 (s, 3H), 1.85 (s, 6H), 2.13 (s, 3H), 5.21(s, 2H), 6.49 (s, 2H), 6.58 (s, 2H), 6.97–7.13 (m, 16H), 8.86 (s, 1H).13C NMR (100.6 MHz, DMSO- d 6): δ(ppm) 15.5, 15.7, 17.7, 21.0, 50.0, 54.9, 114.4, 114.9, 115.1, 115.2, 115.4, 125.1, 128.0, 130.7, 130.8, 131.0, 131.1, 131.4, 138.2, 138.3, 139.0, 139.1, 140.1, 152.9, 159.4, 159.5, 161.8, 161.9, 167.5, 169.4. 19F NMR (376.5 MHz, DMSO- d 6): δ(ppm) −116.7, −116.4. Anal. Calcd for C45H38F4N2O (698.81): C, 77.35; H, 5.48; N, 4.01. Found: C, 77.23; H, 5.56; N, 3.63. L-PhOH Based on the similar procedure described for L-OH, L-PhOH was synthesized via the reaction of 3-((2,4,6-tribenzhydrylphenyl)imino)butan-2-one (3.33 g, 5 mmol) and 4-amino-3,5-dimethylphenol (0.69 g, 5 mmol) with a catalytic amount of para-toluenesulfonic acid (0.087 g, 0.5 mmol) in toluene. L-PhOH was isolated as yellow crystalline solid (1.21 g, 31.0 %). 1H NMR (400.2 MHz, DMSO- d 6): δ(ppm) 0.72 (s, 3H), 1.77 (s, 3H), 1.85 (s, 6H), 5.16 (s, 2H), 5.39 (s, 1H), 6.49 (s, 2H), 6.60 (s, 2H), 6.87–6.97 (m, 12H), 7.11–7.23 (m, 18H) 8.85 (s, 1H). 13C NMR (100.6 MHz, DMSO- d 6): δ(ppm) 16.0, 16.1, 18.1, 52.2, 55.6, 114.8, 125.4, 126.4, 126.5, 126.7, 128.5, 128.6, 128.9, 129.0, 129.2, 129.2, 129.6, 131.4, 137.7, 140.7, 142.5, 143.4, 144.6, 146.1, 153.4, 168.1, 170.1. Anal. Calcd for C57H50N2O (779.04): C, 87.88; H, 6.47; N, 3.60. Found: C, 88.04; H, 6.48; N, 3.68. L-PhFOH Based on the similar procedure described for L-OH, L-PhFOH was synthesized via the reaction of 3-((2,4,6-tris(bis(4-fluorophenyl)methyl)phenyl)imino)butan-2-one (1.54 g, 2 mmol) and 4-amino-3,5-dimethylphenol (0.28 g, 2 mmol) with a catalytic amount of para-toluenesulfonic acid (0.035 g, 0.2 mmol) in toluene. L-PhFOH was isolated as yellow crystalline solid (0.79 g, 44.7 %). 1H NMR (400.2 MHz, DMSO- d 6): δ(ppm) 0.90 (s, 3H), 1.71 (s, 3H), 1.84 (s, 6H), 5.19 (s, 2H), 5.49 (s, 1H), 6.37 (s, 2H), 6.49 (s, 2H), 6.87–6.90 (m, 4H), 6.95–7.08 (m, 18H) 8.86 (s, 1H). 13C NMR (100.6 MHz, DMSO- d 6): δ(ppm) 14.4, 15.7, 16.4, 18.1, 22.5, 28.8, 31.7, 50.5, 53.5, 114.8, 115.2, 115.3, 115.5, 115.5, 115.6, 115.8, 125.5, 128.9, 130.8, 130.9, 130.9, 131.0, 131.2, 131.3, 131.4, 137.9, 138.5, 138.5, 139.2, 139.2, 140.4, 140.4, 140.5, 146.0, 153.4, 159.9, 159.9, 162.3, 162.4, 167.9, 169.7. 19F NMR (376.5 MHz, DMSO- d 6): δ(ppm) −116.8, −116.3. Anal. Calcd for C57H44F6N2O (886.98): C, 77.19; H, 5.00; N, 3.16. Found: C, 77.39; H, 5.14; N, 3.09. Ni-OH NiCl2·6H2O (0.209 g 0.88 mmol) and L-OH (0.55 g, 0.88 mmol) were dissolved in DCM (10 mL) and EtOH (2 mL) and stirred at room temperature overnight. The mixture was concentrated under vacuum pump and wash with diethyl ether (20 mL) and heptane (10 mL). The precipitated compound was filtered and washed by the excess diethyl ether and heptane, affording Ni-OH as orange powder (0.44 g, 66.7 %). ESI-MS (positive-ion mode): m/z 719.2334 ([M – Cl]+). Calcd: m/z 719.2316. Anal. Calcd for C45H42N2ONiCl2 (756.44): C, 71.45; H, 5.60; N, 3.70. Found: C, 71.33; H, 5.27; N, 3.71. Ni-FOH Based on the similar procedure and molar ratios described for Ni-OH, Ni-FOH was isolated as orange powder (0.51 g, 62.2 %). ESI-MS (positive-ion mode): m/z 791.1930 ([M – Cl]+). Calcd: m/z 791.1957. Anal. Calcd for C45H38F4N2ONiCl2 (828.40): C, 65.25; H, 4.62; N, 3.38. Found: C, 65.36; H, 4.66; N, 3.21. Ni-PhOH Based on the similar procedure and molar ratios described for Ni-OH, Ni-PhOH was synthesized as orange powder (0.25 g, 92.5 %). ESI-MS (positive-ion mode): m/z 871.3008 ([M – Cl]+). Calcd: m/z 871.2960. Anal. Calcd for C57H50N2ONiCl2 (908.63): C, 75.35; H, 5.55; N, 3.08. Found: C, 75.00; H, 5.78; N, 2.81. Ni-PhFOH Based on the similar procedure and molar ratios described for Ni-OH, Ni-PhFOH was synthesized as yellow powder (0.66 g, 81.5 %). ESI-MS (positive-ion mode): m/z 472.1342 ([M – 2Cl]2+). Calcd: m/z 472.1350. Anal. Calcd for C57H44F6N2ONiCl2 (1016.58): C, 67.35; H, 4.36; N, 2.76. Found: C, 67.06; H, 4.39; N, 2.52. Ni-OH: Single crystals of C45H42Cl2N2NiO [Ni-OH] were crystallized. A suitable crystal was selected and mounted on a STOE IPDS 2 diffractometer. The crystal was kept at 173(2) K during data collection. Using Olex2, the structure was solved with the SIR2008 structure solution program using Direct Methods and refined with the SHELXL refinement package using Least Squares minimization [50–52]. Crystal Data for C45H42Cl2N2NiO (M = 756.41 g/mol): monoclinic, space group P21/c (no. 14), a = 18.926(3) Å, b = 9.4716(10) Å, c = 21.408(4) Å, β = 100.789(13)°, V = 3769.8(10) Å3, Z = 4, T = 173(2) K, μ(MoKα) = 0.694 mm−1, Dcalc = 1.333 g/cm3, 22,350 reflections measured (3.874° ≤ 2Θ ≤ 50.416°), 6702 unique (R int = 0.0791, Rsigma = 0.0541) which were used in all calculations. The final R 1 was 0.0438 (I > 2σ(I)) and wR 2 was 0.1274 (all data shown in Table S1.). Supplementary crystallographic data (CCDC 2202319) is available free of charge from The Cambridge Crystallographic Data Centre CCDC. Ni-FOH: Ni-FOH was characterized by the similar procedure for Ni-OH. Crystal Data for C45.5H39Cl3F4N2NiO (M = 870.84 g/mol): orthorhombic, non-centrosymmetric space group P21212 (no. 18) with Flack −0.01(1), a = 22.900(2) Å, b = 19.6803(17) Å, c = 10.5741(7) Å, V = 4765.5(7) Å3, Z = 4, T = 173(2) K, μ(MoKα) = 0.624 mm−1, Dcalc = 1.214 g/cm3, 30,791 reflections measured (2.728° ≤ 2Θ ≤ 50.368°), 8293 unique (R int = 0.0549, Rsigma = 0.0400) which were used in all calculations. The final R 1 was 0.0468 (I > 2σ(I)) and wR 2 was 0.1448 (all data shown in Table S1.). Supplementary crystallographic data (CCDC 2202320) is available free of charge from The Cambridge Crystallographic Data Centre CCDC. Ni-PhFOH: Ni-PhFOH was characterized by the similar procedure for Ni-OH. Crystal Data for C58H46Cl4F6N2NiO (M = 1101.48 g/mol): monoclinic, non-centrosymmetric space group Cm (no. 8) with Flack 0.01(1), a = 10.6554(12) Å, b = 18.6022(16) Å, c = 13.4204(14) Å, β = 95.166(9)°, V = 2649.3(5) Å3, Z = 2, T = 173 K, μ(Mo Kα) = 0.631 mm−1, Dcalc = 1.381 g/cm3, 18,123 reflections measured (3.048° ≤ 2Θ ≤ 51.508°), 4991 unique (Rint = 0.0805, Rsigma = 0.0488) which were used in all calculations. The final R1 was 0.0466 (I > 4u(I)) and wR2 was 0.1218 (all data). A solvent mask was calculated and 74 electrons were found in a volume of 216A3 in 2 voids per unit cell. This is consistent with the presence of one heptane molecule per formula unit which accounts for 58 electrons per unit cell. (All data shown in Table S1.). Supplementary crystallographic data (CCDC 2202318) is available free of charge from The Cambridge Crystallographic Data Centre CCDC.The polymerization reactions were carried out in a 300 mL stainless Büchi steel autoclave. The polymerization setup was equipped and connected with multifunctional systems including vacuum, argon pipeline, monomer feeding tank, catalyst injection, temperature control, and magnetic stirrer with controlling units. The temperature inside the reactor was controlled by the connected thermostat. The reactor was initially vacuumed and filled up with argon for three time. The desired amount of mixture of dry solvent (toluene) and co-catalysts was injecting inside to wash and clean the impurities in the reactor. The washing procedure was maintained for 15 min under 90 ℃. Then, the solution mixture was released by the argon pressure, which kept the inert atmosphere in reactor. At the selected polymerization conditions, the distill toluene (30 mL) was injected into the autoclave, followed by the injection of co-catalyst dissolved in dry toluene (50 mL). Ni complexes were purified using the Schlenk manipulations under argon. The required amount of Ni complexes was introduced with dissolution of the rest 20 mL toluene. The reactor was immediately pressurized into certain ethylene pressure while the magnetic stirrer was also initiated at the same time. After the required time for ethylene polymerization, the monomer pressure was released out of the autoclave. The mixed solution containing polymers was removed out and quenched by the mixture of HCl and EtOH (ratio 1:10). The polymer was collected, washed with EtOH, and then dried under vacuum oven at 60℃ for further catalytic calculation and characterization.The synthesis of the here proposed a-diimine Ni complexes could be divided into several parts, namely the synthesis of bulky anilines, a-imino-ketones, a-diimine ligands and Ni complexes (Fig. 2 ). The synthetic procedure of the bulky anilines has previously been reported, which involved the reaction between the normal anilines and diphenylmethanol catalyzed by zinc chlorides [53–55]. Subsequently, one-equivalent 2, 3-butanedione reacted with these bulky anilines with a catalytic amount of para-toluene sulfonic acid in dichloromethane under reflux, producing the a-imino-ketone precursors. The unsymmetrical a-diimine ligands were synthesized via the imine formation reaction of a-imino ketones with 4-amino-3,5-xylenol. The most conventional starting material for the synthesis of a-diimine Ni(II) complexes is the nickel(II) bromide 2-methoxyethyl ether complex ((DME)NiBr2) [11]. As (DME)NiBr2 is very sensitive to moisture, the synthesis of a-diimine Ni complexes has to be carried out under inert atmosphere. The aid of air-stable nickel(II) chloride hexahydrate (NiCl2·6H2O) as the coordination affiliation is rarely reported [17]. Additionally, the price of (DME)NiBr2 is even higher than NiCl2·6H2O. The inert-protection procedure during the synthesis and the high price of chemicals definitely increases the catalyst cost, which hampers their commercialization. Therefore, the more efficient and cheaper NiCl2·6H2O was applied to produce the a-diimine Ni chlorides in ethanol/DCM mixture, which was completely carried out at ambient. The components of the a-diimine ligands and Ni complexes synthesized in this work are not commercially available, nor was their synthesis reported previously.The a-diimine ligands (L-OH, l-FOH, L-PhOH, and L-PhFOH) were characterized by elemental analysis, 1H, 13C, and 19F NMR spectroscopy (Figs. S1–S10). The NMR analysis was also employed to verify the a-diimine Ni (II) complexes (Ni-OH, Ni-FOH, Ni-PhOH, and Ni-PhFOH). However, owing to the paramagnetic behavior of the Ni based complexes, the chemical shifts of the 1H NMR signals were very broad and difficult to determine their complex structures comparing to the ligands (Figs. S11–S14) [56]. As a consequence, the a-diimine Ni (II) complexes were further characterized by Elemental Analysis, ESI-HRMS, and a single-crystal X-ray diffraction study. Particularly, the single crystals of Ni-OH, Ni-FOH, and Ni-PhFOH were isolated from slow-diffusion of heptane (nonpolar solvent) into a dichloromethane solution (polar solvent) and further verified via X-ray crystallographic analysis. It was also found that storage in the solvents was critical to protect the single-crystals from decomposition. The molecular structures of the complexes are shown in Figs. 3–5 with selected bond lengths and angles. In the Ni complexes, the Ni atom is situated at the center of a distorted-tetrahedron structure. Two nitrogen donors belonging to the unsymmetrical a-diimine ligands coordinate with Ni chlorides. The two N-aryls are perpendicular to the planar coordination sphere of Ni and aliphatic backbone. The phenyl rings of the N-aryls surround the coordinated Ni center, which partially shields it and controls the monomer insertion rates. Thus via the various steric and electronic effects, the coordination environment is finely tuned, as evidenced by the variation of bond lengths and bond angles among these different Ni complexes. The characteristic tetrahedral geometry of a-diimine Ni (II) complexes favors the improved catalytic activity in catalyzed ethylene polymerization [57].As a first step, the initiating effects of different activators (co-catalysts) were screened for the in situ polymerization using the synthesized Ni complexes. Numerous alkyl-aluminum compounds have been previously reported as activators for late-transition metal precatalysts [17]. In this work, initiation effect of various activators such as modified methylaluminoxane (MMAO), ethylaluminum sesquichloride (EASC), diethylaluminumchloride (Et2AlCl), and trimethylaluminium (TMA) were studied. Although the activation mechanism of these co-catalysts might be slightly different, the monomer insertion of ethylene into the active cationic alkyl-metal species remains the same. The screening of co-catalyst initiation were performed using Ni-FOH in toluene with a constant ethylene pressure of 10 bar and a temperature of 30 °C (see Table 1 ).As shown in Table 1, all co-catalysts except TMA exhibited remarkable capacity to initiate Ni-FOH. MMAO is the most efficient activator for Ni-FOH in ethylene polymerization resulting in a catalytic activity of 13.0 × 106 g of PE (mol of Ni)−1 h−1. Furthermore, the GPC analysis revealed the high molecular weight (1.36 × 106 g mol−1) for the isolated PE in the MMAO-Ni-FOH system (see entry 1, Table 1). High activities were also observed for the Et2AlCl-Ni-FOH and EASC-Ni-FOH system (12.7 and 8.8 × 106 g of PE (mol of Ni)−1 h−1) (entries 2 and 3, Table 1). However, the molecular weight of the PE decreased for Et2AlCl and EASC compared to the MMAO-Ni-FOH system. This particularly demonstrates the competition between the chain-growth and chain-transfer process [58]. The latter led to a decrease in the molecular weight of PE. Somehow, the Et2AlCl-Ni-FOH and EASC-Ni-FOH system seemed to provide the active species, which typically favored the chain-transfer process in ethylene polymerization. TMA is not a suitable activator for the Ni-FOH complex (entry 4, Table 1). Nevertheless, the main cause of this poor activation performance of TMA-Ni-FOH still remains unclear. As previously reported, TMA was also applied as the significant linker to covalently tether the OH/NH2-containing late-transition complexes to solid substrate for ethylene heterogeneous polymerization [59,60]. It is also likely that TMA reacts with the Ni complex (Ni-FOH) and generates the a-diimine Ni dimethyl or heterobinuclear Ni(I) species, rather than initiating the catalytic metal center for polymerization [61].Furthermore, the influence of various polymerization conditions, such as Al: Ni ratio, polymerization temperature, catalyst lifetime and ethylene pressure, for the MMAO-Ni-FOH catalytic system was investigated to improve the catalyst’s performance (entries 1–16, Table 2 and Fig. 6 ). Initially, the influence of the Al: Ni molar ratio was optimized with an ethylene pressure of 10 bar, a polymerization time of 10 min and a polymerization temperature of 30 °C (entries 1–5, Table 2). The catalytic activities of the systems displayed an upward trend as the ratio was increased from 500: 1 to 1500: 1. The highest activity (21.84 × 106 g (PE) mol−1 (Ni) h−1) was observed at a molar ratio of 1500: 1. Above this molar ratio, the catalytic activity steadily dropped to 7.44 × 106 g (PE) mol−1 (Ni) h−1 (Fig. 6). This decreased activity was plausibly attributed to the presence of the OH moiety in the para-N-aryl, as higher Al: Ni molar ratios generates O− ions from OH. These O− ions are strong electron-donating group, which weaken the catalytic capacity of the Ni centers [62]. All PE samples exhibited the high molecular weight, typically around 1.2–1.3 × 106 g mol−1. However, no obvious correlation between the PE molecular weights and the catalyst activity was observed.The effect of temperature on the performance of the MMAO-Ni-FOH system was tested by conducting polymerization under selected temperature gradients with a constant Al: Ni ratio of 1500:1 (entries 3, 6–8, Table 2). It was observed that the reaction temperature had a significant influence on the catalytic activity and the molecular weight (Fig. 6) of the PE; namely, the molecular weight diminished (from 1.81 to 0.45 × 106 g mol−1) with increased temperatures. The MMAO-Ni-FOH system achieved the highest activity of 21.84 × 106 g (PE) mol−1 (Ni) h−1 at 30 °C. A high molecular weight for the formed PE at 0 °C was attributed to the low rate rotation of the C-N-aryl bond and chain transfer, which decreased the rate of the monomer insertion and activity [63]. Along with a higher activity and molecular weight, a better balance between the chain propagation and chain transfer mechanism could be achieved at 30 °C. As expected, both catalytic activity and the molecular weight decreased significantly at higher polymerization temperatures (60 and 90 °C). A persuasive explanation can be made that the increased temperature gave rise to the increased rotation of C-N-aryl bond. High rate rotation of the C-N-aryl bond led to the fast chain transfer (low Mw) and thermal damages to the metal center (chain termination). A polymerization temperature of 90 °C yielded an activity of 2.8 × 106 g (PE) mol−1 (Ni) h−1 and a molecular weight of 0.4 × 106 g mol−1 (entry 8, Table 2). Meanwhile, higher temperature also reduced the concentration of the ethylene monomer in toluene solution, therefore decreasing the monomer access. However, the catalytic performance of Ni-FOH/MMAO also suggests great thermal stability compared to systems reported in previous works [17].Additionally, variation of the polymerization time was used to check the lifetime of the activated Ni species (entries 3, 9 – 13, Table 2). As shown in Fig. 6, the highest activity (27.7 × 106 g (PE) mol−1 (Ni) h−1) was achieved around 5 min after the beginning of the polymerization; then the activity gradually decreased from 5 min to 120 min, leading to 7.7 × 106 g (PE) mol−1 (Ni) h−1. The reason for this decrease might be due to a monomer diffusion limit. As the synthesized PE was not fully dissolved in toluene, a polymeric diffusion barrier around the metal center was created during the ethylene polymerization. This reduced the possibility of the coordination-insertion process between the Ni center and ethylene monomer. There is the possibility that the Ni complexes were partially deactivated during the long-term polymerization. However, the catalytic activity of the Ni complexes was still maintained at high level even after 2-hour reactions, demonstrating the robust nature of these synthesized catalysts. To further investigate the influence of ethylene pressure, polymerizations were performed between 5 and 20 bar (entries 3, 14 – 16, Table 2), as a higher monomer pressure leads to a higher ethylene concentration in toluene solution. Consequently, higher activity (up to 29.1 × 106 g (PE) mol−1 (Ni) h−1) and molecular weight (Mw = 1.53 × 106 g mol−1) were observed at 20 bar (Fig. 6). This result demonstrates a superior performance of the catalyst under increased polymerization pressure.The structural modifications of the a-diimine Ni complexes were of significant importance to control the catalytic behavior and the properties of the synthesized PE. In order to determine the catalytic performance, Ni-OH, Ni-FOH, Ni-PhOH, and Ni-PhFOH were applied to the ethylene polymerization under the previously optimized conditions (entries 3, 17 – 22, Table 2). All the Ni complexes exhibited a high catalytic (8.7 – 23.7 × 106 g (PE) mol−1 (Ni) h−1) activity for ethylene polymerization at 30 °C. It was apparently observed that the F effect and the incorporation of the sterically demanding groups generated positive influences on the catalytic activity and molecular weight of the resulting PE. Both, Ni-FOH and Ni-PhOH exhibited higher catalytic activity compared to Ni-OH, whereas the catalytic activity of Ni-FOH was also higher than Ni-PhOH. These findings together indicated that the F effect plays a crucial role in improving the catalytic activity at 30 °C. The highest activity of these Ni complexes was achieved using Ni-PhFOH as the pre-catalysts, probably due to the combination of the F effects and the steric enhancement (Fig. 7 ). There was a trend concerning the molecular weight of the produced PE for different ligand structures: Ni-PhFOH (1.71 × 106 g mol−1) > Ni-PhOH (1.52 × 106 g mol−1) > Ni-FOH (1.38 × 106 g mol−1) > Ni-OH (1.15 × 106 g mol−1) observed, which could be understood as follows. The restricted access from axial directions of the metal center suppressed the chain-transfer process more efficiently, which leads to a further increase of molecular weight via chain propagation. However, the F and steric effects did not have a great impact on the catalytic activity of the Ni complexes at 90 °C. Although the catalytic performance was maintained at high activity [> 106 g (PE) mol−1 (Ni) h−1], a minor decrease was nevertheless observed for all Ni complexes (Fig. 7) – probably due to a boost of the rotation rate of the C-N-aryl bond at the high reaction temperature. The coordination between the protons from the alkyl-N-aryl groups and the metal center terminated the catalytic active species. The presence of the -CHPh2 groups at the para-N-aryl led to an increase in the molecular weight of PE, thus Ni-PhFOH and Ni-PhOH systems yielded a higher molecular weight PEs compared to Ni-FOH and Ni-OH systems. Compared to the symmetrical modifications on α-diimine Ni complexes with the incorporation of the 2, 6-dibenzhydryl groups (B in Fig. 1), one of the biggest advantages of the proposed Ni complexes (Ni-PhFOH, Ni-PhOH, Ni-FOH and Ni-OH) is the remarkably high catalytic activity in ethylene polymerization [21,22]. Even after 1 h, the catalytic activity was still maintained at the level of 107 g (PE) mol−1 (Ni) h−1 (entry 11, Table 2). Derivative Ni-Sym. from a previous research [46]. similar to B [21,22]. was selected as a benchmark reference for this work (as shown in Fig. 7). The steric bulkiness of both sides of the N-aryls limited the space for monomer insertion, thus reducing the catalytic activity. Furthermore, the rigid structure of the symmetrical Ni complexes also created fewer opportunities to modify the molecular weights and other properties of the resulting PE. Contrary to this, the unsymmetrical designs of the a-diimine Ni complexes (Ni-PhFOH, Ni-PhOH, Ni-FOH, and Ni-OH) offered both a tailored catalytic activity and a high molecular weight PE (Fig. 7). Meanwhile, this work combines the excellent catalytic features of previously described unsymmetrical a-diimine Ni complexes [64–69]. The incorporation of the aliphatic backbone is more likely to generate the PE samples with higher Mw (up to 1.81 × 106 g mol−1) in comparison to the aromatic backbone. A high catalytic activity and thermal stability of such unsymmetrical a-diimine Ni complexes were also observed in this work. Compared to the previous derivatives with aliphatic backbone, the steric enhancements on both the para- and ortho-position of the N-aryls allows for generation of a stable single-site catalytic Ni center during the ethylene polymerization[69]. In the current work a high catalytic activity, high PE Mw and narrow PDI were observed. In addition, the presence of the terminal hydroxyl group offers reactive site for covalent immobilization of these outstanding a-diimine Ni complexes on inorganic substrates for applications in heterogeneous polymerization.Compared to the currently applied Ziegler-Natta heterogeneous catalysis in ethylene polymerization, one of the main advantages of single-site catalysts is the formation of PE with narrow molecular weight distribution. As shown in Fig. 8 , the PE samples catalyzed by the a-diimine Ni complexes typically exhibit a narrow PDI, enabling better mechanical properties of the resultant polymer [70]. However, broader PDIs (higher than 3) were observed for the PE samples generated at a polymerization temperature of 90 °C for all the Ni complexes. It illustrates that deactivation and chain termination occurs at higher temperatures during ethylene polymerization.In addition, the steric hindrance from the para-N-aryl potentially squeezed the free volume of the phenyl rings at the ortho-N-aryl, which increased the possibility for the coordination between the CH and the active Ni centers [71]. Besides a narrow PDI, branched structure is another potential advantage of the a-diimine Ni complexes, enhancing the polymer mechanical flexibility, such as the tensile strength and elongation at break [72]. The chain-walking mechanism allowed the α-diimine Ni and Pd complexes to produce high molecular weight PE with highly branched structures [73]. The branched microstructures of the resulting PE samples was determined by the high-temperature 1H and 13C NMR (Figs. 9–11 ).The calculation of the branching density and peak assignments of the PE simples were based on the methods, which were previously reported [19,25,74]. As shown in Fig. 9, the 1H NMR spectra of the PE-3′ sample (entry 3 in Table 1) shows a significant quantity of terminal methyl groups in the polymer chains, which is interpreted as evidence for the PE branching (52.5B/1000C). 13C NMR spectra also helps to characterize various types of branching architectures. For instance, the 13C NMR spectra of the PE-3′ sample (Table 1) revealed multiple branching types in Fig. 10, including methyl (50.5 %), propyl (15.2 %), butyl (10.6 %), and long chain moieties (23.7 %). The polymer microstructure of sample PE-3 (entry 3 in Table 2.) exhibited less variation in the branches (26.2B/1000C) than the PE-3′ sample (Table 1).Obviously, only methyl branches were detected in the 13C NMR spectra for the PE-3 sample (Table 2.), shown in Fig. 11. Interestingly, such branched elastomer like PE was obtained via ethylene polymerization in this work by using ethylene only as the monomer feedstock; remember, the difference in the synthesis of PE-3′ (Table 1) and PE-3 (Table 2) was the use of a different co-catalysts (Et2AlCl or MMAO), applied for ethylene polymerization. This surprising finding suggests that the variation of the co-catalysts plays a role in altering the chain-walking behavior at the hence initiated Ni center—the exact details remaining unclear. A possible explanation could be that the co-catalysts initiation can induce the selectivity of the activated Ni center, where either the chain propagation (MMAO and PE-3 (Table 2.) or chain transfer (Et2AlCl and PE-3′ (Table 1) is favored upon ethylene insertion. It was observed that the molecular weight of PE-3 (Table 2) was almost twice as high as PE-3′ (Table 1), also indicating a higher rate of chain transfer if activated with Et2AlCl as compared to MMAO. Therefore, a high rate of chain transfer increased the chances of chain-walking process thus leading to the formation of branched structures. A similar observation was made in catalyzed propylene polymerization, where the MMAO activation formed a much higher syndiotacticity in comparison to the in-situ activated hafnocenes. This stereoselectivity was assumed to be attributed to the effects of counter anion [75]. Moreover, higher temperatures led to an increasing rate of the chain-walking process as well as branching, which was due to the C-N-aryl rotations of the Ni complexes. As shown in Fig. S15, for the PE-8 samples (Table 2), the 13C NMR spectra shows the presence of both the methyl (66 %) and propyl groups (34 %) as the branches (69.9B/1000C), if synthesized at 90 °C. The branching density of the PE sample also suggests a higher rate of the chain-transfer process as compared to PE-3 (Table 2), which was synthesized at 30 °C [57]. The variation of the melting points was mainly due to the different branching densities and structures, which led to the formation of various crystallinity and microstructures of the PE samples. The branching properties of PE samples were directly modified by the so-called chain-walking process in ethylene polymerization, catalyzed by such Ni complexes. The competition between the chain-growth and chain-walking process can be modulated by the applied polymerization conditions or complexes structures. The higher degree of branching densities, the lower melting points are characterized (Fig. 12 ). For example, hyperbranched and amorphous PE can be prepared via the α-diimine Pd complexes, which exhibits higher selectivity to chain-walking process than α-diimine Ni complexes in ethylene polymerization [14,20]. The α-diimine Ni complexes can produce mainly linear (few short-branches) PE with a well-defined melting point. DSC (second-heating-scan) curves revealed diverse microstructures of the selected PE samples, which were initially controlled various ligands structures and polymerization conditions. The melting points of the produced PE range from 67.9 up to 127.5 °C.In conclusion, this work presents a series of novel unsymmetrical a-diimine Ni complexes for ethylene polymerization, combining the benefits of high activity and high molecular weight PE. The two unsymmetrical N-aryl groups created the unique coordination surroundings toward the active metal center. The dibenzhydryl groups (-CHPh2/-CHPhF para 2) on N-aryl suppressed the axial direction of the Ni coplanar and retarded the rate of chain transfer, which gave rise to high molecular weight of polyethylene (Mw = 1.8 × 106 g mol−1). Meanwhile, the less bulky N-aryl provided enough space for monomer insertion, leading to the high catalytic activity (29.1 × 106 g of PE (mol of Ni)−1 h−1). The incorporation of the fluorine atoms on the bulky substitutions brought about the significant increase on the catalytic activities of these Ni complexes and PE molecular weight (Mw = 1.7 × 106 g mol−1). It also indicated that the polymerization conditions played a crucial role in controlling the catalytic behaviors and PE microstructures, including the co-catalysts, polymerization temperature and ethylene pressure. High melting transitions (up to 127.5 °C) were observed among the selected PE samples. The presence of the strong electro-donating hydroxyl group at the para-N-aryl did not negatively influence the catalytic activity, while it provided the opportunities to further functionalized the newly synthesized a-diimine Ni complexes. These unique unsymmetrical structures must generate a gradient polarizing effects at the metal center, i.e. more electron withdrawing on the F-rich and highly steric side. Such F-effects brought about positive influences on the both the catalytic performance of Ni complexes and chain-growth process in ethylene polymerization. It is very interesting to explore the role of such “metal center polarization” in future catalyst designs, which is the “key” to further optimize the catalytic behaviors of such related catalysts. This N-aryl moiety with –OH group also enables the capacity to covalently tether theses Ni catalysts on inorganic nanoparticles (SiO2, Al2O3, MgCl2, etc.) for ethylene heterogeneous polymerization. Ruikai Wu: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing – original draft. Lucas Stieglitz: Investigation, Writing – review & editing. Sandro Lehner: Investigation, Writing – review & editing. Milijana Jovic: Investigation, Writing – review & editing. Daniel Rentsch: Investigation, Writing – review & editing. Antonia Neels: Investigation, Writing – review & editing. Sabyasachi Gaan: Project administration, Writing – review & editing, Supervision. Bernhard Rieger: Writing – review & editing, Supervision. Manfred Heuberger: Project administration, Writing – review & editing, Supervision, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was financially supported by Subitex grant, Switzerland (2020–2025) and China Scholarships Council (No. 201904910562). The NMR hardware was partially granted by the Swiss National Science Foundation (SNSF, Grant 206021_150638/1). The authors are grateful to Feng-Sen Sun (LMU) for his kind help with ESI-HRMS measurements and interpretations.Supplementary data to this article can be found online at https://doi.org/10.1016/j.eurpolymj.2023.111830.The following are the Supplementary data to this article: Supplementary data 1 Supplementary data 2 Supplementary data 3 Supplementary data 4

A series of new tailored a-diimine Ni (II) complexes (Ni-OH, Ni-FOH, Ni-PhOH, and Ni-PhFOH) containing bulky ortho-N-aryl groups with various dibenzhydryl substitutes was successfully synthesized, characterized and applied in ethylene polymerization. The a-diimine ligands and Ni (II) complexes were characterized by 1H, 19F, and 13C NMR, elemental analysis, and high resolution electrospray ionization mass spectrometry (ESI-HRMS). The X-ray crystallographic study of metal complexes Ni-OH, Ni-FOH, and Ni-PhFOH revealed their distorted tetrahedral geometry. An unsymmetrical steric-enhancement design approach was employed to modulate the competition between the monomer insertion and the chain-walking process in ethylene polymerization. This facile design resulted in a high catalytic activity and yielded high molecular weight PE. The catalytic activity of these complexes was optimized by varying the polymerization conditions (temperature, time, and ethylene pressure), use of co-catalysts and variation of the Al/Ni ratio. When activated with modified methylaluminoxane (MMAO), these Ni complexes exhibited the activity as high as 29.1 × 106 g of PE (mol of Ni)−1 h−1), with a molecular weight of 1.81 × 106 g mol−1. Their thermal stability was well pronounced at elevated temperatures; high activity of 2.88 × 106 g of PE (mol of Ni)−1 h−1 and high molecular weight PE (0.86 × 106 g mol−1obtained at 90 °C). PEs with tunable branches and high melting points (127.5 °C) were obtained, which is a typical feature of LLDPE. The incorporation of fluorine atoms on N-aryl groups had a strong positive influence on the catalytic activity of the Ni complexes and favored the chain-growth process in the ethylene polymerization. Surprisingly, the simultaneous presence of terminal hydroxyl group in these complexes did not adversely affect their catalytic activity, while offering the possibility for covalent attachment to the solid supports for future heterogeneous polymerization.

Data will be made available on request.Hydrothermal liquefaction (HTL) is a promising thermochemical process for the conversion of wet biomass into energy-dense bio-oil which can be further refined into platform chemicals and liquid transportation fuels to replace fossil-derived petrochemicals (Bampaou et al., 2022). HTL is typically conducted in the presence of subcritical water (280 – 374 °C, 10 – 22 MPa) which exhibits a low dielectric constant, resulting in non-polar organic solvent like properties to assist with biomass solubilization, and high dissociation to promote fast acid-base catalyzed reactions (Gollakota et al., 2018). Therefore, the hot pressurised water environment aids the solubilisation and depolymerisation of biomass into smaller reaction intermediates which can be further deoxygenated and condensed into the desired bio-oil product. Despite this partial deoxygenation through dehydration and decarboxylation reactions, the bio-oil retains a high heteroatom content requiring significant upgrading through secondary processes such as hydrotreatment, hydrogenolysis and cracking reactions, which increase the cost of processing (Mukundan et al., 2020; Wagner et al., 2018). Besides small amounts of solid char and reaction gas, consisting predominantly of CO2, HTL also produces large volumes of aqueous phase (AP) product, which contains a mixture of short organic acids, phenolics and other polar hydrocarbons, resulting in a high total organic carbon (TOC) and chemical oxygen demand (COD). While AP carbon may be recovered via secondary processes such as aqueous phase reforming, gasification, anaerobic digestion or algae cultivation, these processes are expensive and tend to produce lower value products (Silva Thomsen et al., 2022; Watson et al., 2020).Therefore, there is a considerable need for employing catalysts that can improve solid biomass depolymerisation, convert the undesired AP products into bio-oil, and promote deoxygenation and other heteroatom removal reactions to reduce the cost for subsequent upgrading processes (Mukundan et al., 2022). The most commonly used catalysts are homogenous bases such as NaOH, KOH, K2CO3, Na2CO3, etc., which help to improve carbohydrate conversion and reduce solid formation (Biller et al., 2016) (Wang et al., 2013). However, as these homogenous base catalysts are difficult to recover and recycle post reaction, they may be more suitable for HTL of dry biomass, where the majority of the HTL aqueous phase, and hence the catalyst, can be recirculated (Biller et al., 2016).Unlike homogenous bases, heterogenous catalysts can be more easily recovered from the product mixture through simple physical processes, such as filtration or gravity. Multiple studies have explored the use of noble metals such as Pt, Pd or Ru, which have shown high activity for biomass conversion and oxygen removal (Duan & Savage, 2011; Yang et al., 2016), but due to their high costs are unlikely to be viable at commercial scale. Moreover, despite the easy separation from the HTL AP, heterogenous catalysts become contaminated with solid HTL products, particularly the inorganic ash (e.g., SiO2, CaO, MgO, K2O, Al2O3, TiO2) which account for up to 20 % of total biomass weight (Caillat & Vakkilainen, 2013), impeding their long-term cyclability. A potential solution to this problem is the use of inexpensive ferromagnetic catalysts such as Fe, Co, or Ni and their oxides, which can be magnetically separated from the solid product post reaction, as already demonstrated for other applications such as biodiesel production (Gardy et al., 2018; Shylesh et al., 2010; Zhang et al., 2021). HTL studies with inexpensive NiMo and CoMo catalysts (Prestigiacomo et al., 2019) and carbon nanotube supported Fe, Ni, and Co catalysts (Liu et al., 2021) resulted in significant increases in bio-oil O/C ratios compared to non-catalytic reactions from 0.11 to 0.16, and 0.29 to 0.69, respectively. Despite this, studies on the magnetic catalyst recovery and subsequent recycling are rare in this field, and do not fully address the separation of catalyst and solid HTL products, particularly the inorganic ash. Fe powder catalysts used during HTL of empty oil palm fruit bunch was regenerated by heating the biochar/catalyst mixture at 1000 °C under N2, with the biochar acting as reducing agent to regenerate metallic Fe, while the removal of ash (4.6 % for selected biomass feed) was not discussed (Miyata et al., 2017). Another study employed Fe powder for the HTL conversion of the macroalga Cladophora socialis, pre-treated with HCOOH to reduce its ash content from 21.2 % to 5.7 % (Nguyen et al., 2021). After reaction, the catalyst was magnetically separated, heated to 550 °C under air to burn off the biochar and finally reduced at 700 °C in the flow of H2/Ar gas to regenerate metallic Fe. However, it was found that ash build-up in the catalyst during repeated catalyst recycling led to catalyst deactivation and the use of high temperatures and hydrogen reduction may not be economically viable.The current study focuses on developing a low-cost ferromagnetic catalytic system for biomass HTL, that can be effectively separated from the solid reaction products and reused multiple times with minimal deactivation. Unlike the existing literature, the study specifically addresses the purification and reusability of catalysts post reaction, supported by detailed material analysis, without the need for high-temperature catalyst regeneration. Initially, two different ferromagnetic metal oxide catalysts, FeOx/C and NiOx/C, were tested for HTL of draff (brewer’s spent grains), an example lignocellulosic residue, and baselined against the performance of a homogeneous base catalyst (Na2CO3). The two metals were chosen for their significantly lower cost compared to Co, as well as their high cracking and hydrogenating activities. Based on the improved recovery of FeOx/C over NiOx/C after reaction, the former was selected for subsequent catalytic reusability studies and characterised to identify its active oxide phase and stability. The novelty of this study lies in (i) synthesising the active magnetite phase (Fe3O4) on activated carbon support by simple wetness impregnation method, (ii) the multi-functionality of the catalyst in both depolymerising and deoxygenating the biomass polymer resulting in remarkable bio-oil yields and composition, (iii) separation of the catalyst from the reaction solid phase by simple magnetic retrieval, and (iv) retained catalyst activity for up to 5 reaction cycles without high-temperature catalyst regeneration or hydrogen reduction. This work extends the current literature and knowledge on the metal active phase, catalyst separation and reusability for HTL reactions.Draff was provided by Chivas Brothers from the Strathclyde distillery unit. Iron (III) nitrate nonahydrate (99+ %) and nickel (II) nitrate hexahydrate (99 %), were purchased from Fischer Scientific UK, dichloromethane (99 – 99.4 %) from Honeywell, anhydrous sodium carbonate from Fischer Chemicals, and activated charcoal (DARCO, 100mesh particle size) from Sigma Aldrich. Deionised water was used for all catalyst synthesis and HTL experiments.The activated carbon supported Ni and Fe oxide catalysts, abbreviated as NiOx/C and FeOx/C throughout the article, were prepared by simple wetness impregnation method with an active metal loading of 7.5 wt%. Prior to use, the activated carbon support was dried overnight at 70 °C in the oven. The appropriate amount of aqueous solution of the Ni or Fe precursor was added to the activated carbon support and stirred for 8 h at room temperature. Water was then removed using a rotary evaporator at 50 °C, followed by overnight drying at 100 °C, and subsequent heat treatment at 550 °C for 5 h under the flow of 30 mL min−1 N2 in a tube furnace.The moisture content in draff was calculated from the weight-loss during oven-drying of as-received draff at 110 °C for 12 h. The weight of dried draff is the total solids. The dried draff was heated to 550 °C for 4 h in a muffle furnace in air. The weight loss is the volatile solid while the remaining solid is the ash content (Sluiter et al., 2008).The elemental analysis (C, H, N, S, and O) was performed using a Thermo Fisher Scientific Flash SMART elemental analyser coupled with TCD at Grant Institute, University of Edinburgh. The furnace temperature was set to 950 °C and 1060 °C for CHNS and O, respectively. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) was used to quantify the elemental percentage in the materials. The catalyst was digested by reacting 5 mL of aqua regia with 0.25 g in a microwave accelerated reaction system at 220 °C for 40 min. After cooling, the sample was filtered through a 0.45 µm pore PTFE syringe filter. X-ray powder diffraction (XRD) patterns of the materials were recorded using a Bruker D8 advance with monochromatic Cu Kα radiation (λ = 1.542 Å) at 30 kV and 15 mA with a step size of 0.1°, for the range of 10° ≤ 2θ ≤ 80°. Nitrogen adsorption–desorption data were obtained at −196 °C using a Micromeritics TriStar II 3020 surface area and porosity analyser. Prior to physisorption measurements, all samples were degassed under vacuum at 200 °C overnight. The specific surface area was determined by applying Brunauer– Emmett– Teller (BET) method and pore volume were calculated from the amount of N2 adsorbed at P/Po of 0.99. X-ray photoelectron spectrometer (XPS) by Thermo NEXSA with a monochromated Al kα X-ray source (1486.7 eV), was used to find the oxidation states of catalysts. The peaks were calibrated by using C 1 s line in the carbon spectra at 284.0 eV as a reference. Raman spectroscopy was performed using a Horiba Jobin Yvon Raman microscope under ambient conditions. Pump radiation was supplied by a red argon laser operating at a wavelength of 633 nm.Hydrothermal liquefaction reactions (HTL) were carried out in a 300 mL capacity stainless steel closed high–pressure Parr reactor (Model 4560) at temperatures of 300 °C, 320 °C, and 340 °C, with a reaction of 1 h after reaching the set-point temperature. The reactor was loaded with as received draff (corresponding to 10 g dry weight), and deionized water (dry draff: H2O = 1:10). Catalysed experiments (FeOx/C, NiOx/C, Na2CO3) were conducted with a 1:20 catalyst:draff mass ratio. The catalysts were used as synthesised and did not go through the traditional reduction step. Once complete, the reactor was cooled, and the gas was measured using an inverted water-filled measuring cylinder. Initial screening experiments indicated that the gas consisted predominantly of CO2 (over 98 %), in line with the existing literature. The contents of the reactor were passed through Whatman filter paper (grade 1), and the aqueous phase was recovered. The reactor assembly was flushed with dichloromethane (DCM) to extract any remaining residues and combined with the filter retentate to solubilize and recover the bio-oil into the solvent, followed by rotary evaporation. The remaining solid phase was dried at 70 °C overnight to quantify the overall solid recovery, consisting of biochar, ash, and catalyst.The catalyst recovery from the solid phase post reaction is given in Fig. 1 . To recover the catalysts, the solids were first suspended in DCM, thoroughly mixed using vortex (Grant instrument PV-1) and centrifuged at 3000 RPM. After decanting the supernatant, the solids were oven-dried at 70 °C for an hour, re-suspended in deionized water and catalyst was recovered using a magnetic bar. The recovered catalyst was dried overnight at 70 °C before used for the reusability studies. To compensate for the unavoidable losses associated with catalyst filtration and recovery (2 % to 18 % of catalyst loading), fresh catalyst was added for repeatability studies. The reusability tests were conducted with the same procedure as mentioned above. All the experiments were performed in duplication, and the errors are represented as the standard deviation.Bio-oil was analysed by GC–MS, CHNS(O) for elemental quantification, bomb calorimetry to determine the calorific value, and muffle furnace for measuring the boiling point distribution. The bio-oil was analysed using a Shimadzu GC-2010 Plus with Restek Rtx- 5 Sil MS column (30*0.25*0.25). The injection temperature was 220 °C with a split ratio of 50.0. The column temperature was set to 50 °C and held for 5 min, then ramping to 250 °C at the rate of 15 °C min−1 and held there for 1 min and finally increased to 300 °C at 5 °C min−1 and held for 1 min. Cal2K oxygen bomb calorimeter (Digital data systems) was used to find the calorific value of bio-oil using 0.2 ± 0.02 g of each sample. The tests were performed in triplication. CHNS(O) analysis is discussed in section 2.3. The boiling point distribution of the different bio-oils were analysed using a muffle furnace (in the presence of air). Alumina crucible was used, and the empty weight of the crucible was noted. Then, bio-oil was sampled into the crucible and the final weight was noted. The weight of bio-oil was maintained around 0.7 ± 0.1 g for better comparison. The crucibles were placed in the furnace and the temperature was set to 100 °C at the heating rate of 5 °C min−1. Once set temperature was reached, the oven was turned off and the samples were let to cool, and the weight loss was noted. For the consecutive temperatures (200 °C, 300 °C, 400 °C, 500 °C, and 600 °C), the furnace was initially set to the previously analysed temperature and then ramped to desired temperature and the weight loss was noted. The total organic carbon in aqueous sample was analysed using Analytik Jena Multi N/C 2100 S HTC TOC/ TN CLD system with HT1300 solids combustion module and AS 60 Autosampler. A thermogravimetric analyser (Mettler Toledo TGA/DSC 1 STAR system) was used to find the decomposition profile. The sample was heated from 25 °C to 800 °C in the flow of air (20 mL min−1).The yields (%) of each phase, % deoxygenation in bio-oil, and % carbon distribution into each phase, % carbon recovery, % energy recovery, H to C effective ratio, and % catalyst recovery were calculated using the following formulae: (1) Yield of bio - oil (\%) = Mass of bio - oil g Mass of dry , ash - free draff g ∗ 100 (2) Yield of solids ( % ) = Mass of solid after reaction g - weight of catalyst g Mass of dry , ash - free draff g ∗ 100 (3) Y i e l d o f a q u e o u s p h a s e ( % ) = 100 - y i e l d o f ( b i o - o i l + s o l i d s + g a s ) (4) Yield of gas % = n gas M gas Mass of dry , ash - free draff g ∗ 100 = PV RT ∗ M gas Mass of dry , ash - free draff g ∗ 100 where ngas  = number of mols of gas, Mgas  = molecular weight of gas, assumed to consist exclusively of CO2 (44 g mol−1), P = 1 atm (101,325 Pa), R = 8.314 m3 Pa/Kmol−1, T = temperature (K) at which the gas was collected, V = recorded volume of gas in m3. (5) Deoxgenation % = Molar ratio of O C in draff - Molar ratio of O C in bio - oil Molar ratio of O C in draff ∗ 100 (6) C a r b o n d i s t i b u t i o n % = M a s s o f C i n p r o d u c t M a s s o f C i n d r a f f ∗ 100 (7) C a r b o n r e c o v e r y % = M a s s o f c a r b o n i n b i o - o i l ∗ y i e l d o f b i o - o i l M a s s o f c a r b o n i n d r a f f (8) E n e r g y r e c o v e r y % = H H V o f b i o - o i l ∗ y i e l d o f b i o - o i l H H V o f d r a f f (9) % C a t a l y s t r e c o v e r y = c a t a l y s t u s e d f o r r e a c t i o n g - c a t a l y s t r e c o v e r e d ( g ) c a t a l y s t u s e d f o r r e a c t i o n g ∗ 100 (10) H / C e f f e c t i v e = ( H - 2 O - 2 N ) / C where H, C, O, and N are the moles of hydrogen, carbon, oxygen, and nitrogen, respectively in the sample (Huang et al., 2016).Draff is a key by-product from the malting process of barley in whiskey distilleries, where barley is mixed with water to convert the starch into soluble sugars. The insoluble solid fraction is recovered as draff and usually used as wet animal feed. Scottish based draff typically contains around 17–25 % cellulose, 22–28 % hemicellulose, 12–28 % lignin, as well as proteins (15–24 %) and lipids (∼10 %) (Foltanyi et al., 2020). The draff used in this study had a residual water content of 69.8 %, below the typical HTL water to biomass ratio of 10:1, and hence no additional dewatering or drying is required (see supplementary material). Instead, some make-up water is required, which may be added by recycling some of the HTL aqueous phase product, or by co-feeding pot ale or spent lees, which are aqueous-phase by-products from the malting processes and contain dilute concentrations of organic acids and alcohols (Akunna & Walker, 2017).The dried draff contained 48.9 % carbon, 7.2 % hydrogen, 3.8 % nitrogen, 0.2 % sulphur and 34.5 % oxygen, and the remaining 4.4 % can be attributed to minerals in the ash phase (5.1 %). Using a typical protein nitrogen content of 16 %, the maximum draff protein content can be estimated as 23.8 %, within the typical range stated above. Oxygen is present as different functional groups in draff such as fatty acids, phenolics, aldehydes, ketones, amino acids, etc., as well as oxides in the ash phase, explaining the slight difference between ash content (5.1 %) and non-organic minerals (4.4 %).To determine the optimal temperature for catalytic studies, initial screening experiments were conducted at three typical HTL temperatures (300 °C, 320 °C, and 340 °C) and the % yield and % carbon distribution into bio-oil, aqueous, gas and solid phases are given in Fig. 2 . In Fig. 2a, the maximum solid residue yield of 16 % was observed at 300 °C, reducing to 13.5 % at 320 °C and 340 °C, indicating improved biomass depolymerisation at higher temperature. While 300 °C is sufficient to break the glycosidic bonds in carbohydrates, previous HTL studies with protein-rich biomass have shown that higher temperatures are required for full protein and lipid depolymerisation (Seshasayee & Savage, 2021; Shakya et al., 2015). Bio-oil yields increased from 27.4 ± 1.7 % at 300 °C to 41.2 ± 2.2 % at 320 °C, before reducing slightly to 38.5 ± 3.0 % at 340 °C. At the same time, aqueous phase yields reduced from 52.6 ± 2.7 % at 300 °C to 41.4 ± 0.3 % at 320 °C, indicating increased recombination of polar organic molecules, such as glycoaldehydes, organic acids, furfurals and phenols, into less-polar components which preferentially fractionate into the oil phase as the reaction temperature is increased (Gollakota et al., 2018). The small reduction in oil yields at 340 °C may be caused by increased cracking at high temperatures (Jindal & Jha, 2016), although gas yields remained constant at around 3.6 % at all three reaction temperatures. The data clearly suggest that HTL temperatures of 320 °C are sufficient to depolymerize draff and maximise carbon extraction into the oil phase, avoiding excessive cracking and other undesirable side reactions.Oil yields from this study are significantly higher than those reported previously for non-catalytic HTL of draff at 300 °C for 1 h (∼14 %) (Déniel et al., 2017), but lower than from pyrolysis in a twin coaxial screw reactor at 450 °C (51 %), containing both organic and aqueous phases (Mahmood et al., 2013). These differences may be explained by a variety of factors, including differences in biochemical composition, water loading, heating rates etc. The carbon distributions given in Fig. 2b follow similar trends to the mass yields, but with a significant shift to the oil phase, ranging from 36.4 ± 3.5 % at 300 °C to 58.3 ± 3.5 % at 320 °C, indicating the preferential fractionation of carbon to this product. While an excellent carbon balance of >99 % was obtained at 320 °C and 340 °C, only 88 % of carbon was recovered at 300 °C, due to the high viscosity of the bio-oil/solid product mixture, making it difficult to extract all the products completely.Two ferromagnetic metal oxide materials, FeOx/C and and NiOx/C, were tested as potential magnetically recoverable low-cost catalysts for the conversion of draff, both in the absence and presence of homogenous Na2CO3 base catalyst (Fig. 3 ). While all catalytic runs resulted in a significant reduction in solid yields, the effect was more pronounced for the metal-oxide catalysts on their own than the metal oxide – base combinations (Fig. 3a). These trends are slightly different to the catalytic conversion of corn straw over Zn/Ni/Co-Fe2O4 (Chen et al., 2021) and food wastes over mixed metal oxides based on Al, Si, Fe, and Ca (Cheng et al., 2020) where the presence of base (NaOH and red mud/red clay base catalysts, respectively) reduced the solid residue yield compared to the single catalytic system. In both cases, it was reported that the base to acid site density was crucial for improving bio-oil yield.A potential explanation for the discrepancy with the current study is the difference in biomass composition. While corn straw and food wastes are rich in cellulose and carbohydrates, respectively, draff contains higher fractions of lignin, protein, and lipids with recalcitrant CC and CO bonds which require strong acid/metal sites to cleave. HTL of lignin under basic conditions produces phenolic monomers which are prone to repolymerise into oligomers increasing solid formation (Ciuffi et al., 2021). Base catalysts have also been found to be detrimental for the conversion of protein and lipid-rich feedstocks due to the inefficiency of alkali catalysts to break the peptide and ether linkage bonds (Shah et al., 2022).Consistent with the reduced solid yields, all catalytic reactions resulted in an increase in the HTL aqueous phase (AP), indicating improved biomass depolymerisation. Bio-oil yields were significantly enhanced for the NiOx/C (47.9 %) and FeOx/C (49.3 %) catalysed reactions, but addition of base resulted in a reduction of oil yields for all three reaction systems. It is known that HTL follows a multi-step reaction sequence, starting with the hydrolysis of biomass into reactive intermediates in the aqueous phase, followed by the deoxygenation and condensation into organic bio-oil compounds. Therefore, although the presence of base appears to increase the initial biomass depolymerisation during the non-catalytic experiments, it inhibits the subsequent conversion of AP products, resulting in reduced bio-oil production.The trends are even more pronounced for the product carbon distribution, where the NiOx/C and FeOx/C catalysts increase carbon recovery to the bio-oil to 75.1 ± 2.9 % and 78.6 ± 2.1 %, respectively, while reducing AP carbon by 18.9 ± 0.4 % and 14.8 ± 0.9 % (Fig. 3b). Besides improved bio-oil yields, the lower AP carbon reduces the demand for downstream waste-water treatment, improving the viability of the process. In contrast, addition of base increases the carbon concentration in the AP between 21.9 % for FeOx/C to 69.4 % for the non-catalytic runs, while reducing bio-oil carbon by 5 %, further demonstrating the inhibitive effect of base on the conversion of AP intermediates into bio-oil.Non-catalytic HTL of draff increased the carbon content from 48.9 % in the biomass to 68.7 % in the bio-oil, while reducing oxygen content from 34.5 % to 19.8 %, hydrogen from 7.2 % to 6.8 % and nitrogen from 3.8 % to 2.9 % (Table 1 ). Despite the reduction in hydrogen content, the effective H/C ratio, (H/C)eff, which indicates the required amount of hydrogen and energy during bio-oil refining (Karatzos et al., 2014), increased from 0.64 to 0.70, due to the large reduction in oxygen content. Similarly, the higher heating value (HHV) increased from 22.1 MJ kg−1 in the biomass to 30.3 MJ kg−1 in the bio-oil, corresponding to an energy recovery of 54.3 % into the bio-oil. Oil properties can be improved both by hydrogenation reactions to eliminate O and N as water and ammonia, or through non-hydrogen reactions such as cracking, releasing oxygen in the form of carbon oxides to reduce the heteroatom content, and hence improving the (H/C)eff value. The presence of base catalyst increased the bio-oil carbon content to 73.6 %, while reducing the hydrogen and oxygen contents to 6.7 % and 15.5 %, respectively, raising the (H/C)eff to 0.74, with a HHV of 32.4 MJ kg−1. However, the reduced bio-oil yields and bio-oil carbon recovery lowered the overall bio-oil energy recovery to 49.1 %.The addition of the as-synthesised metal catalysts resulted in a significant increase in both bio-oil carbon (81.3 % and 80.0 %) and hydrogen contents (8.3 % and 9.5 % for FeOx/C and NiOx/C, respectively), leading to high effective H/C ratios (1.06 and 1.24), and HHVs (37.8 MJ kg−1. and 38.6 MJ kg−1), comparable to the HHV of heavy fuel oil (40 MJ kg−1) (Karatzos et al., 2014). Together with the high oil yields, the addition of FeOx/C and NiOx/C result in excellent energy recoveries of 84.4 ± 0.8 % and 83.7 ± 0.7 %, respectively, indicating that the catalysts simultaneously improve bio-oil yield and bio-oil composition. These values are comparable to the highest HTL energy recoveries of 87 % reported to date (for halophytic microalga, Tetraselmis sp.) (Eboibi et al., 2014), and significantly higher than those obtained from non-algae biomass (up to 75.6 % for sewage sludge derived bio-oil) (Anastasakis et al., 2018). Particularly, the high hydrogen content of 9.5 % in the bio-oil produced over NiOx/C is remarkable and may be ascribed to the excellent hydrogenation and effective water dissociation activity of nickel-based catalyst (Subbaraman et al., 2011; Wang et al., 2020). Nickel oxides have been previously found to yield superior bio-oil hydrogen content during HTL of microalgae compared to other metal oxides such as Fe, Co, Mg, and Mo (Wang et al., 2018). Zhu et al. reported the exceptional catalytic activity of Ru/Ni/Ni(OH)2/C for room temperature hydrogenation of naphthalene, where it was reported that the role of Ru was hydrogen activation, while Ni acts as a bridge in transferring the hydrogen species to the naphthalene over the Ni(OH)2 sites (Zhu et al., 2017). Meanwhile, using FeOx/C catalyst, the H % still increased to 8.3 %, indicating good hydrogenation properties. A potential mechanism is the reaction of partially oxidised Fe3O4, with water to generate in-situ hydrogen. For example, Fe3O4 catalyst has been reported to effectively dehydrate cellobiose to 5-hydroxymethylfurfural in water, while retaining its activity and oxidation state even after several uses (Bhalkikar et al., 2015). Addition of base to the metal-oxide catalysed reactions resulted in reduced bio-oil carbon and hydrogen contents, while increasing the fractions of oxygen and nitrogen. Together with the lower bio-oil yields overall, the energy recovery is significantly reduced to 67.9 % for FeOx/C + Na2CO3 and 60.6 % for NiOx/C + Na2CO3, and hence the presence of base appears detrimental to the metal-oxide catalysed HTL of draff.A commonly used qualitative analysis of bio-oil composition is its simulated boiling point distribution, which determines the fraction of compounds within pre-defined boiling point ranges, and hence the downstream refining requirements. Here, boiling point distributions were obtained by heating bio-oil samples in a muffle furnace under controlled heating rates and recording the resulting weight losses at 100 °C intervals (Fig. 4 ). As expected, bio-oil weight losses up to 100 °C were low for all reactions (∼1 – 2 %), as most volatiles would be lost during solvent evaporation of the oil recovery process. Given the high overall carbon recoveries of > 99 % (See Fig. 3b) and low gas yields (3.4 %), volatile formation during HTL of draff appears to be low for both the standard and catalytic reactions, and the low boiling point compounds are most likely attributed to residual water or DCM solvent in the bio-oil.Non-catalytic HTL produced relatively heavy bio-oils, with the majority of compounds in the 400 – 600 °C boiling point range. The results are consistent with the low effective H/C ratio of these oils, suggesting limited cracking and hydrogenation during bio-oil formation. Therefore, significant upgrading will be required to convert these bio-oils into useful transportation fuels or chemicals. It is also notable, that the majority of compounds fall outside the suitable range for standard GC analysis (<240 °C), demonstrating that this method is not suitable for bulk bio-oil characterisation. Therefore, commonly reported compositional analysis by GC provides only limited insight into the true molecular make-up of bio-oils. Addition of base catalyst as limited effect on the formation of low-boiling point compounds but increases the production of compounds in the 300 – 400 °C range from 5.8 to 7.5 %. This may be the result of improved biomass depolymerisation into smaller aqueous phase intermediates and in-situ hydrogen production via formate (HCOO-Na+) formation, aiding the reduction of heavier hydrocarbons (Sınaǧ et al., 2004). At the same time, the increase in lower boiling point compounds is outweighed by the notably reduction in heavier compounds in the 400 – 600 °C region, resulting in an overall decrease of bio-oil production, and hence reduced overall energy recovery.Both metal oxide catalysts resulted in a significant increase in the formation of low-boiling compounds in the gasoline (100 – 200 °C) and particularly jet fuel (200 – 300 °C) range, consistent with the high effective H/C ratios and energy densities. Acidic metal oxides have been proven effective catalysts for C–heteroatom bond, aiding the deoxygenation of shorter chain organics, reducing their solubility in the aqueous to oil phase. Ni and Fe oxide catalysts are also well-known for cracking reactions to produce lighter hydrocarbon compounds (Qiu et al., 2022). The metal oxides also enhanced the production of compounds in the 300 – 400 °C and 400 – 500 °C range, while significantly reducing the formation of heavies (boiling points > 500 °C) from 25 % for the non-catalytic run to 1.5 % and 3.5 % for FeOx/C and NiOx/C, respectively. Crucially, around two thirds of compounds produced with FeOx/C (66.5 %) and NiOx/C (63.5 %) have boiling points below 400 °C, indicating their potential for upgrading to vital chemicals and fuel additives.While the addition of base to the metal oxide catalyst reactions resulted in only small reduction of overall bio-oil yields, particularly for FeOx/C, it shifted the boiling point distribution towards heavier compounds, consistent with the lower hydrogen and energy content in these products. Cracking reactions are commonly catalysed by acid-sites on the metal oxide catalysts, which are likely blocked by the base resulting in reduced bio-oil conversion.Both metal oxide catalysts investigated for HTL of draff showed excellent activity in not only increasing bio-oil production, but also improving bio-oil composition through reduced heteroatom content, increased effective H/C ratio and heating content, and improved boiling point distribution. Despite this, commercial application of these catalysts for the HTL of biomass is only feasible if the catalysts can be effectively recovered and recycled. Particularly the separation of catalyst from solid ash poses a major challenge for biomass HTL, as unlike char, the ash cannot be readily volatilised or incinerated. Draff used in the current study had an ash content of 5.1 %, comparable to the amount of catalyst loaded to the system, and hence most of the ash must be removed prior to recycling to avoid system accumulation. By employing the two ferromagnetic catalysts FeOx/C and NiOx/C, it was hoped that their magnetic properties could be exploited for post reaction recovery. Both fresh materials exhibited magnetic properties and were attracted to a magnet placed outside the sample vial. However, the magnetic affinity of FeOx/C was visibly stronger than that of NiOx/C, increasing the amount of material that could be dragged up at a time (see supplementary material).After reaction, initial attempts to magnetically separate the catalysts from the solid reaction products were unsuccessful, as residual oil and char trapped in the solids acted as a binder between the catalyst and ash, inhibiting the separation. However, after adapting the procedure to improve bio-oil extraction through thorough mixing with DCM and suspending the resulting catalyst-ash mixture in water, the majority of FeOx/C catalyst (96 %) could be recovered. In contrast, none of the NiOx/C could be recovered. NiOx/C displayed two sharp peaks with a 2θ values of 52.15° and 61.06° corresponding to the JCPDS 00-001-1260 card for Ni metal (see supplementary material). The formation of metallic Ni during heat treatment in N2 is ascribed to the low reduction temperature required for Ni oxide reduction to metal. For example, it has been previously reported that Ni nitrate precursor, supported on carbon, first decomposed into Ni oxides, before reducing into metallic Ni due to oxygen transfer to the carbon support, during high temperature treatment (723 K) under N2 flow (Gandia & Montes, 1994). However, post reaction, the catalyst no longer exhibited any peaks corresponding to Ni oxides or Ni metals, explaining the loss of magnetic properties, and indicating poor catalyst stability under hydrothermal conditions. Hence this catalyst appears unsuitable for this process and could not be used for further catalyst reusability studies.To test the stability of FeOx/C catalyst during HTL of draff, its reusability was tested over 5 reaction cycles (Fig. 5 ). The catalyst recovery ranged between 80 % and 98 % and the difference in catalyst weight was compensated using fresh catalyst. Reaction yields were corrected to discount the contribution of fresh catalyst using the following equation: (11) Y corrected = Y actual - m FC m TC × Y FC - Y NC where Y is yield of different product fractions, mFC and mTC are the weights of fresh and total catalyst, respectively, and YFC and YNC are the yields obtained from the initial catalytic and non-catalytic reactions, respectively.The results show a gradual decrease in the carbon recovery to the bio-oil from 80 % initially to 71.7 % after 5 reaction cycles, while the solid formation doubled from 1.8 % to 3.6 %. There were no significance differences in the aqueous and gas yields during the reusability studies. The increased solid formation resulted in increased material losses during product extraction, resulting in a slight decrease in overall carbon balance closure. The increase in solid formation also impeded the catalyst separation post reaction, reducing catalyst recovery from 96 % after the first run, to only 86 % retrieved by 5th reaction cycle. Nonetheless, even after 5 cycles, the catalyst showed a significant improvement in oil yields and reduced solid formation compared to the non-catalytic experiments.The catalysts were supported on commercial activated charcoal, with a surface area of 870 m2 g−1. The N2 adsorption–desorption isotherm profiles (see supplementary material) indicate that the carbon support consists mostly of mesopores (type IV with H3 hysteresis loop, IUPAC classification isotherm), with a pore volume of 0.6 cm3 g−1 and pore size of 5.7 nm. As expected, following FeOx and NiOx deposition the material surface area reduced slightly to 706 and 637 m2/g, pore volume decreased to 0.5 cm3 g−1 and pore size to 3.7 and 3.8 nm, respectively with FeOx/C and NiOx/C catalysts, due to metal oxide deposition inside the pores, however the overall mesopore structure was retained. The bulk metal loading in the catalyst (Fe in FeOx/C and Ni in NiOx/C) was confirmed as 7.4 % and 7.2 % using ICP-AES analysis (see supplementary material).Iron oxide exist in various phases, including α-Fe2O3 (hematite), γ-Fe2O3 (maghemite), and Fe3O4 (magnetite). While both maghemite and magnetite are magnetic, fully oxidised hematite is generally nonmagnetic. Therefore, given the strong magnetic affinity of the material, the presence of significant amounts of α-Fe2O3 in the catalyst is unlikely. Analysis by XRD (see supplementary material) revealed the presence of a broad reflection around 25°, corresponding to amorphous carbon, and several reflections at 35.07 ° (220), 41.38 ° (311), 50.44 ° (400), 62.91 ° (422), 67.22 ° (333), and 74.11 ° (440), which could be matched to the JCPDS-ICCD diffraction pattern 082–1533 of Fe3O4 (Silva et al., 2013). To further confirm the formation of Fe3O4, Raman analysis (see supplementary material) revealed a broad band around 670 cm−1 which may correspond to Fe3O4, however Fe3O4 is a poor Raman scatterer and tends to oxidise under laser irradiation inducing phase changes (Li et al., 2012; Schwertmann & Cornell, 2008). This may explain the presence of band around 243 cm−1 and 290 cm−1, which have been previously attributed to the Eg Fe- O symmetric bending of α-Fe2O3 (Mäkie et al., 2011), and the band around 228 cm−1, which can be attributed to the A1g Fe- O symmetric stretching of α-Fe2O3. The oxidation state of the metal oxides was further characterized by high resolution XPS and the deconvoluted XPS spectra of the Fe 2p region (see supplementary material). Fe 2p exhibits two oxidation states, +2 and + 3. The spectral bands at 712.1 eV and 725.8 eV corresponds to the 2p3/2 and 2p1/2 of Fe + 3 of Fe3O4, while the bands at 710.4 eV and 723.8 eV can be attributed to the 2p3/2 and 2p1/2 of Fe + 2. Stoichiometrically Fe3O4 can be expressed as FeO.Fe2O3, with a ratio of Fe2+ to Fe3+ of 0.33 to 0.67. The results from the deconvoluted peaks of Fe 2p region provided a ratio of 0.3:0.7 which is consistent with the values published earlier (Yamashita & Hayes, 2008). A potential explanation for the formation of Fe3O4, rather than fully oxidised Fe2O3, is the use of N2 during heat treatment, which may limit the availability of oxygen to form the more common Fe2O3 phases. With NiOx/C, the characteristic spectral bands at 853.4 eV and 870.7 eV corresponds to the Ni 2p3/2 and Ni 2p1/2 of Ni-O with a satellite peak at 861.5 eV. The bands at 856.3 eV and 873.9 eV corresponds to the Ni 2p3/2 and Ni 2p1/2 of Ni(OH)2 with a satellite peak at 878.6 eV (Nesbitt et al., 2000). As can be observed, Ni is mainly present as Ni(OH)2 which could be due to the immediate conversion of Ni(NO3)2 precursor to Ni(OH)2 which is strongly bound to the carbon surface (Sturgeon et al., 2014). The formation of Ni metal was not observed in the XPS which could be due to the surface oxidation of the Ni atoms.To gain further insight into the behaviour and performance of the promising FeOx/C catalyst, the as synthesised and spent materials (recovered after 5 reaction cycles) were thoroughly analysed to identify the active sites and surface changes during the HTL reaction. The FeOx/C catalysts retrieved after 5 reaction cycles were re-characterised to identify any changes in physicochemical properties. From ICP analysis, the amount of Fe reduced from 7.4 % in the fresh to 5.9 % in the spent catalyst, indicating some degree of metal leaching over the course of the 5 reactions (see Supplementary Information). The decrease in iron content could be due to the leaching of iron from the catalyst into the solid collected after reaction. As it has been reported, metal leaching from activated carbon support is possible due to the low surface oxygen groups in the carbon that makes a poor support-metal interaction. To check the possible leaching of iron into the solid phase, the solid residue obtained after catalyst recovery was characterised with an iron content of 3.6 %, compared to 2.4 % in the ash obtained from draff (obtained by heating at 550 °C for 4 h). The other minor elements found by SEM-EDS in ash were Ca, P, Mg, Na, and K. Besides leaching, there is also the possibility that some of the biomass ash was retained in the recovered catalyst, reducing its apparent Fe content, and vice versa.However, as described above, the ash retained some of the solid catalyst which may partially account for this increase in Fe content. The XRD patterns of the catalyst retrieved after 5 reaction cycles (before and after ash removal) are less crystalline (see supplementary material), but still display the diffraction patterns related to Fe3O4, as well as small peak corresponding to the α-Fe2O3 phase. These results suggest that Fe3O4 may participate in redox reactions under hydrothermal conditions, which may explain its excellent activity for bio-oil deoxygenation and hydrogenation. High-resolution XPS analysis of the spent materials did not detect any signals related to Fe, which may indicate carbon deposition on the catalyst surface XPS is sensitive only up to 10 atomic layers. Interestingly, a significant amount of N was observed during the survey scan which could indicate surface adsorption of biomass nitrogen. For comparison, an XPS survey scan of biochar produced by non-catalytic HTL of draff was performed and similar signals related to N were observed, supporting this observation (see supplementary material).Overall, despite the small metal losses and phase transformation, the catalyst maintained good stability after 5 reaction cycles as demonstrated by the high activity compared to the non-catalytic experiments. It was demonstrated that the magnetic properties of the catalysts could be effectively exploited to separate the catalyst from the solid HTL reaction products, particularly ash, although the process needs to be further developed to reduce the number of processing steps and increase the lifetime of the catalyst, such as; (i) pre-treatment of biomass to remove ash; (ii) synthesis of a core–shell catalyst where the active species is encapsulated in the core and covered by metal oxide such as Al2O3 or SiO2; (iii) increase of catalyst particle size or iron oxide deposition into monoliths. It has to be noted that the metal oxide should be stable in HTL conditions since γ- Al2O3 tend to phase transfer to boehmite under pressurised water.In summary, a cheap, safe, stable, and magnetically separable Fe3O4 catalyst supported on activated carbon was developed for HTL of lignocellulosic biomass. Catalyst addition not only increased bio-oil mass yields by 49.3 %, to reach non-optimised carbon and energy recoveries of 82 % and 84.3 %, respectively, but also produced an oil with an exceptionally high calorific value of 37.8 MJ kg−1. The catalyst was separated and reused up to 5 times, however, further optimisation of reaction and recovery conditions are required to achieve commercial relevant system. Swathi Mukundan: Data curation, Formal analysis, Investigation, Writing – original draft. Jin Xuan: Funding acquisition, Investigation, Writing – review & editing. Sandra E. Dann: Methodology, Formal analysis, Writing – review & editing. Jonathan L. Wagner: Conceptualization, Methodology, Project administration, Supervision, Writing – review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work was supported by UKRI-EPSRC through grant EP/V011863/1. The authors sincerely acknowledge the facilities and assistance provided by Loughborough Materials Characterisation Centre (LMCC) at Loughborough University. Sincere thanks to MrsBethany Taylor and Dr Tanya Radu from the Department of Water Engineering, Loughborough University for the ICP and TOC analysis. Further, we would like to acknowledge Chivas Brothers for providing the brewer’s spent grains used in this study.Supplementary data to this article can be found online at https://doi.org/10.1016/j.biortech.2022.128479.The following are the Supplementary data to this article: Supplementary data 1

This article reports a safe, low-cost, and industrially applicable magnetite supported on activated carbon catalyst that can be magnetically retrieved from the solid and reused multiple times without the need of a regeneration step. The FeOx/C catalyst improved the bio-oil yield by 19.7 ± 0.96 % when compared to the uncatalysed reaction at 320 °C for the HTL of draff (brewer’s spent grains). The use of homogeneous Na2CO3 base as a catalyst and co-catalyst, improved carbon extraction into the aqueous phase. The exceptional catalytic activity can be attributed to the Fe3O4 phase which can produce in-situ H2 that improves the biomass decomposition and oil property with an energy recovery of ∼84 %. The FeOx/C catalyst was separated using magnetic retrieval and maintained its catalytic activity even up to 5 reaction cycles showing potential as a cheap catalyst for HTL reactions and can be scaled-up for industrial applications.

Excellent mechanical and chemical properties enable polyolefin widely used in automation, electronics and other industrial fields [1–3]. Current development trend indicates that the polyolefin industry will continue to maintain a rapid growth rate of 4%–5% in the next few years [4]. Furthermore, catalysts for olefin polymerization have always been the focus of researchers [5]. Since the Ziegler-Natta catalysts and Metallocene catalysts are hard to realize the synthesis of polyethylene with high branching density [5], Brookhart et al. reported the α-diimine(ΙΙ) nickel complexes as the catalysts for the ethylene polymerization [6,7]. Based on the unique “chain walking” mechanism [8], α-diimine nickel catalysts with varied ligand structures could greatly enrich the microstructure of polymer chains and then affect the properties of the resultant polymers [9,10], which provides a new path for the preparation of polyolefin. In the next two decades, researchers have conducted a series of in-depth researches on the catalytic performance by adjusting external reaction conditions [11–13] and ligand structures [14–16]. As one of the important routes to obtain high-end polyethylene products, the introduction of branched chains would further enrich the microstructure of polyolefin. So the presence of 1-hexene in the ethylene polymerization would contribute to enrich the microstructure and manipulate the product properties [17]. However, systematic investigation of ligand effect on ethylene/1-hexene copolymerization and microstructure manipulation of products have not been reported in detail before.In this contribution, we investigated the ethylene/1-hexene copolymerization performance of three catalysts with different steric effect, naming (Ar1N=C(Me)–C(Me)=NAr1) NiBr2 (Cat. A), (Ar1N=C(An)–C(An)=NAr1) NiBr2 (Cat. B), and (Ar2N=C(An)–C(An)=NAr2) NiBr2 (Cat. C) (where Me = methyl, An = acenaphthene, Ar1 = 2,6-(i-Pr)2C6H3, Ar2 = 4-MeO-2,6-(Ph2CH)2C6H2) respectively, under different reaction conditions. The ligand steric effect of these three α-diimine nickel complexes on branching distribution was detected by nuclear magnetic resonance (NMR) in detail. In addition, electron paramagnetic resonance (EPR) experiments were used to investigate the chemical valence of Ni species and the proposed polymerization mechanism in the presence of co-catalyst.Toluene (99.5%), ethanol (99.7%), chlorhydric acid (36.0%–38.0%) were purchased from Sinopharm Chemical Regent Co. Ltd. Ethylene (99.99%) and argon (99.99%) were purchased from Jin Gong Gas Co. Ltd. Catalysts (Cat. A, Cat. B, and Cat. C, Fig. 1 ) were provided by the University of Science and Technology of China [6,16]. C6D4Cl2 (o-DCB) were purchased from Qingdao Tenglong Weibo Technology Co. Ltd. Polymerization-grade ethylene and argon were further processed by the purification system. Toluene was refluxed in the presence of sodium and indicator benzophenone before it was used for ethylene polymerization.All polymerization process involving air and/or moisture-sensitive compounds were performed under argon atmosphere under strict anhydrous and anaerobic conditions through the Schlenk techniques [18,19]. Olefin polymerization was carried out in a 100 mL three-necked flask. Co-catalyst methylaluminoxane (MAO) and catalyst toluene solution were injected to initiate polymerization. It was quenched by the addition of acidified ethanol (5 wt % HCl) after the polymerization was completed. The product was washed by anhydrous ethanol for more than three times and dried at 40 °C to constant weight.The detailed polymerization conditions were summarized in Table 1. 1H NMR spectra were recorded at 120 °C on a Varian Mercury-Plus 300 MHz spectrometers. Chemical shift (δ) were expressed as parts per million and 1H NMR spectra were referenced using the solvent o-DCB. Sample was prepared as 40 mg/mL and the cumulative scan was 100 times. It was used to investigate the branching density (B) of the products. 13C NMR spectra were obtained on an Agilent DD2 600 MHz spectrometer and CDCl3 was used as the solvent. The concentration of prepared sample was 50 mg/mL and the cumulative scan was 3000 times. The peak of the main chain methylene was set to δ 30 as an internal standard. It was used to detect the branch distribution of the products.Gel permeation chromatography (GPC) measurements were carried out using a PL-GPC220 high temperature gel permeation chromatography at 160 °C. And 1,2,4-trichlorobenzene was used as the solvent. It was used to investigate the molecular weight (M w) and polydispersity index (PDI) of the product.Differential Scanning Calorimetry (DSC) measurements were conducted under nitrogen in the temperature range from −80 °C to 160 °C with heating or cooling rates of 10 °C/min on a TA Q200 instrument. It was used to detect the glass transition temperature (T g) and melting temperature (T m) of the products.EPR measurements were carried out using a Bruker A300 electron spin resonance spectrometer at room temperature. The concentration of sample was 2.5 mmol/L of toluene solution.The experiments focused on exploring the tuning in catalytic performance of Cat. A/B/C (Fig. 1) under different reaction conditions, such as reaction temperature (T), co-catalyst/catalyst ratio (Al/Ni ratio), co-monomer concentration ([1-hexene]) and reaction time (t) etc. Polymerization conditions and results (activity, M w, PDI, T g, T m, B) are summarized in Table 1.Catalyst activity was the most intuitive reflection of reaction conditions on the catalytic performance. Fig. 2 illustrated the reactivity of these three catalysts under different reaction conditions in detail and completely different catalytic performance was observed. Among them, the performance of Cat. A and Cat. B were more greatly affected by temperature and catalytic activity changed obviously from 25 °C to 75 °C [20] as shown in Fig. 2(a). The activity of Cat. A was increased from 1.8 × 105 g polymer/(mol Ni·h) to 4.04 × 105 g polymer/(mol Ni·h). As for the Cat. B, the activity dropped from 8.94 × 105 g polymer/(mol Ni·h) at 25 °C to 4.04 × 105 g polymer/(mol Ni·h) at 75 °C.Comparatively, the activity of Cat. C maintained at a low value (Fig. 2(a)), which might be due to the steric hindrance generated by the ortho-position substituents [21], and the blocked coordination process between metal active site and ethylene monomer. On the other hand, previous studies have already shown that a bulky catalyst with dibenzhydryl group as ortho-position substituent could only exhibit good thermal stability and reactivity under high pressure(≥4 atm) [22].Compared to the change in reactivity, effect of Al/Ni ratio on the activity was all in an increasing trend (Fig. 2(b)), which was similar to the results reported before [23], which might be due to the adequate reactant for the polymerization provided by co-catalyst promoter. When the Al/Ni ratio approaching to 1800, its effect on catalytic activity of Cat. B tended to be weakened due to enough cocatalyst existing. When the comonomer 1-hexene increased from 0 mol/L to 0.30 mol/L, the overall activity of the Cat. A increased (Fig. 2(c)). This increase in the catalytic activity of the catalyst could be attributed to the “co-monomer effect” [24]. However, Cat. B and Cat. C may have slightly reduced activity due to steric hindrance. Besides, Cat. C even maintained good thermal stability in the 2 h polymerization process (Fig. 2(d)), due to the hindrance of a bulky ligand.As the temperature increased continuously from 25 °C to 75 °C, T g value of poly(ethylene-co-1-hexene) in the presence of Cat. A decreased from −46.20 °C to −64.54 °C (Fig. 3 (a)). Cat. B showed a moderately rise till 75 °C, T g value of the product decreased from −57.09 °C to −67.40 °C. The absence of absorption peak in the DSC curves under this condition (T = 75 °C) indicated that the product was completely amorphous [20,25]. In general, the effect of catalyst Cat. C on the crystallinity of poly(ethylene-co-1-hexene) was most obvious with the increase of temperature (Fig. 3(a)), During this process, T g values of the product catalyzed by Cat. C also showed a slight downward trend, which was indirectly resulted by the unique "chain walking" ability of catalyst. In addition, T m decreased with the increase of branching density. Under the microscope, with the increase of e branching density of polyethylene, the number of active segments in the poly(ethylene-co-1-hexene) increases, so a lower temperature is needed to cease the movement of the segments.The effect of Al/Ni ratio on T g was not obvious. As shown in Fig. 3(b), T g values decreased while the Al/Ni ratio changed from 600 to 1800. The T m and T g values of the copolymer prepared by these three catalysts all decreased to a certain extent, and the trend gradually slowed down as the steric hindrance of the catalyst increased. Due to the conformation of Cat. C, poly(ethylene-co-1-hexene) with the highest crystallinity and T m values was tuning over a wide temperature range (58.40–75.12 °C). Similarly, T g and T m values of all the samples decreased slightly with the increase of the concentration of co-monomer 1-hexene (Fig. 3(c)), which may be closely related to the modification of branch topologies. It was shown from Fig. 3(d) that the effect of reaction time on the preparation of poly(ethylene-co-1-hexene) catalyzed by Cat. C was negligible. T m values of the obtained copolymers range from 69.38 °C to 73.44 °C. Besides, the reaction time didn't have a great effect on the phase transition temperature of products catalyzed by Cat. A and Cat. B.Obviously, M w values of the obtained poly(ethylene-co-1-hexene) decreased rapidly with the increase of reaction temperature, (Fig. 4 (a)). When catalyzed by Cat. B, the M w values of poly(ethylene-co-1-hexene) was as high as 2.08 × 105 g/mol at 25 °C, but only 8.78 × 104 g/mol at 75 °C. M w values of poly(ethylene-co-1-hexene) obtained with Cat. A and Cat. C also showed similar but smoother trends. This might be attributed to the fact that the chain transfer rate was greatly accelerated compared to the chain growth rate as the temperature increased [20].Among the catalysts involved, Cat. A and Cat. B was greatly affected by the tuning of Al/Ni ratio (Fig. 4(b)). In contrast, due to its large sterically hindered ligand structure and low reaction pressure, the M w values in the case of Cat. C remained a low value. In addition, as the concentration of 1-hexene increased from 0 to 0.3 mol/L (Fig. 4(c)), the product obtained in the presence of Cat. A had a significant increase in M w after further improvement of the co-monomer concentration (from 1.22 × 105 g/mol to 2.15 × 105 g/mol). Due to the influence of the catalyst N-aryl ligand "hugging" on the metal site and the steric effect of 1-hexene, the insertion of the monomer was restricted. The products obtained by Cat. B and Cat. C seemed to be slightly affected by the concentration of 1-hexene. The effect of reaction time is a good reflection of thermal stability, as shown in Fig. 4(d). In comparison, Cat. C got a clearer distinction in chain growth rate and chain transfer rate, which leading to an almost linear increase in the M w values of polyolefin obtained once got sufficient reaction time. To the best of our knowledge, the PDI of polyethylene obtained by alkyl backbone ligand catalyst was smaller than that of the corresponding acenaphthene backbone catalysts [26,27], which was basically same as that mentioned in Table 1: such as the PDI of the obtained polymerobtained in the presence of Cat. B (Table 1, Run 9. Cat. B) is higher than Cat. A (Table 1, Run 2. Cat. A).Reaction temperature was believed to affect the microstructure greatly. From 75 °C to 25 °C, the branch density of poly(ethylene-co-1-hexene) prepared by Cat. A, Cat. B, and Cat. C vary from 106 branches/1000C (Run 3 in Table 1 ) to 79 branches/1000C (Run 1 in Table 1), from 116 branches/1000C (Run 12 in Table 1) to 87 branches/1000C (Run 10 in Table 1), and 54 branches/1000C (Run 21 in Table 1) to 46 branches/1000C (Run 19 in Table 1), respectively.In order to detect the poly(ethylene-co-1-hexene) microstructure in detail, 13C NMR investigations were further conducted (Fig. 5 ). According to the resonance peak assignments in literatures [28–30], the branch distribution was quantitatively analyzed. The integrated area of 13C NMR spectra was consistent with the theory mentioned above. The order of the integrated l area of butyl and propyl groups (Cat. C < Cat. A < Cat. B) indicated Cat. B has the best catalytic activity under the same condition.In addition, the branch distribution of Cat. A/B/C under different temperatures (25/50/75 °C) has been described in Fig. 6 . It can be found that the increasing temperature had a great impact on the enrichment of branched polyolefin microstructures. As temperature increased, the branches distribution of samples obtained with the three catalysts changed greatly. First, proportion of methyl group on the polyolefin chain decreased gradually, and that of long-chain branches (more than five carbon atoms) increased steadily. At the same time, ethyl or propyl groups gradually present from absence, and the insertion of the co-monomer (1-hexene) promoted to the butyl group with higher proportion. Due to the bulky steric hindrance and the reaction conditions (low pressure), the proportion of intermediate groups of poly(ethylene-co-1-hexene) obtained by Cat. C was substantially zero proportion under the applied temperature (Fig. 6(c)).The microstructure of poly(ethylene-co-1-hexene) was vividly illustrated in Fig. 6(d) that the insertion of 1-hexene enabled us to obtain a more abundant branched structure [8,26,31]. For poly(ethylene-co-1-hexene) prepared with Cat. A and Cat. B, it was especially noticeable that the proportion of butyl increases significantly. However, when Cat. C with significant steric hindrance enhancement was used, poly(ethylene-co-1-hexene) with abundant short-chain microstructure has never been obtained (Fig. 6(c)). Furthermore, this phenomenon confirmed the existence of the “primary insertion” and “secondary insertion” stage (Fig. 7 ). It was suggested that the insertion of the monomer demonstrated a preference for the “primary insertion” stage [32]. Comparatively, the activity of the catalyst would be largely affected by the steric hindrance in the “secondary insertion” stage [33]. Thereby, the monomer might prefer to insert into the primary alkyl species.During the polymerization process, the metal active site at the end of chain may undergo one-step β-H elimination and subsequent monomer reinsertion. Once the chain is grown at this active site, methyl group in the main chain would be formed. Similarly, as the polymerization process continues, after two-step β-H elimination and the following monomer reinsertion, a nickel-active species with α-ethyl groups can be formed, repeatedly forming a branched polyethylene structure. But Pei et al. believed that it was difficult to form a branched group with larger steric hindrance, such as ethyl group [34]. At this point, it coincided with the previous view that the metal active site becomes inactive once it inserts into the polymer chain [35] (i.e. in the “secondary insertion” stage). In addition，Pei and co-workers described the mechanism by which long chain branches form in detail [34]. Branch distribution shown in Fig. 6(d) reached a good consensus with Pei's research.In order to better understand the internal changes of catalytic system and explore the role of MAO in Cat. B/MAO catalytic system, EPR analysis under different Al/Ni ratios was carried out (Fig. 8 (a)). Only Cat. B or MAO in toluene had no signal. However, with the addition of the co-catalyst (MAO) in the catalyst solution, an unpaired electron signal (g = 2.218) belonging to Ni(Ι) species was obviously detected [36], accompanied by a radical signal (g = 2.002) [36]. Due to the interaction of the excessive Lewis acid MAO with the electrophilic acenaphthene-based ligand, unpaired electrons were attracted or captured from metal active center to form a ligand-based and carbon-centered radical, thus, the coordination process between metal active centers and olefin monomers was hindered or weakened to some extent (Fig. 9 ) [37,38], and the effect of this process becomes more obvious with the increase of Al/Ni ratio (the increasing trend of catalyst activity decreased). Most interestingly, as the Al/Ni ratio increased, the intensity of signal (g = 2.218) was significantly enhanced. This means that the concentration of Ni(Ι) species would increase with the growth of Al/Ni ratio. It demonstrated that the existence of inactivated Ni(ΙΙ) species was motivated by MAO and promoted the olefin polymerization. Because this effect on the performance of the catalyst is more significant than that of being reduced to Ni(I), the catalytic activity of the system increases with the increase of the Al/Ni ratio.When Al/Ni ratio was adjusted from 600 to 1800, the signal intensity of Ni(Ι) species decreased (Fig. 8(b)). Considering the fact that the activity of the catalyst increases slowly, it was believed that MAO played the role of reductant more because of the large excess. In other word, more Ni(ΙΙ) species were further reduced to zero oxidation state after conversion to Ni(Ι) species. This view was also confirmed by the appearance of black particle precipitation in the solution [35].In this paper, the catalytic performance of three α-diimine nickel catalysts with different ligands for ethylene/1-hexene copolymerization under different reaction conditions. The results showed that the external reaction conditions (temperature, Al/Ni ratio, [1-Hexene], reaction time) had great influence on the polymerization process. Different ligand structures led to the effect of external conditions differ from one to another, but in short, changes of reaction temperature showed most significant influence on the catalytic performance. Specially, the proportion of the "branch-on-branch" structure (sec-butyl group) of the polymer chain also increased rapidly, well indicating the tuning of the “chain-walking” ability. The main reason for the changes of catalytic performance was the relative rate of chain growth and chain transfer. On the other hand, the change in Ni species was also noticeable. The introduction of co-catalyst also played a role of reductant to a certain extent, and promoting the olefin polymerization process.The authors declare that they have no competing interests.Financial support from the National Key Research and Development Program (2016YFB0302403) is gratefully acknowledged. Besides, we would like to thank Dr. Shengyu Dai for providing the catalysts in this research.

The structure of polyolefin has an important influence on its performance and application. Ethylene/1-hexene copolymerization is one of the important ways to control the structure of the polyolefin. However, research on the ethylene/1-hexene copolymerization catalyzed by nickel complexes with different steric ligands remains to be refined. Here, three α-diimine nickel catalysts are used to study the ligand effect on catalytic performance in the ethylene/1-hexene copolymerization. Reaction activity, molecular weight, phase-transition temperature and branching density of the resultant copolymer are measured to evaluate the catalytic performance. The results indicate that the steric ligands could exert great effect on the copolymerization. As for the chemical valence of Ni species, detailed EPR demonstrate that the presence of excess co-catalyst can reduce Ni(II) to the lower valence and affect the catalytic performance.

In the industry, the catalytic hydrogenation of edible oils is typically carried out in a slurry reactor through a semi-batch process, with H2 gas injected at an elevated temperature and pressure [1]. One of the most imperative aspects of the process is the usage of a catalyst to catalyse the reaction. Although Ni-based catalysts are vastly applied in industrial processes due to their superior performance and cost effectiveness, constant improvement is vital to further enhance the activity and selectivity of the catalysts. Various preparation parameters have been studied e.g. support type, active metal content, reduction temperature, presence of doping agents, synthesis methods, etc. [2–5]. To illustrate, some of these properties such as the pore volume, pore length, particle size, total surface area and Ni crystallite size play a crucial role in determining the activity and selectivity of the catalysts [6,7]. In particular, a high pore volume, shorter pore length, smaller particle size, larger total surface area and smaller Ni crystallite size would potentially benefit the catalytic activity of the catalyst [8,9].The co-precipitation of a Ni salt and a silica source using an alkali source such as sodium carbonate is one of the common and conventional techniques to synthesise Ni catalyst precursors i.e. silica-supported Ni carbonate. The aforementioned substance yields supported Ni catalysts upon reduction [10]. Nitta and colleagues have concluded that compounds of Ni, particularly hydroxide or carbonate compounds of Ni, tend to form strong interactions with the silica carrier or support. Consequently, this results in the generation of Ni hydrosilicate phases, particularly nickel phyllosilicates, on the external layers of the support phase, which function as anchoring sites for Ni particles, leading to their stabilisation and dispersion [11]. This phenomenon is also supported by various other work concerning the synthesis of silica-supported Ni catalysts [2,12–14]. The presence of these metal-support interactions (MSI) could potentially influence the properties and performance of heterogeneous supported catalysts in terms of their active metallic area and dispersion, reducibility and stability or resistance to thermal sintering, and subsequently the activity and selectivity [3,15,16]. As described in some studies, many factors could influence these properties, including the catalyst synthesis technique. For instance, it was reported that the ageing step is of substantial significance in dictating fundamental catalyst properties, which controls the degree of formation of less reducible silicates, therefore affecting final metallic dispersion and surface area properties [17,18]. In particular, ageing time and ageing temperature were reported to play an important role in tuning nickel silicate formation, ultimately influencing catalyst activity and applicability [18].Among numerous catalyst synthesis techniques, ultrasonic technology has emerged over the years as an innovative method to effectively modify the properties and performance of heterogeneous catalysts [19–21]. The incorporation of ultrasonic irradiation into catalyst synthesis could result in various changes in particle morphology, surface composition, metal dispersion, structural or geometric properties, electronic configurations and catalyst reactivity [19,22,23]. In some cases, the use of ultrasound is able to activate less reactive, but also less costly, catalytic metals [24]. To illustrate, in liquids irradiated by ultrasound, the phenomenon known as acoustic cavitation induces the formation and subsequent implosion of numerous short-lived micro-bubbles with extremely high temperatures and pressures [25]. These transient, localised hot-spots facilitate various physical and chemical reactions during catalyst synthesis, which brings about the enhancement and promotion of nucleation rates and dispersion of active metals on the support surface [26]. Nevertheless, it is also imperative to understand the underlying mechanisms involved in driving the positive effects demonstrated by ultrasonic irradiation in catalyst synthesis, which is an aspect that many literature sources lack. For instance, in various catalysts synthesised for photocatalytic, catalytic cracking and gas reforming applications, authors have reported the increase in reducibility, metal dispersion, particle uniformity, BET surface area etc. [19,27–34] with the use of ultrasound during synthesis but detailed explanations for the reason behind the improvement were generally lacking. It is also important to note that performing routine calibration experiments is imperative to accurately reflect the actual acoustic power dissipated by the ultrasonic source, which could allow effective replications as well as comparisons between different bodies of work. This in fact is a critical aspect that a large majority of literature sources lack, whereby only the electric powers specified by the ultrasonic generators (as provided by manufacturers) are detailed [19,27–29,32,34,35].Prompted by this knowledge gap, this present work aims to study the phenomenon and mechanisms that take place during the synthesis of sequentially precipitated catalysts with ultrasound employed during the ageing step. Specifically, nickel-silica hydrogenation catalysts were synthesised via sequential precipitation, with ultrasonic irradiation applied at varying ultrasonic intensities during the ageing step, whereby the actual acoustic powers supplied were determined. The catalysts were evaluated based on several characterisation tests, in which the variation in catalyst phase compositions due to sonication was analysed and the ultrasonic irradiation synthesis mechanism was outlined. To further appraise the performance of the synthesised catalysts, the partial hydrogenation of sunflower oil was carried out to ascertain the activity and selectivity of the catalysts.Chemicals used in this work were of analytical grade, obtained from R&M Chemicals Malaysia, utilised without supplemental purification. The ultrasonic system employed in this study was fitted with a 20 kHz probe. The diameter of the probe tip was approximately 1 cm (Sonics and Materials, VCX 750, 750 W). O’life Sunflower oil (Sime Darby Food and Marketing Sdn. Bhd., Malaysia) was employed as the feed for the hydrogenation reaction. The iodine value (IV) of the feed was 124, with its composition listed in Table 1 .To obtain the calorimetry data for the ultrasonic system used, a beaker of 200 ml deionised water was subjected to irradiation with ultrasonic amplitudes of 20, 30 and 40%, total duration being 30 min. The temperature of the system was not regulated. The power output of the ultrasonic system, Q, was obtained using the following equation: (1) Q = m C p Δ T Δ t where m is the total mass of water, Cp specific heat capacity of water (4.18 kJ kg−1 °C−1), ΔT/Δt (°C s−1) is the temperature gradient. Temperature readings were repeated at 30 s intervals and taken at three varied positions in the beaker, noting the mean value.To ascertain the heat dissipated by the system, water at an equal volume to the ultrasonic set was heated using a portable electric heater of 1000 W, while being stirred. Similarly, the process took 30 min and readings were repeated at 30 s intervals and taken at three varied positions in the beaker, noting the mean value. The power output due to heating, QH , can be acquired using Eq. (2). Hence, the heat dissipated, QHL , can be obtained using the following equation: (2) Q H L = Q H - Q Subsequently, the acoustic energy intensity provided by the probe, IUS , is calculated via the following equation: (3) I U S = Q + Q H L A where Q + QHL is the output power, Pout , and A is the cross-sectional area of the surface producing ultrasonic waves, determined as 0.8 cm2 for the ultrasonic probe.In addition, the ultrasonic density provided by the probe to the liquid body, ρUS , is calculated via the following equation: (4) ρ U S = Q + Q H L V whereV is the volume of water (cm3) in the beaker used for the calorimetry test.Solutions of nickel sulphate hexahydrate (Ni(SO4)2·6H2O) and magnesium sulphate heptahydrate (Mg(SO4)2·7H2O) were mixed with a molar ratio of 3:1 and subsequently heated to 50 °C. Then, 10 wt% sodium carbonate (Na2CO3) was dosed until the pH of the precipitated suspension was 8.8, in a duration of 10 min. The suspension was then heated to 90 °C, followed by the addition of 2 wt% sodium metasilicate pentahydrate (Na2SiO3·5H2O) solution, in a duration of 10 min. The suspension was then allowed to age for 30 min at 90 °C. Throughout the whole synthesis process, the suspension was stirred constantly. After terminating the synthesis procedure, the resulting mixture was filtered and thoroughly rinsed using deionised water three times. Next, the filtered precipitate was dried at 100 °C for 5 h in an oven. Lastly, the dried precipitate was calcined in a chamber furnace at 400 °C for 4 h. The unsonicated sample is labelled as A.For catalysts subjected to ultrasonic irradiation, the dosing of 2 wt% Na2SiO3·5H2O solution was delivered under the influence of a sonicator, in the same duration of 10 min as the unsonicated sample for consistency. The probe was immersed at a depth of 2 cm, pulsed at 2 s on/off, under temperature-controlled conditions. Three sonicated catalysts were synthesised with ultrasonic amplitudes of 20, 30 and 40%, corresponding to ultrasonic intensities of 7.07, 20.78 and 27.72 W cm−2, labelled as B, C and D, respectively. The aforementioned ultrasonic amplitudes were chosen considering their interesting impacts on catalyst synthesis in a previous work [36].The morphology of the samples was characterised with a field emission scanning electron microscopy (FE-SEM, Fei Quanta 400F). SEM scans were obtained with a Fei Quanta 400F microscope. The beam current was 1 µA and the accelerating voltage was 20 kV.A field emission scanning electron microscope (FE-SEM) fitted with an Oxford Instruments X-MAX energy dispersive X-ray (EDX) analyser was used to obtain the elemental information of the catalysts. The accelerating voltage was 20.0 kV with an acquisition live-time of 45 s.The PANalytical X’Pert-PRO diffractometer was used to obtain diffractograms for the catalysts. Cu-Kα X-ray radiation used was of the wavelength 1.54060 Å. The beam current was 40 mA. While the voltage was 45 kV. Crystallite sizes of the samples was ascertained via the Debye-Scherrer’s formula.The specific surface areas (SBET) and the porosities of the catalysts were acquired with the Micromeritics 3Flex Surface and Catalyst Characterisation Analyser by using the Brunauer-Emmett-Teller method. Adsorption and desorption runs were executed at −195.681 °C (77 K) using N2 gas. The catalysts were degassed at 150 °C for 4 h before the adsorption experiments.Temperature programmed reduction was carried out with the Micromeritics AutoChem II 2920 chemisorption analyser fitted with a thermal conductivity detector. Approximately 30 mg of the sample were carefully inserted into the quartz U-tube reactor and then placed in the tubular furnace. Catalysts were subjected to pre-treatment under Ar flow at 20 cm3/min, with the temperature raised from ambient to 100 °C and held for 60 min to remove physisorbed and/or weakly bound species. Subsequently, the catalysts were analysed in 9.47% of H2/Ar at a flow rate of 25 cm3/min from ambient to 900 °C at a heating rate of 10 °C/min. The vapour produced is removed via a cold trap filled with chilled coolant. H2 adsorbed by the catalysts was detected by the thermal conductivity detector and ascertained via the peak area of the H2-TPR profile. The TPR curves were further processed and deconvoluted using the OriginPro software via Gaussian multi-curve fitting.Pulse chemisorption analysis was conducted using the Micromeritics AutoChem II 2920 chemisorption analyser. Samples were first degassed in an inert argon gas flow for 60 min at 100 °C. Then, the catalysts were reduced at 500 °C in H2 for 2 h and degassed once more for 30 min prior to the chemisorption analysis. For the analysis, hydrogen gas was pulsed every 6 min until no further uptake was detected. The Ni metal dispersion (%) is calculated with the following equation: (5) % D i s p e r s i o n = ( V m V m o l M % W a ) ( F s ) where Vm is the volume of hydrogen gas chemisorbed (cm3/g STP); Vmol is the molar volume of the adsorptive (cm3/mole STP); M% is the percentage of Ni metal by weight as grams of Ni per gram of sample; Wa is the atomic weight of Ni (g/mole); Fs is the stoichiometry factor, taken as 2 for hydrogen on Ni.The active metal surface area (m2/g) is calculated with the following equation: (6) M e t a l s u r f a c e a r e a = F s n a N A A g where Fs is the stoichiometry factor, taken as 2 for hydrogen on Ni; na is the number of moles of gas adsorbed (cm3/g STP); NA is Avogadro’s constant; Ag is the cross-sectional area of the active adsorptive atom (nm2) (with the assumption that a single Ni atom occupies 0.0649 nm2) [37].The average metal particle size (nm) is calculated with the following equation [38]: (7) A v e r a g e p a r t i c l e s i z e = 6 ( A S m ) ( % D i s p e r s i o n ) ( ρ ) X 100 where ASm is the active metal surface area (m2/g); ρ is the density of the metal (g/cm3).X-ray photoelectron spectroscopy (XPS) was conducted with the JEOL JPS-9030 photoelectron spectrometer. Al Kα (1486.6 eV) was used as the excitation source to probe the sample surface information at a depth of 1 – 12 nm. The pressure in the analysis chamber during experiments was less than 5 × 10−10 Torr. A hemispherical electron-energy analyser working at a pass energy of 30 eV was used to collect core-level spectra. The samples were dispersed in ethanol and placed on silicon wafers, which were mounted on a sample holder and directly transferred into the analysis chamber. Step size was adjusted to 0.1 eV, dwell time was set at 100 ms and the high-resolution spectra were recorded with 10 scans. Charge effects were corrected by using the C 1 s peak at 285 eV. A Shirley background was applied to subtract the inelastic background of core-level peaks. The model peak to describe XPS core-level lines for curve fitting was a product of Gaussian functions. The XPS spectra were processed and deconvoluted using the OriginPro software.Atomic absorption spectroscopy (AAS) was conducted with a Perkin Elmer AAnalyst400 AA spectrometer to obtain the concentration of leached nickel from the hydrogenated oil samples. Sample readings were taken three times to obtain an average value.Prior to carrying out the hydrogenation reaction tests, the catalyst samples involved were reduced in pure hydrogen gas for activation. The samples were reduced in a tubular furnace. Before reduction, the samples were heated to 500 °C with a heating rate of 10 °C/min in N2. Upon reaching 500 °C, the samples were subjected to a flow of hydrogen gas and held for 2 h. Lastly, fully hydrogenated palm stearin (iodine value <0.5) was used to coat the catalysts to prevent oxidation, thus forming fat-coated catalyst granules of 22 wt% Ni.Partial hydrogenation was carried out in a 1.5 L pressurised batch reactor (Buchiglasuster Eco-Clave). The temperature of the reactor was regulated by a high precision temperature regulatory system (Huber Unistat Tango Nuevo), with the use of a silicone oil jacket around the circular reactor and a temperature probe inside the reactor. Briefly, 750 ml of sunflower oil was added into the reactor with the aid of a vacuum pump and heated to a set-point temperature of 180 °C. Prior to dosing the catalyst, the reactor was vacuumed to remove any air or moisture that will poison and deactivate the catalyst. Subsequently, a catalyst dosage of 2 g/L was used and hydrogen gas of 5 barg was charged into the reactor to initiate the reaction, while the slurry was vigorously stirrer at 1500 rpm. Oil samples (4–5 ml) were taken from the bottom of the reactor for each time interval. The reaction was carried out for 90 min. Obtained samples were labelled and filtered using filter papers to remove the catalyst particles.The iodine value measures the degree of unsaturation of fats and oils, which represents the mass (g) of iodine consumed per 100 g of oil. Products collected from the reactor were subjected to iodine value tests, in accordance with the American Oil Chemist’s Society (AOCS) Official Method Tg 1a-64, performed 3 times to obtain average values. The resulting iodine value was calculated with the following formula: (8) I o d i n e v a l u e = B - S × N × 12 . 691 m a s s o f s a m p l e , g where B is the volume of titrant of blank set, mL; S is the volume of titrant of sample set, mL; N is the normality of the sodium thiosulphate solution.Gas chromatography was carried out in order to ascertain their respective fatty acid compositions of the reaction products. A PerkinElmer Clarus 500 gas chromatograph fitted with a PerkinElmer COL-ELITE-2560 capillary column (100 m × 0.25 mm ID × 0.20 μm df). Helium was utilised as the carrier gas at 1.3 ml/min. Prior to the tests, esterification of the hydrogenation products was carried out to convert them into fatty acid methyl esters (FAME). The oven, injector and flame ionisation detector (FID) temperatures were 175, 210 and 250 °C, respectively. The injection volume was 1 μL and the split ratio was 100:1. The relative areas of the fatty acid peaks obtained from the chromatogram were used to determine the product distribution of the oil samples.The ultrasonic system was assessed for its calorimetry and results are presented in Table 2 . The heat loss, QHL , calculated was 0.8 W.The morphology of the catalysts was interpreted via the SEM scans in Fig. 1 . While the non-sonicated catalyst exhibited rough surfaces with the presence of conglomerates, the morphology of the sonicated catalysts all appeared to be smooth and uniform, which confirmed the capability of ultrasound to reduce the extent of particle aggregation [39]. As demonstrated later, the combination of ultrasound irradiation and co-precipitation generated a synergistic effect whereby a uniform environment was provided for the nucleation and growth of metal particles, while concurrently averting small particles from agglomerating [40], thus preventing the formation of aggregates such as those in the non-sonicated catalyst. In particular, the increased ultrasonic intensity instigated more violent micro-bubble implosions, hence inducing stronger shock waves and micro-jets with velocities of approximately 400 km/h. Consequently, vigorous inter-particle collisions and surface pitting could occur, simultaneously suppressing the aggregation of catalyst particles, thus producing particles with increased dispersion [41]. This observed phenomenon resonated with the work of other researchers, in which a lower degree of particle agglomeration was noticed with the use of ultrasound during synthesis [40,42].From the EDX and XPS analyses of the synthesised catalysts in Table 3 , it is apparent that they were constituted of the required elements to form the catalysts, which indicates the efficacy of the synthesis procedure. As the typical approximated penetration depth of an EDX electron beam is 0.4 μm [43], the measured atomic percentage was considered to represent the bulk atomic percentage of the catalysts. On the other hand, characterisation by XPS detects the elemental composition of the samples at the surface layer, which corresponds to a penetration depth of approximately 1 – 12 nm, hence demonstrative of the superficial composition of the catalysts [43]. As the catalysts investigated were all synthesised according to the same starting materials, it will be useful to draw comparisons based on their bulk and surface atomic distributions to ascertain the changes due to the difference in synthesis conditions. As shown in Table 3, there were decreasing trends for the Mg/Ni and Si/Ni atomic ratios with the increase in ultrasound intensity during the ageing process from data obtained from both EDX and XPS analyses. It is worth noting that while the difference was small for the Mg/Ni ratios between the EDX and XPS methods, there was a substantial gap in the Si/Ni ratios between the two aforementioned methods. We believe the observed change in Mg/Ni and Si/Ni atomic ratios was mainly due to the effect of ultrasonic intensity (or lack thereof) on the sequentially co-precipitated Ni-Mg hydroxycarbonates and siliceous materials during the ageing process.A conceptualisation demonstrating the action of ultrasound during the ageing process is presented in Fig. 2 . It is herein noted that, for the sake of accuracy, prefixes such as hydro- or hydroxy- are used to represent the actual phases of the precursors present during synthesis. After the synthesis process, the aforementioned phases were subjected to dehydration and calcination and subsequently transformed into their respective oxides or silicates, which were then used for characterisations mentioned in this present work. In conventional synthesis of sequentially precipitated catalysts without the presence of ultrasound, the formation of Ni silicate proceeded as the silica precursor was added during the ageing process, which occurs under an alkaline condition at elevated temperatures. Prior to the ageing process, co-precipitates of Ni-Mg hydroxycarbonates were present in the suspension. As the ageing process was initiated, silicate ions (in the form of Na2SiO3) were dosed into the ageing solution, which then: (i) attached epitaxially to the co-precipitates [44], forming nickel hydrosilicates, in which Ni-Mg hydroxycarbonates would then be anchored; (ii) attached to existing hydrosilicates and underwent polymerisation to form silica clusters [45]. This attachment of silicate ions is illustrated by Step 1 in Fig. 2. In the presence of ultrasonic irradiation, the Ni-Mg hydroxycarbonate co-precipitates can experience erosion, causing the shrinkage of the size of the co-precipitates, which will eventually lead to higher Ni dispersion as we will see in Table 4 . The increase in ultrasound intensity also led to lower Mg/Ni ratios, beginning with sample A at 0.329 (theoretical ratio = 0.333 with Ni:Mg = 3:1) and down to 0.268 for sample D. During the formation of Ni-Mg hydroxycarbonate nanoparticles, Mg has the tendency to migrate to surficial layers due to its lower surface energy compared to Ni. It is noted that Ni has a surface energy of 2.080 J/m2, while Mg has a surface energy of 0.688 J/m2 [46]. As a result, it is quite likely that a significant amount of Mg was eroded away from the surface layers of Ni-Mg hydroxycarbonate co-precipitates when the ultrasonic intensity was increased. As shown in Table 3, the difference in the Mg/Ni ratio for the surface and bulk (XPS – EDX) decreased as the ultrasonic intensity increased from samples A – C, indicating the increased extent of Mg erosion from the catalyst surface. However, sample D registered an increase in the Mg/Ni ratio difference between the surface and bulk, likely due to the re-attachment of the Mg ions at higher ultrasonic intensities, as similar investigations reported the inefficiency of ultrasonic irradiation at high powers causing the re-agglomeration of particles [47]. In a similar manner, higher ultrasound intensity has also disrupted the formation of hydrosilicates and silica clusters and led to the drop in the bulk Si/Ni ratio, from 0.333 to 0.244.Following the attachment of silicate ions to form metal hydrosilicate and silica clusters, the individual silicate-attached Ni-Mg hydroxycarbonates would clump together through the -O-Si-O- bonding provided by silicate ions in the ageing solution [45,48]. Small aggregates begin to form from the clumping of nano-sized silicate-attached Ni-Mg hydroxycarbonates (Step 2 of Fig. 2), while further clumping of small aggregates lead to bigger aggregates, eventually forming the catalyst particle network (Step 3 of Fig. 2). The significantly higher Si/Ni ratios as observed by XPS (from 0.468 to 0.333) compared to EDX (from 0.333 to 0.244) was mainly a result of silicate ions attaching to and growing on the outer surface of the aggregated catalyst particles as silica clusters. Attachment of silicate ions within the pores of the catalyst particles is thought to be insignificant compared to the outer surface due to slower mass transfer in the narrow pores. In the present work, ultrasound irradiation was only active during the first 10 min of the ageing process for the sonicated samples, while the remaining ageing period was carried out without ultrasonic irradiation. As higher surface Si/Ni density was also observed for catalyst samples sonicated with high intensity, it can be inferred that the growth of silicate ions on the surface of the catalyst particles took place predominantly following the termination of irradiation, since a high ultrasonic intensity can disrupt the formation of hydrosilicate and silica clusters. It should be noted that the discrepancy between the bulk and surface layer Si/Ni ratios in this present work was in stark contrast with the typical synthesis method of simultaneously co-precipitating the active and support phases. Compared to the sequential precipitation method used in this work, whereby the support phase was added after the co-precipitation of active phases, the simultaneous co-precipitation of both active and support phases could result in a more uniform and homogeneous distribution of the elements [49,50], thus the clustering of the elements would be less drastic in that case.The XRD diffractograms for the catalysts are shown in Fig. 3 . Peaks representing the phases of NiO (ICDD: 01-073-1523), MgO (ICDD: 01-089-7746) and nickel silicate (Ni3Si2O5(OH)4) (ICDD: 01-083-1648) were detected. Diffraction patterns of all samples exhibit reflections at the 2θ value of 36°, 43°, 63°, 75° and 79°, which correspond to NiO(111), NiO(200), NiO(220), NiO(222) and Ni(311) planes, respectively [51,52]. These results agreed well with those of several other investigations [53,54]. The SiO2 support was amorphous, which did not contribute to any peaks on the diffractogram [55,56]. The NiO crystallite size was ascertained using the Scherrer equation, referencing the characteristic peak at 43°, with the results presented in Fig. 3. In general, the diffractograms registered reflections that appeared broad, which signified the poorly crystallised state of the catalysts. This also indicated the presence of small and well-dispersed phases and highly developed surfaces [57,58]. In addition, the lack of sharp peaks also indicated the absence of large crystalline domains or metal oxide clusters, which could be of great advantage for metal dispersion during catalyst reduction [59]. On the other hand, the presence of nickel silicate, formed by a strong interaction between Ni and Si, is also detected at the 37° peaks. Poorly crystallised nickel silicate compounds with imperfect nickel antigorite structures are commonly formed when the co-precipitating nickel salt and silicate solutions are at temperatures lower than 100 °C, with structures such as Ni3Si2O5(OH)4 [60]. Due to the incorporation of ultrasound during synthesis, the samples registered an increase in amorphicity, evidenced by the decrease in crystallite size of 2.56 nm for the unsonicated catalyst, to a range of 2.28 – 2.41 nm, for the sonicated catalysts. Pertaining to the sonicated catalysts, the crystallite size decreased with the increase in ultrasonic intensity. This may be owed to the highly turbulent mixing induced by sonicating the suspension during synthesis, in which unique conditions due to acoustic cavitational bubble collapse were produced. This highly turbulent mixing is facilitated by the presence of acoustic streaming and high mass transfer [61] and subsequently resulted in a higher packing disorder in the samples, forming samples that were more amorphous [25]. Furthermore, as the cavitation bubbles collapsed violently, localised regions of extremely high cooling rates, in the range of 1011 K/s, were generated, which could inhibit the growth of crystals [25].H2-TPR was used to probe the redox properties of the unsonicated and sonicated catalysts, with results presented in Fig. 4 . The reduction profiles of the catalysts are characterised by multiple reduction peaks fused together into a much broader reduction band from 100 to 800 °C, with a peak ca. 400 – 500 °C. This observation denotes the homogeneous distribution of small particles over the support, with the presence of complex and intimate interactions between the active phase and support, which supports the XRD diffractograms and is also in accordance with other researchers [6,45,57,62], as well as Ni catalysts prepared via other techniques such as sol–gel [63]. From the figure, it is observed that the unsonicated catalyst had the highest maxima of 448.9 °C, while the sonicated catalysts had maximum reduction peaks in the range of 431 – 435 °C. Moreover, sample D, irradiated with the highest ultrasonic intensity presented the lowest reduction peak at 431.4 °C. The presence of ultrasound during synthesis has marginally increased the reducibility of the catalysts, which is also a finding supported by other relevant investigations concerning ultrasound-assisted catalysts synthesis [29,64]. The increase in reducibility also signified the decrease in extent of metal-support interactions, in which phases with weaker interaction with the support (i.e. NiO) would be more readily reduced to Ni0.To gain more insight into the TPR results, the broad peaks were deconvoluted into symmetrical peaks via Gaussian multipeak curve-fitting to evaluate the relative percentage of the different reducible species, as well as their peak positions, presented in Fig. S1 (Supplementary information). In total, three main peaks could be identified, which were labelled as α, β and γ. The reduction temperatures reported can be classified into different types of nickel species, which are directly associated with the degree of interaction with the silica support. This also indicates that Ni species of different extents of interaction coexist within a catalyst sample. Henceforth, the three peaks are classified based on their reducibility and explained based on their relative proportion and individual peak position. It is noted that no reduction peaks corresponded to the reduction of Mg species due to their extreme difficulty in reduction under the present conditions applied [59]. The first low temperature peak (α) at ca. 181 – 198 °C represented a minor portion approximately 6% or less of the total reduction profile. These are ascribed unambiguously to the reduction of higher Ni(III) oxides in trace quantities (Ni2O3 to NiO), which is a common occurrence for precipitates calcined at temperatures 400 °C or lower, as per this present study [13,63]. However, since the XRD results exhibited no obvious Ni2O3 signals, it is suggested that the Ni2O3 phase present was too sparse in amount and also highly dispersed on the support [52]. On the other hand, the β peaks at ca. 411 – 422 °C were due to the NiO species possessing weak interactions with the silica support [65,66]. Lastly, the γ peaks at ca. 495 – 514 °C were due to the Ni species having strong interactions with the silica support, hence forming Ni silicate, which presented difficulty in reduction [65,66].Based on Table 4, one can see that the composition of α for all samples remained in the range of approximately 3–6%, representing their minor role in the catalysts. However, the percentages of β and γ varied significantly with the presence of ultrasound and its intensity. In general, the weaker NiO phase, β, observed an increase in proportion with the increase in ultrasound intensity, from 32.1% in the unsonicated catalyst and up to 42.5% in the sonicated counterparts. Accordingly, the Ni phase with stronger interactions, γ, also observed a declining trend with the increase in ultrasound intensity, registering 61.7% in the unsonicated catalyst and down to 53.4% for sample D, which suggests the possibility of Ni silicate erosion or inhibition by ultrasonic irradiation. Interestingly, the increase in ultrasonic intensity from 20.78 to 27.72 W cm−2 did not lead to significant variations in the relative percentages of β and γ, which indicates that there is a limit to the extent of Ni silicate erosion caused by ultrasonic irradiation. Considering the bond linkages of -Si-O-Ni- present in Ni hydrosilicate species [45] and the bond dissociation energy of Ni-O as 391.6 kJ mol−1 [67], calculations have shown that it is in fact plausible for acoustic micro-bubble implosions to generate sufficient energy for the bond breakage of Ni-O [68], which erodes away the Ni hydrosilicate structure . On the other hand, with a significantly higher bond dissociation energy at 440 kJ mol−1 for the Si-O bonds present in the bond linkages of -Si-O-Si-, it became more difficult to instigate bond breakage once silicate ions were attached to the hydrosilicates to form silica. Consequently, this allowed the silica clusters to act as a protective layer, enclosing the Ni silicate phases to impede its erosion caused by ultrasonication. Instead of Ni silicate erosion, it appeared that the increasing ultrasonic intensity has caused considerable agitation that suppressed the build-up of silica clusters by deterring the polymerisation of silicic acid [69], leading to the decreasing trend of Si/Ni ratios as shown previously in Table 3.Apart from the compositional change in Ni phases in the catalyst, the increase in ultrasonic intensity has also resulted in the shrinkage of Ni-Mg hydroxycarbonate nanoparticles, leading to higher Ni surface area and dispersion. In particular, Ni dispersion was increased from 8.79% to 17.81%, while the Ni surface area was increased from 58.55 m2/g Ni to 118.53 m2/g Ni, as seen in Table 4. This suggested that Ni was more dispersed across the support due to the action of ultrasonic irradiation during synthesis, whereby the increase in intensity amplified this phenomenon. As discussed earlier, the improvement in Ni dispersion and surface area could be due to the erosion of Ni-Mg hydroxycarbonates caused by ultrasound irradiation, with higher intensity lead to more severe erosion. Given that the pH remains unchanged at 8.8 during the application of ultrasound irradiation, it is thought that the eroded Ni and Mg ions would re-precipitate and form smaller hydroxycarbonate nanoparticles, which in turn lead to higher Ni dispersion and Ni surface area. Generally, the increase in total metal surface area and dispersion are good indicators of enhanced catalytic activity, as more active sites equate to more area available for reaction to occur.The chemical and electronic properties of the calcined catalysts were studied using XPS. As per Fig. 5 , the spectra show features associated with Ni 2p3/2. For all catalysts, the Ni 2p3/2 is accompanied by the presence of a satellite peak at 861 – 863 eV, congruous to the existence of Ni2+ instead of metallic Ni0 [70]. It is noted that these satellite peaks manifest due to the paramagnetic state of the Ni2+ species and electron shake-up. Notably, these peaks are typically ca. 6 eV higher than the main Ni 2p3/2 peak [71], as noted by the ΔEsat values in Table 5 .The Ni 2p3/2 binding energy of the unsonicated catalyst resonated well with reported values of 856.3 – 856.7 eV [72,73]. However, it is discovered that the Ni 2p3/2 binding energy of the sonicated catalysts experienced a shift to lower values at 855.7 – 855.9 eV. Since XPS analysis is surface sensitive, the discrepancies in binding energies indicated that the Ni species on the surface have been altered electronically [2], in this case due to the incorporation of ultrasound into the synthesis procedure of the catalysts. Furthermore, a decrease in binding energy also indicated a lowered extent of metal-support interaction between the Ni phase and Si phase [2,74]. In this case, a shift to lower binding energies due to ultrasound has also been reported by other researchers [75], which led to a change in electron density that may affect the bonding of chemical intermediates to the active sites, thus ultimately affecting the reaction pathways during hydrogenation [76–78]. Furthermore, this indication of weakening in metal-support interactions is also corroborated with the TPR studies demonstrated earlier, noting the shift in main reduction peaks to lower temperatures in sonicated samples.To gain more insight into the XPS results, the Ni 2p3/2 XPS core level region was deconvoluted via Gaussian multi-peak curve-fitting and fitted with three doublets assigned to NiO, Ni silicate and associated satellite features, as shown in Fig. 5. As the NiO binding energies have been reported to be significantly lower at a range of 854 – 855 eV [72,73,79], it is suggested that another Ni phase exists at a higher binding energy, which is assigned to the Ni silicate phase. Hence, this agreed well with the two Ni phases detected in the XRD and TPR analyses. In addition, it is known that a relationship can be derived using the difference between Ni 2p3/2 and Si 2p binding energies, ΔENi-Si [80]. The existence of Ni silicates in the catalyst sample would give ΔENi-Si values of 753.2 – 753.8 eV [73,81,82]. From Table 5, the ΔENi-Si values of the catalysts synthesised agreed excellently with this, thus substantiating the presence of the aforementioned phases. Literature has also reported similar binding energy values of nickel silicates [73].The total area of respective signals, when averaged over the whole system, allows one to approximate the relative amount of species considered [83]. In this case, the area under the curve of each deconvoluted peak in Fig. 5 were used to estimate the relative surface proportion of each nickel species, which are presented in Table 5. Echoing the results from the TPR analysis, sonication has also imparted considerable discrepancies on the NiO and Ni silicate surface compositions. The proportion of NiO on the surface was increased from 18.6% in the non-sonicated catalyst to 29.5 – 40.7% in the sonicated catalysts, in which this proportion has also increased with higher ultrasonic intensities. Conversely, the Ni silicate phase has also seen a decrease in relative proportion when ultrasound was used during the synthesis, as it dropped from 81.4% in the unsonicated catalyst to a range of 59.3 – 70.5% in the sonicated counterparts. This aforementioned trend observed in the XPS analysis was similar to those obtained from the TPR analysis, albeit with a greater magnitude of change. This was due to the fact that XPS presents details on surface composition, which could vary slightly from those of the bulk provided by TPR measurements [84].It was reported that the precipitation of silicates without the use of surface modifiers tend to clump to each other, which results in the formation of secondary aggregates [48], hence in the case of no sonication (samples A) and low power sonication (sample B), silicate ions would adhere to existing silicate networks in an unperturbed manner during suspension ageing, resulting in a higher degree of aggregation within a shorter period of time. With such a phenomenon, the relatively unhindered growth of the silicate network is more prominent, which brought about longer and more severe ageing conditions on the outer surface of the catalyst particles for samples A, as validated by the XPS analysis showing the high concentration of Ni silicates on the surface (81.4%). As opposed to the above case, the inclusion of ultrasonic irradiation during support loading and the early stage of ageing could effectively disperse the aggregates and suppress the growth of the catalyst particles, mainly by delaying the growth of the silicate network. The erosion caused by high intensity ultrasonic irradiation coupled with the delayed growth of silica on the outer surficial layers eventually limited the extent of ageing, as exhibited by the lower Ni silicate concentration in samples C (61.2%) and D (59.3%). Such a delayed growth in particles could have also played a role in enhancing the Ni dispersion in sonicated catalysts, as smaller particles result in an increase in total exposure of Ni active sites. Herein, the conceptualisation of the two different states of ageing is shown in Fig. 6 .The textural and structural properties of the calcined nickel catalysts were evaluated using N2 physisorption, in which the adsorption–desorption isotherms for the catalysts can be seen in Fig. S2 (Supplementary information). According to the IUPAC classification, the sonicated samples (B – D) exhibited Type IV(a) isotherms, characteristic of mesoporous substances with capillary condensation accompanied by hysteresis. Contrariwise, the non-sonicated sample A exhibited a hybrid of the Type II and Type IV(a) isotherms, denoting the presence of mesopores and macropores, which was also accompanied by a hysteresis loop. Regarding the hysteresis loops exhibited by the catalysts, they can be ascribed to the Type H2(b) hysteresis loop, which signifies the presence of ink-bottle pores with a wider size distribution of neck widths, typical of mesoporous ordered silicas obtained after hydrothermal treatment [85].The incorporation of ultrasound is said to have altered the textural properties of the catalyst, resulting in more uniform mesoporous structures with higher BET surface areas, as shown in Table 6 . The increase in surface areas was also noted in other studies, in which ultrasound acted as a dispersant tool for catalyst phases, which might increase the total surface area available for reaction [36,64]. The imploding micro-bubbles in the irradiated suspension as a result of acoustic cavitation facilitated the nucleation and fine dispersion of particles, creating phases of more uniform and well-defined morphology that resulted in increased surface areas from 192.5 m2/g to a range of 228.9 – 289.7 m2/g [86]. According to Coenen [87] and Ghuge et al. [45], it is the nickel antigorite (hydrosilicate) phase instead of the silica support that gives rise to a high BET surface area, with the former typically exhibiting >300 m2/g while the latter presenting surface areas in the range of 20–50 m2/g. If one is to refer to the results from the TPR studies (Table 4), it can be assumed that sample A with the highest Ni silicate percentage (61.7%) would exhibit the highest BET surface area. However, Table 6 shows otherwise with sample A exhibiting a significantly lower BET surface area compared to the sonicated samples. As discussed earlier, the amount of silica clusters attached to the hydrosilicate declined with the usage of ultrasound, which in turn led to higher hydrosilicate to silica ratio in the samples synthesised with ultrasonic irradiation, thus giving rise to a higher BET surface area. In fact, this observation is also correlated with the increase in ultrasonic intensity from sample B – D, whereby the formation of more silica clusters were suppressed, causing a further increase in hydrosilicate to silica ratio, thereby progressively increasing the BET surface area as per Table 6. As a result, sonicated catalysts exhibited higher range in the overall BET surface area than the unsonicated catalyst. Nevertheless, the pore volume and average pore width of the catalysts were similar in range, noting that the non-sonicated sample possessed a larger average pore width due to the presence of macropores, which can also be observed in the pore size distribution in Fig. 7 . One can see that all samples had a uniform pore size distribution in the mesoporous range, while the non-sonicated sample also contained pores in the macroporous range. The presence of larger macropores is likely due to the unperturbed and prolonged ageing [88,89] relative to that experienced by the sonicated catalyst, which contributed to the increase in average pore width for catalyst A. Overall, the presence of pores above 3.5 nm was a good indication that the catalysts were suitable for the hydrogenation of edible sunflower oil, which are approximately twice the size of the triglyceride molecules of 1.5 – 2 nm [8,44]. However, the presence of narrower pores with average width slightly below 3.5 nm in catalyst B might have posed significant diffusional and mass transport problems for the movement of triglycerides. Table 6 collates the relative pore flow rate for each catalyst, with catalyst B as the reference. Employing the flow continuity equation, it is inferred that the unsonicated catalyst A had a 96.4% higher flow rate than sample B. The smaller flow area due to smaller pores denote the increased hindrance in mass transfer, which results in the slower diffusion rate of reactants and products in and out of the pores, leading to decreased activity relative to samples possessing higher pore flow rates.All synthesised catalysts were subjected to catalytic activity tests via the partial catalytic hydrogenation of sunflower oil. The decline in IV throughout the stipulated reaction time of 90 min for each catalyst sample was noted and presented in Fig. 8 . The hydrogenation performance was tracked by observing the decrease in the IV. The sunflower oil utilised as the reactant has an initial IV of 124, which was tested with the above-mentioned iodine value test. As anticipated, the IV for all sample sets decreased as the reactant was progressively saturated during hydrogenation. Furthermore, the gradient of the graph signified the rate of decline in the IV, whereby a sharper gradient represented an increase in hydrogenation activity. Catalyst C, synthesised with an ultrasonic intensity of 20.78 W cm−2 registered the highest activity among the catalysts synthesised, attaining a final IV of 37.9 after a reaction time of 90 min. On the other hand, the unsonicated catalyst produced an IV drop to 53.8 in 90 min. The least active catalyst was catalyst B, synthesised with an ultrasonic intensity of 7.07 W cm−2. Although possessing higher Ni surface area and dispersion than its unsonicated counterpart, catalyst B presented a smaller pore size below 3.5 nm. Coenen [7] postulated that while it is beneficial to possess a high Ni dispersion and surface area, oil/fat hydrogenation is also structure-sensitive, whereby pores narrower than 3.5 nm would considerably impair the mass transport and pore diffusivity of bulky triglyceride molecules, hence causing pore congestion and affecting overall catalytic activity. The initial activity of catalyst B was higher than that of catalyst A due to its advantages in Ni dispersion and surface area. However, at IV = 90, the activity of catalysts A and B began to diverge as shown in Fig. 8, the percentage of bulky triglyceride molecules (C18:2 and cis-C18:1) for sample B remained relatively high compared to that of sample A, as shown in Fig. 9 . As the narrower pores in sample B effectively impede the diffusion of these molecules, the reaction and conversion of such molecules would be affected negatively, thus resulting in a slower drop in IV. In addition, as presented in Table 6, the unsonicated catalyst A has a 96% higher flow in the pores than that of catalyst B, indicating that despite a higher intrinsic catalytic activity possessed by catalyst B, its mass transfer at the pores was severely impacted thus affecting the overall catalytic activity.Based on the catalyst characterisation results, the geometric effects imparted by ultrasonic irradiation have led to the sonicated catalysts possessing superior catalytic activity compared to their unsonicated counterpart, with the exception of the sonicated catalyst B due to its disadvantages in pore characteristics. The action of ultrasound during synthesis has induced the formation of more reducible Ni species available on the catalyst superficial layers, which on the other hand are composed of particles of lower agglomeration, thus leading to increased Ni surface area and dispersion after reduction. The presence of more well dispersed active sites ultimately allowed more reactants and intermediates to adsorb and react. Nevertheless, it was also discovered that catalyst D, albeit sonicated with the highest ultrasonic intensity, performed marginally poorer than catalyst C, whereby the hydrogenation activity of the latter overtook the former during mid-hydrogenation. This could be due to the large Ni surface area exhibited by catalyst D, allowing more surface area contact with the reacting medium, hence increasing the propensity of Ni lixiviation. Indeed, AAS analyses performed on the hydrogenation products of catalysts C and D confirmed that the concentration of Ni present in the hydrogenation product of catalyst D was ca. 30% higher than that of catalyst C at 90 min (0.00294 mg Ni/g oil vs. 0.00226 mg Ni/g oil), which caused the gradual deactivation of catalyst D over time.The hydrogenation activity of the synthesised catalysts was demonstrated via the IV tests, which is a direct representation of the number of carbon–carbon double bonds present in a sample, without any discrimination towards the product composition. Product analyses are also very crucial in distinguishing the behaviour of catalysts during edible oil hydrogenation. The synthesised catalysts were appraised by monitoring the evolution of saturated fat (C18:0), trans-C18:1 and cis-C18:1 at particular IV levels which represent a reaction time up to 90 min, as depicted in Fig. 9. The respective yield and selectivity of the catalysts are shown in Table 7 .From Fig. 9 it can be seen that the composition profile of the hydrogenated fats depicted similar trends for all catalysts used. The elimination of carbon–carbon double bonds, as well as cis–trans isomerisation during hydrogenation can be tracked throughout the plots. Overall, a depletion of C18:2 and increase in C18:0 is observed, accompanied by a variation of cis-C18:1 and trans-C18:1 composition throughout the reaction. Another observation was the overall increase in trans-C18:1 coupled with the overall decrease in cis-C18:1 for all hydrogenation runs. It has been documented that trans-C18:1 are preferentially formed during the initial phase of hydrogenation [90]. On the other hand, cis-C18:1 was either isomerised to trans-C18:1 or fully hydrogenated to C18:0. It can be seen that the formation of trans-C18:1 was generally less favoured in the hydrogenation runs using sonicated catalysts, whereby the selectivity of trans-C18:1/cis-C18:1 at IV 90 and 70 were at the lower range of 10 – 12.5% and 18.42 – 24.39%, respectively. Meanwhile, the unsonicated catalyst presented higher trans-C18:1/cis-C18:1 selectivities of 20.46% and 30.43% at IV 90 and 70, respectively.While Fig. 9 gives a useful outline of the catalytic performance of each sample over the course of the reaction, it is also highly crucial to analyse the product distribution correlated to a specific target IV. In particular, the concentration of trans-C18:1 and C18:0 are of paramount significance in deciding the final oil quality. Fig. 10 demonstrates the percentage of C18:0 and trans-C18:1 fatty acids at an IV of 70. In this case, this particular level was chosen for comparison as it is a common target IV for oleomargarine products and it also corresponds to the point where the formation of C18:0 starts to be significant [6]. This extent of hydrogenation resembles to that in the manufacture of fatty acid components required for shortenings and margarine in the industry [91]. Employing the sonicated catalysts, the hydrogenation products obtained at an IV of 70 presented trans-C18:1 at equal to or less than 10%, while the hydrogenation products of the unsonicated counterpart registered trans-C18:1 levels of 13.7%, with the lowest trans-C18:1 level observed using catalyst D. On the other hand, the sonicated catalysts had similar percentages at approximately 29% in C18:0 production, while the unsonicated catalyst was slightly higher at 34%.Generally, as the overall hydrogenation proceeded, the isomerisation of fatty acid products increased with conversion. However, it is noted that the mechanisms related to hydrogenation and cis/trans isomerisation are heavily interlinked. With regards to the formation of trans fatty acids (TFAs) during hydrogenation, the addition–elimination mechanism proposed by Horiuti and Polanyi [92] is commonly adopted to illustrate the process. The mechanism states that both the hydrogenation and isomerisation processes are governed by a half-hydrogenation state mechanism. Firstly, carbon–carbon double bonds of C18:2 or C18:1 molecules are adsorbed on active sites. It is noted that C18:2 bonds tend to adsorb to the surface to a stronger degree and are thus first hydrogenated. Therefore, this also explains the consecutive drop in linoleic content throughout the reaction as per Fig. 9. After adsorption, one of the double bonds are half-hydrogenated by the addition of one hydrogen atom from the active sites. Further reaction resulting in the saturation of C–C bonds necessitates the addition of another hydrogen atom. Nevertheless, typically without the presence of a second hydrogen atom, the first hydrogen atom detaches from the half-hydrogenated intermediate, thus re-establishing the double bond with either a cis-C18:1 or trans-C18:1 configuration [56]. Particularly, the latter configuration is more thermodynamically favoured than the former [93]. As hydrogenation is supplemented by isomerisation reactions, it is suggested that the change in the electronic characteristics of the sonicated Ni catalysts could also influence the isomerisation reactions. In this case, the enhanced removal of the chemisorbed half-hydrogenated intermediate from the active sites could prevent the further isomerisation to TFAs. This alteration in electronic properties may be used to explain the decrease in TFA formation for the sonicated catalysts, relative to the unsonicated counterpart [2,94]. Hence, one of the functions of ultrasound in this case of catalyst synthesis might be as a selectivity modifier for the Ni catalyst. Through sonication, the bonding strength of the adsorbates on the active sites were altered, which resulted in the increase of the energy barrier required for fatty acid isomerisation or the decrease of the energy barrier for hydrogenation that hastens cis-C18:1 to C18:0 conversion [94]. In particular, the shift from a higher to lower binding energy in the sonicated catalysts as observed in the XPS results (Table 5) complemented this finding. The changes in the electronic properties and electron densities of catalysts as observed by the shift in binding energies is known to alter the adsorption of reacting species and hence the selectivity towards the final products [95–97]. In this case, the decrease in binding energy denotes the increase in electron density of Ni species on the catalyst surface, thus leading to weaker interactions between the adsorbed or hydrogenated intermediates and the active sites, allowing easier desorption thus diminishing the transformation of unsaturated fat molecules or cis isomers to trans isomers [98], which was also a phenomenon reported by other researchers such as Iida et al. [99]. Similarly, Li et al. [75] has noticed a negative shift in binding energy for Ru species in Ru-B catalysts after sonication, which facilitated the adsorption of the oxygen atom of carbonyl groups in the cinnamaldehyde molecule, ultimately enhancing the hydrogenation of the molecule.To further assess the hydrogenation performance of the synthesised catalysts in this work, comparisons to novel catalysts obtained from recent literature were made, with a compilation presented in Table 8 . Compared to most studies, it can be seen that catalyst C in this present work was able to achieve an IV drop to 70 in a shorter reaction time of 40 min, while containing a low amount of saturated fats and trans-fats, particularly the latter. Although noble metal (e.g. Pt and Pd) catalysts required milder reaction conditions (approximately 100 °C) and produced lesser saturated fats, the present catalyst was able to achieve a relatively low trans-fats percentage of 10%. It should also be noted that the saturated fats content for noble metal catalysts were comparatively lower due to their lower extent of reaction. Compared to the other Ni-based hydrogenation catalysts such as Ni/ZnO/Al2O3, Ni/SiO2, Ni-Mg-Ag/D and Ni-Ce/Al2O3, the novel sonicated catalyst in this work has produced at least half the amount of trans-fats, while maintaining comparable if not better hydrogenation activity. Nevertheless, it should be noted that these studies obtained from the literature were executed with disparities in reaction conditions and oil variation, hence the overall activity and trans-fats and saturated fats contents would have also been affected. For instance, the Ni-Ce/Al2O3 catalyst although requiring only 33 min of reaction time for hydrogenation, was employed to hydrogenate a feed oil with the lowest initial IV of 115. Therefore, this present work has exemplified the potential of incorporating ultrasonic technology into the synthesis of such Ni hydrogenation catalysts, with positive and applicable end results i.e. the enhancement of hydrogenation activity and the lowering of detrimental trans-fats for edible oil applications. For the latter application, edible oil manufacturers and catalyst developers who seek an alternative technique to lower their trans-fats production substantially could take the ultrasound-assisted catalyst synthesis procedure into consideration for their processes.This present work was performed to investigate the synthesis of silica-supported nickel catalysts using the ultrasonic technique and its application in the partial hydrogenation of edible oil. It was successfully shown that the incorporation of ultrasonic irradiation into the ageing process of sequentially precipitated catalysts have led to benefits in terms of the catalyst characterisation properties and their performance in the catalytic hydrogenation of sunflower oil. Further understanding into the proposed mechanisms indicated that the usage of ultrasonic irradiation, as well as variations in its intensity, had profound impacts on the overall characteristics and compositions (hydroxycarbonates, silicates and silica clusters) comprising the catalysts. It is noted that ultrasonic irradiation had a significant influence on the ageing properties of the catalysts, which directly affected the formation of silica clusters, followed by hydroxycarbonates and Ni silicates. For instance, ultrasonic irradiation led to the erosion of hydroxycarbonate phases and the suppression of silica clusters, which once formed, serve as protective enclosing to avert Ni silicate erosion. As the ultrasonic intensity was increased, greater Ni dispersion was achieved due to increasing extents in hydroxycarbonate erosion, while increase in BET surface areas was also noted due to the decrease in silica to hydrosilicate ratio as a result of the suppression of silica clusters. On the other hand, it was also observed that upon the termination of ultrasonic irradiation, the growth of the silica network on the surface of the catalysts during ageing also led to a higher Si composition on the external layer of the catalyst particles. In general, these effects in turn imparted alterations to the geometrical and electronic properties of the catalysts, improving catalyst reducibility, surface area and dispersion of the material.Hydrogenation using the catalyst synthesised at an ultrasonic intensity of 20.78 W cm−2 achieved the fastest decrease to IV = 70 in 40 min, compared to 58 min achieved by its non-sonicated counterpart. Furthermore, the usage of ultrasonic irradiation prompted the modification of product selectivity by altering the electron density and subsequently the adsorption capability of the nickel species with hydrogenation intermediates. This led to a lower production of trans-fats in the sonicated catalysts compared to the non-sonicated catalyst. To illustrate, at IV = 70, the sonicated and non-sonicated catalysts produced 7 – 10% and 13.7% of trans fats, respectively. Generally, the effects of sonication on nickel catalysts used for sunflower oil hydrogenation were positive and promising, which could potentially be extended to other oil types or catalyst systems. Overall, the present work has shown that the incorporation of ultrasound during catalyst synthesis offers attractive benefits, in which the activity and selectivity of the synthesised catalysts could be significantly modified by a brief exposure to ultrasonic waves. Mitchell S.W. Lim: Investigation, Data curation, Visualization, Writing - original draft. Thomas Chung-Kuang Yang: Supervision, Resources, Funding acquisition. T. Joyce Tiong: Conceptualization, Methodology, Resources. Guan-Ting Pan: Data curation, Investigation. Siewhui Chong: Data curation, Visualization, Validation. Yeow Hong Yap: Writing - review & editing.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The authors would like to acknowledge the technical guidance provided by Van Wu at the Precision Analysis and Material Research Center (National Taipei University of Technology). This research was financially supported by The Ministry of Science and Technology, Taiwan under the grant number 108-2221-E-027-072.Supplementary data to this article can be found online at https://doi.org/10.1016/j.ultsonch.2021.105490.The following are the Supplementary data to this article: Supplementary data 1

Sequentially precipitated Mg-promoted nickel-silica catalysts with ageing performed under various ultrasonic intensities were employed to study the catalyst performance in the partial hydrogenation of sunflower oil. Results from various characterisation studies showed that increasing ultrasonic intensity caused a higher degree of hydroxycarbonate erosion and suppressed the formation of Ni silicates and silica support, which improved Ni dispersion, BET surface area and catalyst reducibility. Growth of silica clusters on the catalyst aggregates were observed in the absence of ultrasonication, which explained the higher silica and nickel silicate content on the outer surface of the catalyst particle. Application of ultrasound also altered the electron density of the Ni species, which led to higher activity and enhanced product selectivity for sonicated catalysts. The catalyst synthesised with ultrasonic intensity of 20.78 Wcm−2 achieved 22.6% increase in hydrogenation activity, along with 28.5% decrease in trans-C18:1 yield at IV = 70, thus supporting the feasibility of such technique.

Hydrogen is called to play a key role in future energy demand as a substitute for fossil fuels [1]. In fact, in the post-pandemic crisis scenario, the alternatives for the production and use of renewable H2 are taking on great relevance, as for example in the economic recovery package of EU (NextGenerationEU) with €750 billion focused around the European Green Deal, where renewable energy projects highlight, especially wind, solar and kick-starting a clean hydrogen economy, with full economy decarbonization as the target for 2050 [2].Currently, the most common and economic process for H2 production is the steam reforming (SR) of natural gas, accounting for 76% of the global production, and whose global warming potential per kg of hydrogen produced is 11.956 kg CO2-eq [3]. Although the optimal technology for “green H2″ production is the water electrolysis, in the energy transition period until this solution becomes a reality H2 could be obtained with limited CO2 emissions from lignocellulosic biomass, by thermochemical process, particularly gasification, partial oxidation and SR of biomass derivatives (as bioethanol and bio-oil). Some of these processes are currently in a pilot-scale demonstration or at a commercial stage but they require improvements to produce larger competitive volumes [4]. The SR of bio-oil, obtained by fast pyrolysis of lignocellulosic biomass, has gained increased attention [5], due to the good prospects of a strategy to combine the delocalized bio-oil production (with well-developed technologies and with low infrastructure costs) [6], with centralized bio-oil SR in a bio-refinery with units designed ad hoc for selective H2 production. The liquid state and higher volumetric energy density of bio-oil facilitates its transportation, storage and treatment compared to biomass [7]. In addition, the SR of bio-oil avoids the costly dehydration steps required for the use of bio-oil as fuel or for its valorization in other catalytic processes [8].Bio-oil is composed of an oxygenate mixture with the presence of different functional groups (carboxyl, ester, carbonyl, ether, phenolic and hydroxyl groups) and variable water content (depending on the origin of the biomass). The SR reaction of oxygenated hydrocarbons (CnHmOk) to produce syngas (H2 + CO) can be described by the following equation: (1) C n H m O k + ( n - k ) H 2 O → n C O + ( n + m 2 - k ) H 2 In addition to the main reaction (Eq. (1)), the water gas shift (WGS) reaction (Eq. (2)) takes place, and thus the overall SR equation for the oxygenates is defined by Eq. (3). (2) CO + H 2 O ↔ C O 2 + H 2 (3) C n H m O k + ( 2 n - k ) H 2 O → n C O 2 + ( 2 n + m 2 - k ) H 2 H2 yield is also affected by reactions occurring in parallel to oxygenates SR and WGS reactions, such as decomposition/cracking (Eq. (4)), which affects the catalyst stability due to coke deposition, SR of decomposition products (CH4 and hydrocarbons, (Eqs. (5) and (6)), and interconversion of oxygenates (Eq. (7)). (4) C n H m O k → C x H y O z + g a s ( C O , C H 4 , C O 2 , C a H b , H 2 , … ) + c o k e (5) C H 4 + H 2 O ↔ C O + 3 H 2 (6) C a H b + a H 2 O → a C O + a + b / 2 H 2 (7) C n H m O k → C x H y O z Moreover, the reactions for coke formation from gaseous products (Eqs. (8) and (9)) and its gasification reaction (Eq. (10)) should be considered, as they may affect catalyst stability and also the products yields when they are highly promoted. (8) Hydrocarbon decomposition: CaHb → (b/2)H2 + aC (9) Boudouard reaction: 2CO ↔ C + CO2 (10) Coke gasification: Coke + H2O → CO + H2 One of the main problems or bottlenecks of bio-oil SR is the rapid deactivation of the catalyst, which justifies that it receives a great attention, in order to prepare stable catalysts. Deactivation studies have generally been carried out with model oxygenates [9–20], with mixtures of oxygenates [21–23] , and studies with raw bio-oil are limited. [24–29]. The results show the importance of the nature and location of the coke in the deactivation of the catalyst. Thus, the formation of carbon filaments has a reduced incidence in the deactivation, whose responsibility falls mainly on the formation of amorphous coke encapsulating the Ni sites. There is also a general tendency to relate the formation of deactivating amorphous coke to the SR of some families of oxygenates (phenols, carboxylic acids, furfural and saccharides, mainly).This paper delves into the clarification of deactivation by coke of a catalyst derived from NiAl2O4 spinel, which has been previously proven to have high activity and selectivity to H2 in the reforming of raw bio-oil and, more interestingly, it can be fully regenerated by coke combustion at 850 °C (with spinel reconstruction) [30]. For that purpose, we have studied the influence on the deactivation behavior and coke deposition of individual oxygenate compounds with varied functional groups present in bio-oil (acetic acid, acetaldehyde, acetol, ethanol, acetone, catechol, guaiacol and levoglucosan). The evolution along time on stream of the conversion and products yields in the SR of each individual compound have been analysed, as well as the amount, nature, morphology and location of the coke deposited on the catalyst used by means of several techniques: temperature programed oxidation (TPO), X-ray diffraction (XRD), N2 adsorption–desorption, scanning and transmission electron microscopy (SEM, TEM) and Raman spectroscopy. The experimental conditions used are similar to those previously used in the SR of raw bio-oil with the same catalyst [31], which has allowed a direct comparison of the catalyst performance in the SR of each individual oxygenate with that obtained in the SR of raw bio-oil. The results have allowed stablishing the main responsible of catalyst deactivation during SR of bio-oil, as well as the coke characteristics that mainly affect the deactivation of the catalyst. Consequently, interesting information is obtained to adjust the composition of the raw bio-oil in order to attenuate catalyst deactivation by coke. In addition, the oxygenates of greatest interest as model compounds for the comparison tests of catalyst deactivation for the SR of raw bio-oil are identified.The pure oxygenate compounds selected as representative of the major families of oxygenates in bio-oil are the following: acetic acid (AA) (Romil LTD, purity > 99.9 %), acetaldehyde (AD) (Merck KGaA, purity ≥ 99 %), acetone (A) (AppliChem GmbH, purity ≥ 99.9 %), acetol (AT) (hydroxyacetone, Alfa Aesar GmbH, purity = 95 %), ethanol (E) (Merck KGaA, purity ≥ 99.9 %), 1,2-benzenediol or catechol (C) (Sigma-Aldrich, purity ≥ 99 %), levoglucosan (L) (1,6-Anhydro-β-D-glucopyranose, Acros Organics, purity > 99%), and 2-methoxyphenol or guaiacol (Alfa Aesar GmbH & Co, purity > 98 %) dissolved in 50 wt% of ethanol (G + E) due to its low solubility in water. Acetone, acetaldehyde, 1,2-benzenediol (catechol) and 2-methoxyphenol (guaiacol) are representative of relevant families of compounds in bio-oils such as ketones, aldehydes and phenols (among these, mainly guaiacols and catechols) [32,33]. Acetic acid, levoglucosan and acetol are present in remarkable concentrations in the bio-oil obtained from pyrolysis of pine sawdust [26,34]. The study of the catalyst behavior in the SR of ethanol is interesting because it may be cofed with bio-oil for its stabilization and because the SR of bio-oil/bio-ethanol mixture (BO + E) is an interesting route for sustainable H2 production from two biomass derived feeds [35].The catalyst precursor (Ni-Al spinel, NiAl2O4) was prepared by co-precipitation method with a nominal Ni content of 33 wt% from Ni(NO3)2·6H2O and Al(NO3)3·9H2O with a NH4OH 0.6 M solution as a precipitating agent. The precipitation was carried out at 25 °C until the pH was fixed at 8. After aging for 30 min, the precipitate was filtered, washed with distilled water to remove the ammonium ions and dried at 110 °C for 24 h. Lastly, the catalyst was calcined at 850 °C for 4 h [30].The physical properties of the fresh catalyst and deactivated samples (BET surface area, pore volume and mean pore diameter), were characterized by adsorption–desorption of N2 in a Micromeritics ASAP 2010. Temperature Programed Reduction (TPR) was carried out in a Micromeritics AutoChem 2920 for determining the reducibility of the metal species. The amount and nature of coke deposited on spent catalyst samples has been determined by Temperature Programed Oxidation (TPO) in a TA-Instruments TGA-Q5000IR thermobalance, coupled in line with a mass spectrometer (Thermostar Balzers instrument) for monitoring the signal of CO2. The coke content has been quantified from the CO2 spectroscopic signal, due to Ni oxidation during combustion process masks the thermogravimetric signal in samples with low coke content [30]. The X-Ray Diffraction (XRD) analysis of the reduced fresh and spent catalysts was carried out in a Bruker D8 Advance diffractometer with a CuKα1 radiation, from 10° to 80° with step of 0.04° in 2θ and measurement time of 103 min. The scanning electron microscopy images of the fresh or spent catalysts were taken with a Hitachi S-4800 N field emission gun scanning electron microscope (FEG-SEM), with an accelerating voltage of 5 kV and secondary electron detector (SE-SEM) and a Hitachi S-3400 N microscope with an accelerating voltage of 15 kV, using a backscatter electron detector (BSD-SEM). The transmission electron microscopy (TEM) images were obtained in a Phillips CM-200 microscope using an accelerating voltage of 200 kV. The Raman spectra were carried out in a Renishaw InVia confocal microscope using an excitation wavelength of 514 nm, taking a spectrum in several areas of the sample for assuring reproducibility.The N2 adsorption–desorption isotherm and the BJH pore distribution of the fresh-reduced catalyst are shown in Figs. S1a and S1b, respectively, in the Supporting Information. An isotherm of type IV according to the IUPAC classification is observed in Fig S1a, which is associated with capillary condensation taking place in mesopores, with a hysteresis of the type H2, attributed to a difference in mechanism between condensation and evaporation processes occurring in pores with narrow necks and wide bodies (often referred to as 'ink bottle' pores). The BET surface are, pore volume and mean pore diameter for the fresh-reduced catalyst (Table 1 ) are 65.1 m2/g, 0.24 cm3/g and 13.1 nm, respectively. The TPR profile of the fresh catalyst (Figure S1c) has a maximum H2 uptake at 760 °C, corresponding to the reduction of Ni species incorporated in the NiAl2O4 spinel structure [30,36]. The XRD pattern of the fresh catalyst prior reduction (black curve in Figure S1d) shows intense peaks at 2θ = 37.2, 45.2 and 65.7° corresponding to the cubic structure of NiAl2O4 spinel, whereas the XRD of the fresh-reduced catalyst (blue curve in Figure S1d) shows peaks corresponding to Ni0 (diffraction angle at 44.5° in (111) plane, 51.8° in (200) plane and 75.5° in (110) plane, JCPDS n° 00–004-0850) and Al2O3 (37.3°, 45.6° and 66.8°, JCPDS n° 01–077-0396). This result indicates that the reduction treatment (at 850 °C for 4 h) completely converted NiAl2O4 spinel into reduced Ni crystals supported on Al2O3 (Ni/Al2O3), as previously reported [36]. The mean Ni crystal size in the reduced catalyst (calculated with the Debye-Scherrer equation using the diffraction peak at 2θ = 52°) is 9 nm (Table 2 ).Runs have been carried out in an automatized reaction system (MicroActivity-Reference, PID Eng & Tech,) that has been described in detail elsewhere [37], with a fluidized bed reactor. The catalyst (with particle size of 150–250 µm to avoid internal diffusional limitations) is mixed with inert solid (SiC, 37 µm particle size) in order to ensure good fluid dynamic behaviour of the catalytic bed (inert/catalyst mass ratio > 8/1).Prior to each steam reforming reaction, the catalytic bed is reduced in-situ by using H2-N2 flow (10 vol% of H2) at 850 °C for 4 h, thus forming the active Ni0/Al2O3 catalyst. The operating condition for the kinetic runs have been: atmospheric pressure; 600 and 700 °C, that are suitable for attaining high conversion in the SR of bio-oil; space time of 0.034 gcatalyst h/goxygenate in order to favor catalyst deactivation by coke formation during not excessively long runs (of 5 h duration); steam-to-carbon (S/C) molar ratio of 3 (except for levoglucosan, with S/C = 6 due to its low water solubility), which is suitable for promoting WGS reaction (necessary to enhance H2 yield) but without excessive penalty of energy requirements. This S/C ratio has been set by co-feeding water (307 Gilson pump) with the feed (injection pump Harvard Apparatus 22). The reaction products were analysed in a Micro GC Varian CP-490 connected in-line to the reactor through an insulated line (130 °C) to avoid condensation of the products. The gas chromatograph is equipped with three analytic channels: molecular sieve MS5 for quantifying H2, O2, N2, CH4 and CO; PPQ column for light hydrocarbons (C2-C4), CO2 and water; and Stabilwax for oxygenated compounds (C2+) and water.In order to quantify the results, the following reaction indices were used: (11) Carbon c o n v e r s i o n t o g a s e s : X = F out, gas F in where Fout, gas is the molar flow rate of the total carbon in gaseous product (CO2, CO, CH4 and light hydrocarbons, in C units contained) at the reactor outlet, and Fin is the molar flow rate of the oxygenate at the reactor inlet in C units contained. (12) H 2 y i e l d : Y H 2 = F H 2 F H 2 o where FH2 is the H2 molar flow rate in the product stream and F H 2 o , is the stoichiometric molar flow rate, which is calculated as (2n + m/2 – k)/n Fin, according to the global stoichiometry for the SR of each oxygenate (CnHmOk) (including the WGS reaction) (Eq. (3)).In order to assess the catalyst activity, selectivity and stability for the different feeds studied, the evolution with time on stream (TOS) of carbon conversion to gas and yield of H2 at 600 and 700 °C is shown in Fig. 1 (acetic acid (a), acetaldehyde (b), ethanol (c), acetol (d) and acetone (e)) and Fig. 2 (catechol (a), mixture guaiacol + ethanol (b) and levoglucosan (c)).At 600 °C, the initial H2 yield (fresh catalyst) varies between 42 % (for acetone, Fig. 1e) and 61 % (for acetaldehyde, Fig. 1b), with similar values (near 50 %) for the rest of oxygenates, thus evidencing similar reactivity at this temperature. The increase in temperature up to 700 °C enhances the carbon conversion to gas and H2 yield at zero time on stream in the SR of all oxygenates, with this increase being more significant for acetaldehyde, ethanol and acetol (almost 100 % conversion), and also for acetic acid and catechol (around 93 % conversion). However, it is less noticeable for the guaiacol + ethanol mixture, acetone and levoglucosan, thus evidencing the lower reactivity towards SR reactions at high temperature of the latter oxygenates. The low increase with temperature of the carbon conversion for the guaiacol + ethanol mixture (Fig. 2b), compared to that obtained with ethanol (Fig. 1c), gives evidence of a low effect of temperature for guaiacol, whose reactivity for SR reactions is noticeably lower than that of ethanol, acetic acid, acetaldehyde, acetol and catechol. Moreover, the initial H2 yield in the SR of ethanol at 700 °C (70 %) is lower than that obtained with the other oxygenates (around 80 %), in spite of its high carbon conversion (100 %). This result reveals the higher selectivity of the catalyst for H2 forming reactions (steam reforming and WGS) in the SR of acetic acid, acetol and acetaldehyde compared to ethanol, whose reforming produces significant CH4 formation (not shown). The high H2 yield (80%) obtained in the SR of levoglucosan at 700 °C, in spite of its incomplete carbon conversion, should be attributed to the high S/C ratio used in the SR of this oxygenate (6), that significantly promotes WGS reaction.Regarding the stability of the catalyst, overall, the conversion and H2 yield remain constant or even increase (for acetic acid (Fig. 1a), acetol (Fig. 1d) and catechol (Fig. 2a)) at 600 °C and slightly decrease with TOS at 700 °C in the SR of all the studied oxygenates, except for guaiacol + ethanol mixture. The increase in conversion and H2 yield with TOS can presumably be explained by the formation of a remarkable amount of filamentous coke (as shown later), which leads to an improved Ni dispersion and better accessibility of reactants due to the tip-growth mechanism of carbon filaments [38]. Conversely, in the SR of the guaiacol + ethanol mixture (Fig. 2b) there is a fast decrease in conversion and H2 yield at both 600 and 700 °C, until the values corresponding to the thermal reaction routes (in the absence of catalyst) are reached. Therefore, this result evidences a much faster deactivation rate of the catalyst in the SR of this mixture than for rest of the oxygenates studied.In order to identify the causes responsible for the deactivation of NiAl2O4 catalyst, and for a better understanding of its deactivation behavior in the SR of the different oxygenates, a thorough characterization of the spent catalyst samples has been performed, by using complementary techniques, that include TPO, SEM and TEM images, XRD, TPR, Raman spectroscopy and N2 adsorption–desorption. These techniques have allowed determining the amount, nature, morphology and structure of the coke deposited in the catalyst, as well as the changes in the metal sites and porous structure of the catalyst. It should be noted that Ni oxidation was ruled out as deactivation cause, as no significant reduction peaks were observed in the H2-TPR profiles of selected deactivated catalyst samples (results not shown here), thus indicating the absence of oxidized species. This is an expected result, which was previously observed in the oxidative steam reforming (OSR) of raw bio-oil with this type of catalyst [39], and it is due to the highly reducing environment in the SR reaction, with a high H2 content. The results of the rest of characterization techniques are presented in the following sections. The spent catalyst samples have been denoted as X-N, where X identifies the oxygenate feed (AA = acetic acid; AD = acetaldehyde; E = ethanol; AT = acetol; A = acetone; C = catechol; (G + E) = guiaiacol + ethanol; L = levoglucosan) and N is the SR temperature (600 or 700 °C). Figs. 3 and 4 show the TPO profiles of the deactivated catalysts used in the SR of the oxygenates at different temperatures, obtained from the spectroscopic signal of CO2 released during coke combustion (as explained in section 2.3). These results provide qualitative information on the nature and/or location of the coke in the structure of the catalyst [40]. Several authors differentiate the amorphous and filamentous coke contents of coke deposited on Ni catalysts by deconvolution of the TPO profiles. Thus, Hu et al. [41], relate each type of coke to one of the two peaks of the TPO, so that the amorphous/paraffinic coke burns at lower temperature than the graphitic/filamentous coke. This identification of the two types of coke allowed verifying that with the addition of Fe there is a higher attenuation of the deposition of the graphitic/filamentous coke in the Fe-Ni/Al2O3 catalysts used in the steam reforming of toluene. The same authors characterize the coke over a Ni/α-Al2O3 catalyst in the steam reforming of two hydrocarbons (toluene and methylnaphthalene) and two oxygenates (phenol and ethanol) distinguishing amorphous coke from carbon nanotubes (CNTs), whose combustion is identified with the peak at higher combustion temperature [42]. Subsequently, these authors have verified the relevant effect of the steam reforming temperature (in the 500–800 °C range) on the content of the two types of coke and on the quality of the CNTs, proving the existence of a maximum of both at 650 °C [43]. The identification of these two types of coke of different nature by deconvolution of coke combustion TPO profiles was also used to quantify the formation of CNTs on Ni catalysts from the volatiles from polyolefins pyrolysis [44].The size and location of combustion peaks in Figs. 3 and 4 evidence differences in the amount and nature of coke deposited with the different oxygenates and at different temperature. Noticeably, in the TPO profiles of Figs. 3 and 4 there is apparently a unique combustion peak, with maximum in the 500–550 °C range for the SR at 600 °C. This maximum shifts towards higher combustion temperature for the catalyst used in the SR at 700 °C, which suggests that the coke evolves into a more condensed and graphitic-like structure, with lower H/C ratio, and therefore, a higher combustion temperature is required [40].A shoulder burning at low temperature is also observed in the TPO profiles for the samples deactivated at 600 °C with catechol and guaiacol + ethanol (Fig. 4), which is more noticeable for the latter, but is not observed in the SR of the non-phenolic oxygenates (Fig. 3). This result confirms the previously reported relevant role of phenolic compounds as precursors of the coke burning at low temperature (deposited near metal sites, causing its partial or total encapsulation), deposited in the SR of bio-oil over this catalyst [8]. Taking into account the similar amount of this coke type deposited for catechol and for the mixture (guaiacol + ethanol), with only 50 wt% of guaiacol, it can be concluded that the latter is more prone to its formation.The results of TPO profiles also provide information on the content of coke deposited, estimated from the total area under the TPO profiles, because the calculation of coke content from TGA results (Figure S2) is masked by the mass increase due to Ni oxidation. The results of coke content are gathered in Table 1, which also includes the average coke deposition rate, calculated assuming linear coke deposition over the reaction. The coke content notably decreases with reforming temperature in the SR of non-phenolic compounds, but, conversely, it increases in the SR of phenolic compounds (catechol and guaiacol), which suggests a different mechanism of coke formation and evolution for the two groups of compounds. At 600 °C, the amount of coke follows the order: ethanol > acetone > acetic acid ≈ catechol > guaiacol + ethanol > acetaldehyde ≈ acetol > levoglucosan, whereas at 700 °C the order is guaiacol + ethanol > catechol ≫ acetone > ethanol > acetic acid > acetaldehyde > acetol > levoglucosan. It should be noted that S/C ratio used in the SR runs with levoglucosan (S/C = 6), is significantly higher than that used with the rest of oxygenates (S/C = 3), which contributes to the lower coke content obtained with levoglucosan at any temperature.Comparing the results of Table 1 with the deactivation results (Figs. 1 and 2), it is noteworthy that there is no direct relationship between the amount of coke and the deactivation rate. This result has been also reported in previous works on oxygenates reforming [38,45–49] , and is explained by the fact that other characteristics of the coke (morphology, structure and location) have a greater impact on deactivation than its content. In addition, it is observed that with similar TPO profiles in terms of peak position (as is the case of coke for the SR of guaiacol + ethanol and catechol at 600 °C) the deactivation rate is different (much faster in the SR of guaiacol + ethanol). Consequently, although the TPO profile of the coke provides valuable qualitative information on the level of condensation and heterogeneity of the coke, to understand the deactivation of the catalyst it is necessary to complete the information on the coke with other characterization techniques of the deactivated catalyst, which will be shown in subsequent sections.The textural properties of fresh and deactivated samples (BET surface area, average pore diameter and pore volume) have been determined by means of N2 adsorption–desorption and are displayed in Table 1. The N2 adsorption–desorption isotherms of spent catalyst samples are shown in Fig. 5 (ethanol and (guaiacol + ethanol) feeds) and Figure S3 of Supplementary Information (rest of feeds).All the samples have isotherm of type IV, but differently to the fresh-reduced catalyst, a H3-type hysteresis cycle is observed for most of the isotherms of catalyst samples used in SR of pure oxygenate compounds, which does not exhibit any limiting adsorption at high P/P0, and is associated to aggregated of plate-like particles giving rise to slit-shape pores [50]. Overall, the shape of the isotherms in Figs. 5 and S3 for spent catalysts does not change with SR temperature, but there are significant differences in the values of textural properties (BET surface area, mean pore diameter and pore volume, gathered in Table 1). In the region of low partial pressures (P/P0 ≈ 0), the volume adsorbed in samples of catalyst used in the SR of aliphatic oxygenates and catechol at 600 °C increases noticeably due to the high BET surface area of these samples (which is more than double that of fresh catalyst). This result can be explained by the deposition of porous carbon structures, such as carbon filaments (in agreement with the SEM images shown later, in Figs. 9 and 10), and is consistent with the high stability observed in the SR at 600 °C of these oxygenates. In the samples of catalyst used at 700 °C and with the lowest values of coke deposition (SR of acetaldehyde, acetol and levoglucosan), the physical properties resemble those of the fresh catalyst. On the other hand, the significantly lower total volume adsorbed at high pressures (P/P0 ≈ 1) in the sample of the catalyst used in the SR of (guaiacol + ethanol) mixture at 700 °C (Fig. 5a) evidences the partial blockage of the mesopores, thus causing a decrease in BET surface area and mean pore volume (in spite of a high presence of carbon filaments). This partial blockage of the porous structure would contribute to a rapid deactivation, as observed in the SR at 700 °C of this feed (Fig. 2b).The XRD was carried out to analyze the crystalline structure of the catalyst and of the coke, and also to determine the average size after reaction of Ni metal crystals, by means of Debye-Scherrer equation, at 2θ = 51.8° (Ni0 (200) plane). Fig. 6 shows the XRD diffractograms of spent catalyst samples used in the SR of oxygenates at 600 °C (graph a) and at 700 °C (graph b). The XRD diffractogram of the fresh catalyst is also shown in Fig. 6 for comparison. The same diffraction peaks as in the fresh-reduced catalyst are observed in the spent catalyst samples. Therefore, the presence of NiO is not detected, in agreement with H2-TPR results, which corroborates the high reducing capacity of the reaction medium to keep the active metal in a reduced state.Moreover, the XRD pattern of most of the spent catalysts shows the presence of a broad peak at a diffraction angle 2θ = 26 °, which corresponds to high crystallinity cokes (graphite carbon), a characteristic peak usually identified in catalysts used in the steam reforming of pure oxygenate compounds (or some mixtures), such as ethanol [51], acetone [52] or acetic acid [53]. The intensity of this peak is in a reasonable agreement with carbon amounts (Fig. 5). Thus, its intensity is high for all the samples deactivated at 600 °C, except for levoglucosan, and for the samples deactivated at 700 °C with phenolic compounds or acetone in the feed, but it is not observed for levoglucosan, acetaldehyde and acetol, due to their low coke content (<3%). Nevertheless, there is not a linear relationship between the intensity of this peak and the amount of coke, which is consequence of differences in the crystallinity level of the different coke deposits.The calculated average values of Ni0 particles for fresh-reduced and spent catalysts are gathered in Table 2. It should be noted that the calculation is possible only with low-moderate coke content (below 120 wt%), because a high coke content hinders the measurement of metal crystal size from XRD diffractograms (as it masks the Ni0 diffraction peaks). The values of average size of Ni0 crystallites of all used catalysts in Table 2 are around that of fresh catalyst (9 nm), except for SR of levoglucosan, that slightly increases. This moderate sintering of Ni0 crystals could be the responsible of the moderate deactivation rate observed in the SR of levoglucosan (Fig. 2c), in spite of the low coke content deposited in these experiments (Fig. 5). Figs. 7 and 8 show the BSD-SEM images for the catalyst used in the SR of oxygenates at 600 and 700 °C, respectively. The BDS-SEM images allow determining the presence of some type of elements on the external surface of the particles based on the brightness intensity [47]. Thus, the high brightness intensity of the fresh catalyst (Figure S4) indicates the presence of heavy elements (Ni and Al) constituting the catalyst phases (Ni crystals and Al2O3). In contrast, the particles of the spent catalysts are generally homogeneous and exhibit a low brightness intensity (dark appearance), which indicates the majority presence of coke on the particle external surface. However, the catalyst used in the SR of levoglucosan at 700 °C (Fig. 8h) shows an homogeneous high brightness intensity, similar to that of the fresh catalyst, which confirms the very low coke deposition observed in the TPO results. On the other hand, the catalyst used in the SR of acetaldehyde at 700 °C (Fig. 8b) shows heterogeneous particles, some with high brightness intensity and others with a dark appearance, which is indicative of the heterogeneous coke deposition.Additionally, these images also show differences in the particle shapes and textures that can be correlated with the coke content (Table 1). When the coke content is low (below 20 wt%), the particle shape of the spent catalysts (SR of levoglucosan at 600 °C (Fig. 7h) and acetaldehyde, acetol and levoglucosan at 700 °C (Fig. 8b, 8d and 8 h, respectively)) is similar to that of the fresh catalyst, being irregular with a smooth surface and sizes in between 150 and 250 μm (original catalyst particle size). When the coke content is moderately high (between 20 and 120 wt%), the particle texture of the spent catalysts (SR of acetic acid, ethanol and acetone at 700 °C (Fig. 8a, 8c and 8e, respectively)) changes to a rough surface keeping the original catalyst particle size. In particular, the catalyst used in the SR of acetic acid at 700 °C (Fig. 8a) shows bare catalyst particles with fragments of coke shells, evidencing the low mechanical strength of the superficial coke shells. When the coke content is high (above 200 wt%), the particles of the spent catalysts (rest of the experiments) have a rough surface and are remarkably smaller than the original catalyst particle size, which may indicate a collapse of the catalyst particles due to the excessive coke growth. Figs. 9 and 10 show the SE-SEM images of the spent catalyst surfaces. In general, at 600 °C (Fig. 9), the images show the formation of carbon filaments from all the model compounds with different characteristics (heterogeneous in size and texture). In particular, the carbon filaments from SR of ethanol (Fig. 9c) show a rough surface, indicating the growth/deposition of carbon along the filaments. Additionally, the formation of an amorphous carbon phase is observed in the catalyst used in the SR of guaiacol + ethanol (Fig. 9g), and in comparison with the catalyst used in the SR of ethanol (Fig. 9c), this carbon phase is inferred to be formed from guaiacol. At 700 °C (Fig. 10), the SE-SEM images clearly show the predominant formation of carbon filaments from acetic acid, acetaldehyde, ethanol, acetone and catechol. The aforementioned peculiar feature of the carbon filaments from ethanol is highly noticeable at this temperature (Fig. 10c), indicating the growth/deposition of carbon along the filaments is favored. A second carbon phase in between the filaments is observed in the SR of guaiacol + ethanol (Fig. 10g), probably due to the formation of pyrolytic carbon from guaiacol (as explained in discussion section) which is more predominant on some regions of the catalyst surface (Figure S5). On the other hand, the surface of the spent catalysts with low coke content (SR of acetol and levoglucosan, Fig. 10d and 10 h, respectively) resembles that of the fresh catalyst, which confirms no significant coke deposition.To study the location of coke on the catalyst surface, Figure S6 shows contrasts of BSD-SEM and SE-SEM images for selected spent catalyst samples. In the spent catalyst with carbon filaments, Ni crystals are often visualized on the tip of the filaments, but not for all the cases. Interestingly, for the catalyst used in the SR of acetic acid at 600 °C (Figure S6c), a large filament was captured showing various Ni crystals along it, which indicates that various Ni crystals may be involved in the growth of large filaments. To complement these observations, selected spent catalyst samples were also analyzed using TEM, and the images (Figures S7 and S8) evidence the formation of hollow carbon filaments (carbon nanotubes) with thick walls (probably multiwall carbon nanotubes, MWCNT) and the presence of Ni crystals on the tip of or along the filaments with no evidence of sintering. Particularly, the catalysts used in the SR of guaiacol + ethanol at 600 and 700 °C (Figure S7) showed two carbon phases (amorphous and filaments). The presence of Ni crystals on the tip of the filaments is an expected observation based on the tip growth mechanism commonly described for the formation of carbon filaments on different Ni catalysts used in the SR of oxygenates [38,42,54,55]. It also explains the catalyst stability observed in the experiments for the SR of acetic acid, acetaldehyde, ethanol, acetol, acetone, catechol and levoglucosan (Figs. 1 and 2) in spite of the high content of filaments, because Ni crystals are exposed and accessible for the reactants. However, the rapid catalyst deactivation observed for the SR of guaiacol + ethanol in Fig. 2b is associated to the formation of a second carbon phase at 600 and 700 °C, which is also observed in the SR of raw bio-oil [31,34]. This carbon phase is formed from guaiacol and has an amorphous nature at 600 °C and a pyrolytic nature at 700 °C based on the coke combustion characteristics (Fig. 4a and 4b, respectively). Fig. 11 shows the Raman spectra of selected spent catalyst samples to further study the structural properties of coke. All the samples show the typical D (corresponding to disordered aromatic structures, at ∼ 1343 cm−1) and G (corresponding to condensed, ordered or graphitic aromatic structures, ∼1589 cm−1) bands as commonly found for various carbon structures, and the corresponding second-order bands in the 2500–3500 cm−1 region (Figure S9) [31,56,57] . At 600 °C, the G and D bands have similar features for all the spent catalyst samples (SR of ethanol, acetol, guaiacol + ethanol and levoglucosan) with noticeable different intensities for the D band. The intensity ratio between the D and G band (ID/IG) determined from deconvolution (procedure described in the SI document and results summarized in Table 3 ) is notoriously higher for the coke formed from the SR of ethanol, and consecutively decreases for the coke corresponding to acetol, levoglucosan and guaiacol + ethanol. At 700 °C, the D and G band features are significantly different. Thus, the coke formed from ethanol has narrow D and G bands, the G band has a shoulder at 1605 cm−1 and noticeable higher ID/IG ratio. The spectra of the coke formed from levoglucosan present a high noise level, which is coherent with the low coke content determined from the TPO analysis.Moreover, the Raman spectra mostly correspond to carbon nanotubes (CNT) with different structural qualities, in coherency with the results of SEM and TEM analyses, which revealed the presence of CNT in the catalysts used in the SR of ethanol, acetol, guaiacol + ethanol and levoglucosan at 600 °C and ethanol and guaiacol + ethanol at 700 °C. The spectrum for ethanol at 700 °C is very close to that of MWCNT, exhibiting dominant narrow D and G bands [58–60]. The intensity ratio between the D3 (assigned to amorphous carbon) and G bands (ID3/IG) (listed in Table 3) provides an indicator for measuring the quality of carbon nanotubes [59], indicating that those formed from ethanol at 700 °C would have the highest purity having the lowest ID3/IG ratio (0.06).It is also observed that the Raman spectra for the catalyst used in the SR of guaiacol + ethanol at 600 and 700 °C is typical of carbon structures with different degree of order [61], which is in agreement with the formation and deposition of a second carbon phase between the filaments. This result is in agreement with the BSD-SEM (Fig. 8g) and SE-SEM (Fig. 10g) images discussed above. Based on the ID/IG ratio, being higher at 700 °C than at 600 °C, the second carbon phase is predominantly amorphous with ordered domains below 2 nm, but it is more structured at 700 °C, which is consistent with the higher combustion temperature observed in the TPO analysis (Fig. 4b). This relationship between TPO and Raman spectroscopic analyses has been also observed for this catalyst used in the SR of raw bio-oil [31], evidencing the formation of carbon filaments and amorphous carbon with different degree of order, which indicates an analogy between coke deposition in the SR of guaiacol and raw bio-oil.The deactivation of the catalyst in the oxygenates SR can be explained by the steps in Fig. 12 , where the nature of the coke is key.The characterization of deactivated catalyst samples (amount and morphology of coke deposits, sections 3.2.1, 3.2.4 and 3.2.5, as well as physical, metallic and textural properties, section 3.2.2 and 3.2.3) has allowed establishing the deactivation causes of the NiAl2O4 spinel derived catalyst in the SR of the different oxygenates at 600 and 700 °C. Firstly, Ni oxidation has been ruled out as a deactivation cause, due to the absence of reduction peaks (TPR) or NiOx diffraction peaks (XRD measurements) in all the spent catalysts, which is coherent with the highly reducing atmosphere along the SR reactions, and is in agreement with the results reported for the SR of raw bio-oil [31]. Secondly, Ni sintering does not appear to be a relevant cause of deactivation of this catalyst, since it is not observed a significant increase in the average Ni0 crystal size, except for SR of levoglucosan, where a slight deactivation is observed (Fig. 2c). Nevertheless, a similar moderate Ni sintering has been reported in the SR of raw bio-oil at 700 °C with this catalyst [31], which does not explain the rapid deactivation for this reaction. Consequently, the main cause of the rapid deactivation of the NiAl2O4 derived catalyst in the SR of bio-oil and of the guaiacol + ethanol mixture must be attributed to coke deposition.By relating the results of TPO analysis and SEM images of the deactivated samples (sections 3.2.1 and 3.2.4) to the deactivation rate of the catalysts, it has become clear that deactivation is directly related to the nature of the coke, in agreement with previous results in literature for different catalysts [38,45–49] . Thus, a large amount of filamentous coke is deposited in the SR of most of the pure oxygenates studied (especially at 600 °C), but it does not cause a significant impact on the activity of the catalyst. The increase in SBET (Table 1) and the BSD-SEM images (Figs. 7 and 8) for the catalyst used in the SR of oxygenates (such as acetic acid, acetaldehyde, ethanol, acetol and acetone) evidences the deposition of a porous and filamentous coke with contents in the catalyst above 20 wt%, but that does not hinder the access of reactants to metal sites in the reaction time studied. However, for a high time on stream, it can create a slight plug on pores or it may grow as clumps of entangled filaments that encapsulate metal particles [47] , which can originate a decrease in activity as that observed at high reaction time in the SR of ethanol (Fig. 1c).The low values of SBET in the catalyst used in SR of guaiacol + ethanol (only slightly above that of fresh catalyst) is explained by the formation of both i) filamentous coke that is probably stacked on the surface of the catalyst and causes an increase in BET surface area (with high contribution of ethanol to the formation of this type of coke), and ii) an amorphous carbon phase in between the carbon filaments, probably due to the formation of pyrolytic carbon from guaiacol, which is promoted at high temperature, and that clogs the porous structure and contributes to the rapid deactivation observed for the mixture (guaiacol + ethanol) (Fig. 2b). This formation of pyrolytic coke by repolymerization of phenols in bio-oil is well established in the literature [62].According to the literature, the importance of the properties of the catalyst in the nature of the coke should be pointed out. Thus, Zhang et al. [48], observed the prevalent formation of amorphous coke from guaiacol on Ni/Al2O3 catalyst, whereas carbon nanotubes are preferentially formed on Ni/SBA-15 catalyst.Due to the filamentous nature of coke, the catalyst stability is high in the SR of non-phenolic oxygenates, (Fig. 1), as well as in the SR of catechol (Fig. 2a) and levoglucosan (Fig. 2c). Conversely, the catalyst undergoes complete deactivation after 300 min reaction in the SR of the guaiacol + ethanol mixture at both temperatures studied. Taking into account the high stability observed in the SR of ethanol (Fig. 1c), it can be concluded that guaiacol is the responsible of the rapid catalyst deactivation observed in Fig. 2b.The origin of long and heterogeneous carbon filaments in the SR of aliphatic oxygenates (Figs. 9 and 10) can be attributed to the reaction of CO (Boudouard reaction, Eq. 9) and CH4 decomposition (Eq. 8) [51,63,64]. As CH4 decomposition is favoured above 750 °C, in the conditions of this study the main origin of this coke is probably the exothermic Boudouard reaction, whose extent is favoured at lower temperature. Moreover, in the SR of ethanol at 600 °C with the same catalyst, the contribution to the formation of filamentous coke by the route of dehydration to ethylene over the acid sites of the Al2O3 support followed by the ethylene decomposition on the Ni-Al2O3 interface has been proved [36]. Also, acetone is an important precursor of filamentous coke [14,65], which can explain the higher amount of coke deposited in the SR of acetone than in the SR of acetic acid.The formation of filamentous coke is also significant in the SR of the phenolic compounds, as revealed by SEM images (Fig. 9f, 9g, 10f and 10g), and the high combustion peak located at high temperature in the TPO profiles (Fig. 4). But differently to the SR of aliphatic oxygenates, the presence of a small coke fraction burning at low temperature (amorphous and encapsulating coke) is observed in the SR at 600 °C of catechol and more notoriously in the SR of the guaiacol + ethanol mixture (Fig. 4a). For this latter feed, the formation of this amorphous carbon phase could explain the lower amount of filamentous coke deposited at this temperature (Fig. 9g) compared to the SR of ethanol (Fig. 9c). Thus, the formation of encapsulating coke on metal sites hinders the mechanisms of filamentous coke formation, which requires diffusion of C species through Ni metal particles, their precipitation on the base of the Ni crystallite and the formation of a carbon filament growing in size [51,66]. This synergy in the mechanism of formation of each type of coke from each oxygenate makes it difficult to understand the mechanism of coke formation from a complex mixture such as raw bio-oil.The difference in the results of coke amount at 600 and 700 °C can be explained by the effect of temperature on the reactions involved in their formation (mainly Boudouard reaction (Eq. 9) and polymerization reactions) and their elimination (gasification reaction, Eq. (10)). Thus, the polymerization and gasification reactions are favored with the increase in temperature, whereas the Boudouard reaction is disfavored. Consequently, the increase in the reaction rate of gasification and the lower extent of Boudouard reaction explain the sharp reduction of coke amount on the catalyst observed in the SR of aliphatic oxygenates at 700 °C [67]. Nevertheless, in the SR of phenolic oxygenates the coke amount is higher at 700 °C, especially for the guaiacol + ethanol mixture, because guaiacol polymerization (with pyrolytic carbon formation, Fig. 10g) is favored to a greater extent than gasification. A similar result was previously reported for other heavy oxygenates like glucose and m-xylene [14,20].This effect of temperature on coke formation is very important in the reforming of levoglucosan, where the coke amount is 16.7 wt% at 600 °C and 0.5 wt% at 700 °C. Considering the ease of cracking of this oxygenate [68] it can be understood that the increase in the cracking rate favors the SR of the intermediates to a greater extent than their polymerization, which explains the low coke deposition. The extent of thermal cracking is different for each oxygenate in the bio-oil depending on its functionality, which affects the results of the raw bio-oil SR, and in particular deactivation. Moreover, increasing the temperature above 700 °C also favors the gasification of the coke retained in the catalyst, attenuating its development towards filamentous structures. However, as aforementioned, this strategy has the unfavorable effect of Ni sintering. It should be noted that this problem is minimized with the NiAl2O4 spinel derived catalyst, which recovers its spinel structure by controlled calcination, recovering the dispersion and size of the Ni0 crystals in successive reaction-reduction-regeneration cycles [30].In a previous study of the deactivation of the same NiAl2O4 spinel derived catalyst in the SR of raw bio-oil [31] was found that the coke is mainly constituted of short and heterogeneous filaments, representing much lower amounts than those formed in this work from aliphatic oxygenates, being remarkable the presence of amorphous and encapsulating coke. Based on the results of the present work, the formation of this coke may be attributed to the high content of guaiacols and catechols, and heavier phenolic compounds in raw bio-oil, whose polymerization significantly inhibits the mechanisms of formation of abundant and long carbon filaments on the catalyst surface from aliphatic oxygenates present in bio-oil.Consequently, the phenolic compounds have a relevant role in the deactivation of the NiAl2O4 derived catalyst during the SR of raw bio-oil. Nevertheless, the decrease in carbon conversion [31] is faster than that observed in the SR of the guaiacol + ethanol mixture (Fig. 2b), which evidences the significant contribution of other compounds in bio-oil, most probably heavier phenolic compounds, to the deactivation of the catalyst. Moreover, a synergistic effect of the presence of different compounds in bio-oil (with different functionalities) could also contribute to a more rapid deactivation in the SR of bio-oil than in the SR of each pure oxygenated compound. Consequently, in order to establish a mechanism that faithfully represents the reality of coke formation and catalyst deactivation in the SR of raw bio-oil, the study of pure oxygenated model compounds is not sufficient, but studies of co-feeding of binary mixtures and progressively more complex mixtures are required. However, based on the results of this work, it is advisable to separate the phenolic components from the raw-bio-oil to mitigate the deactivation by coke, although this implies a decrease in the H2 yield and the formation of a byproduct stream.The deactivation of the NiAl2O4 spinel derived catalyst in the SR of oxygenates at 600–700 °C is a consequence of coke deposition, whose effect on the deactivation rate highly depends on the oxygenates nature, which determines the coke nature and its deactivation ability. Thus, the formation of filamentous coke from the aliphatic oxygenates by the Boudouard reaction has a reduced deactivation effect, because it does not blocks the porous structure of Al2O3. However, the formation of amorphous and Ni-encapsulating coke in the SR of guaiacol leads to a rapid deactivation of the catalyst. The increase in temperature from 600 to 700 °C has low impact on deactivation because it favors the extent of encapsulating coke formation reactions by polymerization but attenuates the formation of filamentous coke by promoting its gasification.Presumably, in the SR of raw bio-oil a synergy between the mechanisms of coke formation from the different oxygenates present is to be expected. But according to the results of this work, the formation of encapsulating coke from phenolic oxygenates is preferential and inhibits the formation of filamentous coke from aliphatic oxygenates. Consequently, the results of this work can be applied to: i) use guaiacol as oxygenate model to test the stability of new catalysts and adapt the reaction conditions in order to minimizing deactivating coke, and ii) design of pretreatment methods of bio-oil in order to eliminate the guaiacol and phenolic components in order to minimizing the formation of this coke.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.This work has been carried out with the financial support of the Ministry of Science and Innovation of the Spanish Government (grant RTI2018-100771-B-I00 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”), the European Commission (HORIZON H2020-MSCA RISE 2018. Contract No. 823745) and the Department of Education, Universities and Investigation of the Basque Government (Project IT1645-22, IT1218-19 and PhD grant PRE_2021_2_0147 for L. Landa). The authors thank for technical and human support provided by SGIker (UPV/EHU/ERDF, EU).Supplementary data to this article can be found online at https://doi.org/10.1016/j.fuel.2022.124009.The following are the Supplementary data to this article: Supplementary data 1

The catalyst stability, mainly affected by coke deposition, remains being a challenge for the development of a sustainable process for hydrogen production by steam reforming (SR) of bio-oil. In this work, the influence of oxygenates composition in bio-oil on the deactivation by coke of a NiAl2O4 spinel derived catalyst has been approached by studying the SR of a wide range of model oxygenates with different functionalities, including acetic acid, acetone, ethanol, acetaldehyde, acetol, catechol, guaiacol and levoglucosan. A fluidized bed reactor was used in the following conditions: 600 and 700 °C; steam/carbon ratio, 3 (6 for levoglucosan); space–time, 0.034 gcatalyst h/gbio-oil (low enough to favor the rapid catalyst deactivation), and; time on stream, 5 h. The spent catalysts were analyzed with several techniques, including Temperature Programed Oxidation (TPO), X-ray Diffraction (XRD), N2 adsorption–desorption, Scanning and Transmission Electron Microscopy (SEM, TEM) and Raman Spectroscopy. The main factors affecting the catalyst stability are the morphology, structure and location of coke, rather than its content, that depend on the nature of the oxygenate feed. The deposition of pyrolytic and amorphous coke that blocks the Ni sites inhibiting the formation of filamentous carbon causes a rapid deactivation in the guaiacol SR. Conversely, the large amount of carbon nanotubes (CNTs) giving rise to a filamentous coke deposited in the SR of aliphatic oxygenates only causes a slight deactivation. The increase in the temperature significantly reduces coke deposition, but has low impact on deactivation.

Hydrogen is potential as a fuel source has touted for decades, but the technology has never gotten off the ground on a sizeable scale. This is due to complexities with its safety, efficient storage, and very crucial concerns regarding the cost of infrastructure for H2 delivery. This novel finding may have found the solutions to both these encounters. Ammonia as a clean and invulnerable energy carrier comprising H2, offers a potential solution to the complications of storage and expense for usage to produce on-demand in situ hydrogen (David et al., 2014, Yao et al., 2020). As a global commodity, million tons of ammonia produce per annum majorly from the Haber–Bosch process. Moreover NH3 is stable and non-explosive, thus secure to transport, handle, and store (Grinberg et al., 2015). Ammonia can be easily liquefied under moderate circumstances (−33.4 °C at atmospheric pressure or 8.46 bars at 20 °C) (Grinberg et al., 2016).Last several years, several metals like as Cobalt (Podila et al., 2015, Podila et al., 2016, Zhang et al., 2013, Zhang et al., 2014a, Podila et al., 2017), Iron (Ohtsuka et al., 2004), Nickel (Hu et al., 2018, Zhang et al., 2014b, Kurtoğlu et al., 2018), Platinum (Wu et al., 2019, Vajo et al., 1985), Rhodium (Leewis et al., 2006) and Ruthenium (Yin et al., 2004a,b,c, Duan et al., 2010, Huang et al., 2019, Ju et al., 2019) have been examined for ammonia decomposition using various supports and Ru is detected as the best metal catalyst. However, overpriced and unavailability of ruthenium discouraged for its practical application in fuel cells. Thus, there is an inspiration to hunt for economical substitutes such as transition metal oxides. Nickel and Cobalt based catalysts showed better activity for ammonia decomposition.Various materials as support have been examined for H2 production from NH3 such as MgO, SiO2, Al2O3, TiO2, ZrO2, Carbon (activated), porous (meso and micro) materials, multi walled (MWCNTs), etc. (Podila et al., 2015, Zhang et al., 2014b, Yin et al., 2004a,b,c, Ju et al., 2017, Wang et al., 2021, Bell et al., 2020, Gu et al., 2021, Lorenzut et al., 2010). For all tested supports, the multi-walled carbon nano tubes supported metal system has noticed to be the most active, and this is due to high surface area and electronic conductivity. High cost, methanation are the difficulties for the usage of carbon nanotubes as a support material in this reaction. On other hand, studies published that support basicity is essential to a productive catalyst for the NH3 decomposition reaction. The oxides of lanthanum and cerium are favourable for ammonia decomposition reaction and improve catalyst activity because of their basicity (Huang et al., 2019, Lucentini et al., 2019, Lucentini et al., 2021, Huang et al., 2020).The perovskite-type oxides (ABO3) are fascinating materials because they display high ion conductivity, high electron conductivity, and fabulous chemical stability over a wide range of temperature, which has used as catalyst for many catalytic reactions (Amini et al., 2019, Xiaolong et al., 2010, Kajita et al., 2002). Perovskite-like oxides can be edited to create a broad range of catalysts by changing either the A-site and/or the B-site cations with other metal cations, creating solid solutions that favors modification of the physicochemical properties (Sarshar et al., 2011). The La2O3 and CeO2 as support or as promoters has been studied for Ni and Co catalyst for ammonia decomposition. It was found that these oxides enhance catalyst intrinsic activity, and they are active for ammonia decomposition even at high metal loading in these catalysts (Lucentini et al., 2021, Y. Yu et al., 2020, Zhang et al., 2005a).On this basis, it can see those compounds with perovskite structure ABO3 (A = La, B = Ni/Co) shows interesting behavior towards ammonia decomposition reaction. Accordingly, to gain improved activity, we have sought to extend our studies by introducing the Ce along with La in-catalyst preparation. To the extent of our knowledge in literature, only one report is available for ammonia decomposition reaction using perovskite type material (Pinzón et al., 2021). Moreover the authors used much diluted ammonia as reactant (5 %NH3 in Ar).Therefore, our ambition in this study is to synthesize ABO3 (A = La, Ce B = Ni/Co) type perovskite materials for ammonia decomposition reaction by combustion synthesis method. The Citric Acid (CitA) used as organic molecule as fuel. The NH3 decomposition activity will study over these perovskite type materials for environmentally friendly hydrogen production.The ABO3 type perovskite material synthesized using combustion synthesis method. The citric acid (CitA) used as organic fuel and the molar ratio between metal nitrate and CitA maintained as 1:1. The chemicals Lanthanum Nitrate (Merk 99%), Nickel Nitrate (Fluka), Cobalt Nitrate (Aldrich), Cerium Nitrate (Aldrich) and Citric Acid (Aldrich) were utilized in catalyst preparation. To prepare LaNiO3 the required molar amount of lanthanum nitrate and nickle nitrate dissolved in 50 mL of deionized water. To obtain a molar ratio of metal nitrate and CitA as 1, dissolved a required amount of CitA in 50 mL of de-ionized water and mixed to the metal nitrate solution. The primary solution (metal nitrate + CitA) was heated at 80 °C temperature to evaporate water until the solution become viscous gel. Next, the viscous gel heated at 90 °C for 24 h in a water bath. Then the gel dried at 150 °C for 24 h in an oven. The resultant swelled solid ground to powder and executed heat treatment at 650 °C for 6 h in a muffle furnace. The prepared perovskite material denoted as LaNi. The same procedure followed for LaCoO3, La0.5Ce0.5NiO3 and La0.5Ce0.5CoO3 perovskite type materials. These samples specified as LaCo, LaCeNi and LaCeCo respectively.The surface area results and pore-size distribution analysis carried out using Nova Station Quanta chrome device with N2 as the sorbate at −196 °C. The samples out-gassed under vacuum for two hours at 200 °C before analyzing sample. The multi point BET method and DFT method was used for surface area and pore size distribution measurements.X-ray diffraction studies performed using devise from Inel. The fine powder of calcined catalysts loaded in holder and patterns recorded using Co Kα (λ = 1.78 Å) radiation. The phases were identified using PDF database.The micromeritics Auto Chem HP 2950 instrument employed to perform H2 temperature -programmed reduction (H2-TPR) tests. For this a 150 mg of samples was placed into the instrument reactor cell and the catalyst was reduced using 10% H2 in argon with 50 mL/min flow. The catalyst temperature increased at 10 °C·min−1 from 50 to 800 °C. The outlet flow examined with a detector of thermal conductivity (TCD).Sample basicity was studied by CO2 temperature-programmed desorption (CO2-TPD). The analysis performed on micromeritics Auto Chem HP 2950 instrument with TCD detector. Prior to analysis, each sample treated with hydrogen flow at 550 °C for two hours. Then He gas used to flush the sample for 1 h and cool down to 40 °C in a He gas flow. After cooling the sample was exposed to CO2 gas for one hour at 40 °C. Next, the sample replaced with helium gas for one hour. Finally the sample was heated progressively up to 700 °C with a rate 10 °C·min−1.The morphology of prepared samples was studied using of Field Emission Scanning Electron Microscope (FEI Quanta FEG450) unit with an Everhart Thornley detector (ETD) and a back scattering electron detector (VCD). The elemental mapping and compositions of the catalysts measured by EDS.The catalysts performance tests conducted in a fixed-bed reactor at normal pressure. Prior to the test, a 0.1 g sample reduced in N2 and H2 mixture (1:1) gas flow for five hours at 550 °C. After the reduction step, the catalysts exposed to N2 gas for 1 h at 550 °C and cool to 300 °C. 100% NH3 gas fed into the reactor with 6000 h−1 GHSV. The tests operated at 300–600 °C temperature range. The temperature increased via a sequential increments with 50 °C intervals. At each temperature, the reaction was executed till to reach steady state. Reaction products were analyzed online by using gas chromatography unit with TCD and a Poropak Q column. The stability of catalyst tested by conducting the time on study test at 550 °C for 50 h and the products were analyzed continuously.The physical properties of the all-calcined catalysts obtained from the corresponding nitrogen adsorption–desorption isotherms (Fig. 1 A). It is clear from the Fig. 1A that all catalysts showed isotherms of type IV with H3 type hysteresis. The results specify existing aggregates of plate like particles with slit shaped pores. The specific surface area, pore volume and pore width of all prepared perovskite type catalysts are displayed in Table 1 . It can see that the LaNiO3 catalyst showed higher surface area than the LaCoO3 oxide. LaNi and LaCo catalyst showed similar average pore width, but the pore volume is high in case LaNi catalyst. After the substitution of 50% of cerium in place of La the surface areas of both Ni and Co catalyst increased. However, the LaCeNi catalyst showed two times higher surface area than the surface area of LaCeCo catalyst. C.A. Franchini et al. (2014) studied Ce substitution in LaNiO3 mixed oxide with different Ce loadings for glycerol steam reforming. According to this report the replacement of La3+ (1.17 Å) with Ce4+ (0.97 Å) the cerium oxide lattice remarkably affects and increases mean particle diameter. On other hand, many studies reported that segregated CeO2 will form along with perovskite with the substitution of 50% of Ce in place of La (Soongprasit et al., 2012, Su et al., 2014, Pecchi et al., 2008, Kirchnerova et al., 2002). This segregated CeO2 may help in increase of surface area of catalyst. Fig. 1B presents the pore-size distribution of calcined catalysts. From Fig. 1B it clear that the pore-width of Ce substituted catalyst increased also increase pore volume. The catalysts prepared in the present work showed highest surface areas compared to the surface areas of catalysts reported in literature especially in case of Ni catalysts with and without Ce substitution (Amini et al., 2019, Franchini et al., 2014, Soongprasit et al., 2012, Pecchi et al., 2008, Pereniguez et al., 2012).The powder Xrd patterns of all calcined samples are presented Fig. 2 . The diffraction patterns of LaCo and LaNi catalysts in Fig. 2 related to characteristics peaks for perovskite type oxide with rhombohedral phase. The Xrd results showed LaCo and LaNi catalysts contain single phases LaCoO3 [PDF: 01-086-1665] and LaNiO3 [PDF: 00-012-0751] respectively. The LaCo catalyst showed more crystallinity than the LaNi catalyst (from the signal intensities in Fig. 2). Consequently, decreased in surface area will detect in LaCo catalyst than the LaNi catalyst. These results are in good resemblance with surface area results in Table 1 where we detected lower surface area for LaCo catalyst than the LaNi catalyst. After substituting cerium new phases are detected in Co and Ni catalysts. The LaCeCo catalyst showed formation of CeO1.66 [PDF: 01-089-8430], Co3O4 [PDF: 01-074-2120] along with LaCoO3 perovskite oxide phase. Similarly, the LaCeNi catalyst showed formation of CeO1.66 [PDF: 01-089-8430], NiO [PDF: 00-044-1159] along with LaNiO3 perovskite oxide phase.The TPR profiles of LaCo, LaNi, LaCeCo and LaCeNi samples are shown in Fig. 3 . The LaCo catalyst displayed three major peaks. The two overlapped peaks between 200 and 500 °C related to the reduction of surface adsorbed oxygen on the catalyst and reduction of Co3+ to Co2+. Next the peak between 500 and 700 °C could be conferred to the reduction of Co2+ to Co0 (Pereñíguez and Ferri, 2018, Zhao et al., 2017). The hydrogen consumption profile of LaNiO3 showed three peaks. The first two unresolved reduction peaks in 300–400 °C corresponds to the partial reduction of perovskite network leading to the formation of the intermediate oxygen deficient La2NiO4 perovskite phase. The third peak registered at 477 °C matches to reduction of La2NiO4 to Ni0 (Franchini et al., 2014, Batiot-Dupeyrat et al., 2003).The 50% substitution of La by Ce in perovskites brings some modification in redox behavior. In general after addition of Ce, reducibility of active site is favored due to the lattice oxygen mobility toward oxygen deficient areas (Soongprasit et al., 2012). The H2-TPR profiles of the LaCeCo showed three major signals. The combined low temperature peaks having peak maxima at 313 and 367 °C connected to reduction of Co3+ to Co2+ and reduction of segregated CeO2. The high temperature peak with maxima at 510 °C is corresponds to the reduction of Co2+ to Co0 species (Sarshar et al., 2011). The TPR peaks of the LaCeNi catalyst are shifted to lower temperatures (Fig. 3d) indicating easier reduction of Ni species upon substitution of Ce. The half substitution of cerium in place of lanthanum in LaNiO3 leads to an increase of the first peaks with respect to the third one. This result can be ascribed to the existence of isolated NiO and CeO2. The total H2 consumption results of all catalysts for TPR are displayed in Table 1. The hydrogen consumption increased for LaCeNi catalyst compared to that of LaNi catalyst. On other hand, the hydrogen consumption decreased for LaCeCo catalyst in comparison to that of LaCo catalyst. However the reducibility shifted towards low temperature peak in LaCeCo catalyst than in LaCo catalyst (Fig. 3). The results emphasize that the addition of Ce encourages metal reducibility at lower temperature in both Co and Ni systems. In addition substitution of Ce enhances metal reduction in case of Ni system (Table 1) (Amini et al., 2019, Lima et al., 2006, Gallego et al., 2009).The basicity of catalyst is crucial for ammonia decomposition reaction (Yin et al., 2004a,b,c). Basicity strength and basic site distribution of catalysts were studied using carbon dioxide temperature programmed desorption study. The CO2-TPD patterns are displayed in Fig. 4. All catalysts showed desorption peaks at two temperature regions. The first desorption peak observed below 100 °C and the second desorption peak observed in 200–500 °C region. As per the literature reports, the low temperature desorption peak below 100 °C corresponds to weakly adsorbed CO2 on catalyst surface and the desorption peak registered at 200–550 °C related to moderate basic sites. The quantification results of CO2 desorption in 200–550 °C range presented in Table 1. The results clearly refer to that the LaNi catalyst is more basic than the LaCo catalyst. After Ce substitution in La based perovskite the moderate basic sites increased. The LaCeCo catalyst displayed highest basicity among all prepared catalysts.The morphologies of all prepared samples studied by scanning electron microscopy which are presented in Fig. 5 . The morphology of LaNi catalyst in Fig. 5a exhibited flake-like structure. At higher magnification the image Fig. 5e clearly showed that the flake-like structure is with interconnected porous in nature. The LaCo catalyst showed an open structure which contain nano-sized interconnected spherical particles (Fig. 5b & f). After substitution of Ce the morphology of catalyst changed. The LaCeNi and LaCeCo catalysts showed morphology of irregular agglomerated particles with porous nature. It is observed from Fig. 5c and Fig. 5d that the agglomeration seems more in LaCeCo catalyst than in LaCeNi catalyst.For further investigation of metal homogeneity in perovskite-type catalyst elemental mapping performed in several areas. The representative mapped regions are presented in supplementary information. The elemental mapping image of LaNi and LaCo catalysts in Fig.S1 and Fig.S3 clearly showed the distribution of metal is homogeneous. The EDS analysis of LaNi and LaCo catalysts are presented in Fig.S2 and Fig.S4. The quantification results showed 47 and 53 atomic percentage of La and Ni for LaNi catalyst and 48 and 52% of La and Co for LaCo catalyst respectively. The elemental mapping image of LaCeNi and LaCeCo catalysts are presented in Fig. 6 . The LaCeNi catalyst showed homogeneous metal distribution from metal mapping image displayed Fig. 6A. The selected area EDS spectra of LaCeNi catalyst presented in Fig.S5. The quantitative results showed presence of 28, 28 and 44% of La, Ce and Ni in LaCeNi catalyst. The elemental mapping image of LaCeCo catalyst showed in Fig. 6B. An agglomeration of cobalt clearly observed from the Fig. 6B. On other hand, the La and Ce metals distributed homogenously. The EDS results displayed in Fig.S6 and observed 29, 28 and 43% of La, Ce and Co in selected region.The results of catalytic performance for ammonia decomposition reaction over perovskite type catalysts are presented in Fig. 7 . The activity experiments performed from 300 to 600 °C. As the NH3 decomposition reaction endothermic, all prepared samples showed increased activity with increased temperature (Podila et al., 2016, Bell et al., 2020). The results in Fig. 7 clearly showed that the LaNi and LaCo catalyst are active for hydrogen production from NH3 decomposition. As reported by literature the nitrogen desorpton is the rate limiting step for ammonia decomposition. Thus metal nitrogen binding energy is a crucial parameter in the design of good NH3 decomposition catalyst (Bell and Torrente-Murciano, 2016, Yin et al., 2004a,b,c). From the literature reports after Ru the nitrogen binding-energy of Co-based catalysts is nearest to the ideal value (Huang et al., 2020, Torrente-Murciano et al., 2017). However, the ammonia decomposition activity also depends on catalytic support and metal support interaction. In this present study, the La based Ni perovskite type catalyst showed higher activity than La based Co perovskite catalyst. This is due to metal support interaction. Zhang etal reported that La and Ce promoter influence the physico chemical properties of Ni catalyst and causes for increased activity for ammonia decomposition (Zhang et al., 2005b, Zhang et al., 2015, Deng et al., 2012). After the substitution of Ce in La based Ni/Co perovskite derived catalyst the activity rasied substantially. Exclusively LaCeNi catalyst showed 99% ammonia conversion at 550 °C.The Fig. 8 shows the activation energy (Ea) values calculated from Arrhenius graphs (ln(k) vs. 1/T) using the synthesized perovskite derived catalysts for NH3 decomposition from 400 to 550 °C at 6000 h−1 space velocity. The catalysts LaCo, LaNi, LaCeCo and LaCeNi displayed a clear activation énergies 55.7,43.0, 48.4 and 35.9 kJ mol−1 respectively. A comparison of Hydrogen formation rate (H2 mmol/gcat/min) and apperent activation energy (Ea) values of perovskite-derived catalysts from this work and other Ni/Co catalysts reported in literature displayed in Table 2 . From the literature, it is clear that many reported Ni/Co systems showed less hydrogen production rate compared to that of perovskite-derived catalysts in this work. There are few reports showed good hydrogen production rate but the catalyst preparation of these systems are difficult for large-scale application (Hu et al., 2019). Recently Pinzon et al reported lanthanum based perovskites for ammonia decomposition but with 5 %NH3 in Ar reactant (Pinzón et al., 2021). It is clear from Table 2 that the catalysts LaCo and LaNi showed good hydrogen production rate and after Ce substitution i.e., LaCeCo and LaCeNi catalysts the hydrogen production rate increased.The activation energy values of corresponding perovskite-derived catalysts are far lower than the Ea values of Ni/Co-based catalysts for decomposition of NH3 such as 40 %Ni/MgLa [54 KJ mol−1] (Y. Yu et al., 2020), 20 %Ni/LaMg [181 KJ mol−1](Hu et al., 2019), 10% Ni/Al2O3 [53.9 KJ mol−1](Y. Yu et al., 2020), 4 %Ni/Al2O3 [99.5 KJ mol−1](Deng et al., 2012), 4 %Ni/CeO2 [71.0 KJ mol−1](Deng et al., 2012), 5 %Co/MgLa [67 KJ mol−1](Podila et al., 2016), 20 %Co/LaMgO [167 KJ mol−1](Hu et al., 2019), 5 %Co/CNTs [92 KJ mol−1](Pei Yu et al., 2020), 90 %Co/Al [123 kJ mol−1] (Gu et al., 2015) and LiNiO3 [108 kJ mol−1] (Pinzón et al., 2021) etc. The La based Ni catalyst showed lower activation energy in comparison to that of Co catalyst. The replacement of 50% of La with Ce induced significant rise in activity which results decrease in activation energy. These results are obvious that the addition of cerium will enhance the catalytic ability for NH3 decomposition reaction. Among all synthesized catalysts, the LaCeNi catalyst exhibited lowest activation energy. Liu et al. (2016) reported that the La2O3 increases weak and medium strength basicity also it improves the reducibility of metal especially in case of Ni species. The tpr results in Fig. 3 clearly showed the reducibility more favorable in La-based Ni than Co system. Thus, the La based Ni perovskite catalyst showed higher activity than the La based Co perovskite catalyst. After the 50% substitution of La by Ce the medium strength basicity and metal reducibility boosted (Figs. 3 and 4). Thus, the activity increased in comparison to that of without cerium. On other hand the surface area of catalyst increased significantly after the substitution of cerium. Hence the La-Ce based Ni/Co perovskite catalysts showed increased activity. The agglomeration of cobalt in LaCeCo catalyst (from Fig. 6B) is the reason for lower activity than LaCeNi catalyst. The time on study test performed using LaCeNi catalyst at 550 °C for 50 h and the results displayed in Fig. 9 . It is noteworthy that the LaCeNi catalyst showed extremely stable performance in terms of NH3 conversion for 50 h at 550 °C.The La based Ni/Co perovskite catalysts were prepared and used as catalyst precursor for NH3 decomposition. The LaNi sample showed very attractive activity at lower temperature compared to that of LaCo catalyst. The 50% substitution of La by Ce has a substantial effect on the catalytic activity. The NH3 conversion raised extensively both in Ni and Co catalysts. The combination of La, Ce and Ni i.e. LaCeNi catalyst displayed the best performance out of all the synthesized catalysts. The higher catalytic activity of the LaCeNi sample is due to raised surface area, easily reducible and appropriate basicity. The uniform inter-distribution of metal-oxide components confirms good dispersion of nickel species as a result is the achieving extremely stable catalyst.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.The Project was funded by Deanship of Scientific Research (DSR) at King Abdulaziz University, Jeddah, under the grant No. G: 779-135-1441. The authors, therefore, acknowledge with thanks DSR for technical and financial support.Supplementary data to this article can be found online at https://doi.org/10.1016/j.arabjc.2021.103547.The following are the Supplementary data to this article: Supplementary data 1

The La based perovskite type LaMO3 (M = Ni, Co) oxides were prepared by combustion synthesis method using citric acid as organic fuel. These catalyst precursors tested for ammonia decomposition. The LaNiO3 and LaCoO3 catalysts showed good activity for NH3 decomposition. The LaNiO3 catalyst displayed greater activity than LaCoO3. This due to high surface area and easily reducibility of Ni species. A 50% of La was substituted by Ce in both LaNiO3 and LaCoO3 catalysts. A remarkable effect on catalytic performance was observed with the partial substitution of La by Ce in perovskite catalyst especially at lower temperatures. The La0.5Ce0.5NiO3 catalyst exhibited highest activity among all prepared samples. The achieved superior activity is due to boost in surface area, reducibility and suitable basicity. The SEM elemental mapping of La0.5Ce0.5NiO3 catalyst concluded that metal oxide constituents dispersed homogeneously. The La0.5Ce0.5NiO3 catalyst showed excellent stable catalytic performance during 50 h time on study at 550 °C.