Datasets:

Hantao

/

ChemImages

like 1

5862_elsa202000024-fig-0005.jpg

FIGURE 5Open in figure viewerPowerPointElectrochemical detection on EP and NE. (A) Schematics for the sensitive and selective detection strategy of EP in presence of ZnO/MWCNTs/GCE (B) Cyclic voltammograms for the electrochemical behavior of EP at different electrodes in PBS of pH 7.0, (a) Blank solution, (b) Bare GCE, (c) ZnO/GCE, (d) MWCNTs/GCE and (e) ZnO/MWCNTs/GCE. Inset: CV response of ZnO/MWCNTs/GCE at PBS having pH 7.0. Reprinted from Colloids Surfaces A Physicochem. Eng. Asp., 584, Shaikshavali, P.; Madhusudana Reddy, T.; Venu Gopal, T.; Venkataprasad, G.; Kotakadi, V. S.; Palakollu, V. N.; Karpoormath, R., A Simple Sonochemical Assisted Synthesis of Nanocomposite (ZnO/MWCNTs) for Electrochemical Sensing of Epinephrine in Human Serum and Pharmaceutical Formulation, 124038, Copyright (2019), with permission from Elsevier (C) Confocal microscopy of PC12 incubated with 30 mM KCl providing the section of extracellular NE, binding of NE at the surface of C, N doped NiO broccoli-like hierarchy (CNNB-1) with corresponding {110} crystal plane and the oxidation of NE. (D) The amperometric response of NE with successive addition from 0.5-5 µM on CNNB-1 nanoelectrode, and (E) the linear plot of concentration (μM) vs. the current (μA). Reprinted from Biosens. Bioelectron., 100, Emran, M.Y. Y.; Mekawy, M.; Akhtar, N.; Shenashen, M. A. A.; EL-Sewify, I. M. M.; Faheem, A.; El-Safty, S. A. A., Broccoli-Shaped Biosensor Hierarchy for Electrochemical Screening of Noradrenaline in Living Cells, 122–131, Copyright (2018), with permission from Elsevier

All these pathophysiological and social concerns have raised the relevance of the maintenance of an optimum level of DA in the human body, and hence the timely detection of DA is equally significant. The principle of electrochemical detection of DA belongs to the redox reaction of DA into dopamine orthoquinone (DAQ) and vice versa involving 2e-/2H+, which generates a current proportional to its concentration (Scheme #elsa202000024-fig-0009#1).[]

5862_elsa202000024-fig-0013.jpg

SCHEME 5Open in figure viewerPowerPointSchematic depiction of the electrocatalytic reaction mechanism of histamine at the modified electrode

Electrochemical techniques are progressing technique to compete with cumbersome conventional techniques like fluorimetry, electrophoresis, and chromatography for the detection of 5-HT.[] The selective, sensitive, and stable detection of 5-HT using electrochemical techniques is due to the multi-step electro-oxidation of 5-HT involving 2 electrons and 2 protons at the electrode surface as shown in Scheme #elsa202000024-fig-0010#2.[]

5862_elsa202000024-fig-0006.jpg

FIGURE 6Open in figure viewerPowerPointElectrochemical detection of Histamine using metal and metal/metal oxide hybrids. (a) A possible measurement mechanism of sensing of Histamine at copper electrode. (B) Typical voltammogram of histamine on a copper plating electrode. The voltammograms are obtained from blank (a), and the sequential addition of 100(b), 300 (c), 500 (d), and 700 M (e) histamine, respectively. Reprinted from Sensors Actuators, 255, Lin, Y. T.; Chen, C. H.; Lin, M. S., Enzyme-Free Amperometric Method for Rapid Determination of Histamine by Using Surface Oxide Regeneration Behavior of Copper Electrode, 2838–2843, Copyright (2017), with permission from Elsevier. (C) HRTEM OF Au/MnO2 composite (D) The Amperometric responses of the different electrodes (bare SPCE, AuNPs modified SPCE, MnO2 modified SPCE and Au/MnO2 composite modified SPCE) toward various concentrations of histamine. Reprinted from Microchem. J., 155, Knežević, S.; Ognjanović, M.; Nedić, N.; Mariano, J. F. M. L.; Milanović, Z.; Petković, B.; Antić, B.; Djurić, S. V.; Stanković, D., A Single Drop Histamine Sensor Based on AuNPs/MnO2 Modified Screen-Printed Electrode, 104778, Copyright (2020), with permission from Elsevier

Norepinephrine (NE) and epinephrine (EP) (alternatively known as noradrenalin and adrenalin, respectively) are catecholamine neurotransmitters that are produced primarily by the adrenalin glands along with some of the neurons located in the medulla oblongata. Both these neurotransmitters are biogenic monoamines and exhibit a similar core structure where the hydrogen atom attached to the nitrogen in NE is replaced by a methyl group in EP (Table 2).[] As for all catecholamine neurotransmitters, the precursor used for the biosynthesis of NE and EP is the amino acid tyrosine. The initial neurotransmitter produced is DA, which then produces NE with the help of an enzyme dopamine-β-hydroxylase.[] The methylation of NE results in the formation of EP. The detailed process of the formation of NE and EP is given in Scheme #elsa202000024-fig-0011#3.

5862_elsa202000024-fig-0014.jpg

SCHEME 6Open in figure viewerPowerPointPlausible depiction of the non-enzymatic hydrolysis reaction of acetylcholine

Electrochemical detection dominates over other conventional techniques such as chromatography, electrochemiluminescence, and electrophoresis for the nanomolar level detection of EP and NE.[] Similar to other monoamine catecholamines, EP and NE can undergo electro-oxidation owing to the presence of electroactive groups. Under the applied potential, EP or NE gets oxidized to their respective quinone forms resulting in the corresponding oxidation peak.[] The corresponding electro-oxidation of EP and NE is given in Scheme #elsa202000024-fig-0012#4.[]

5862_elsa202000024-fig-0007.jpg

FIGURE 7Open in figure viewerPowerPointTMOs based electrodes for the analysis of Ach. (A,B) SEM micrographs of lichen-like nickel oxide nanostructure with different magnifications. (C) Cyclic voltammograms of MCPE in the absence (curves a) and presence (curves b) of 10 mM ACh. Reprinted from Biosens. Bioelectron., 48, Sattarahmady, N.; Heli, H.; Vais, R. D., An Electrochemical Acetylcholine Sensor Based on Lichen-like Nickel Oxide Nanostructure, 197–202, Copyright (2013), with permission from Elsevier. (D) Plausible electrocatalytic mechanism of DA and ACh at Cu@Cu2O-BNDC/GCE. (E) CV curves of Cu@Cu2O-BNDC/GCE in 0.1 M NaOH solution in the absence (green) and presence (red) of 1 mM ACh at a scan rate of 50 mV s−1. Reprinted with permission from 196. Copyright (2019) American Chemical Society

The real sample detection of His covers two wider aspects; detection in the mast cells or other samples of mammalian body and detection in the food samples like fish sauce.[] The electrochemical detection of His is achievable at bare electrodes, but the requirement of high overpotential and reduced rate of electron and charge transfer processes necessitates the modification of electrodes with suitable materials.[] At the electrode surface, His oxidizes irreversibly to its acetonitrile form by involving 4 electrons ad 4 protons.[] The corresponding reaction is provided in Scheme #elsa202000024-fig-0013#5.[]

5862_elsa202000024-fig-0015.jpg

SCHEME 7Open in figure viewerPowerPointProposed mechanism for the non-enzymatic detection of glutamate by the modified electrodes

ACh is involved in numerous functions both inside and outside the CNS. In the peripheral system, it plays a major role in controlling the voluntary movements at the neuromuscular junctions and the paradoxical vasoconstrictions causing the dilation of normal blood vessels.[],[] Within the brain, it is involved in cognitive functions, including sleep and arousal, consciousness, memory, learning, and attention.[] Hence, along with chronic diseases like Parkinson's, Alzheimer's disease, and depression, its abnormal level may result in various disorders such as abnormal temperature and pressure, impaired memory and learning, and declined motor coordination.[],[] Hence, its effective detection is clinically important. However, the absence of electroactive, fluorophore, or chromophore groups had caused certain hindrance in the effective detection of ACh.[] Electrochemical detection, owing to its cost-effectiveness and simplicity dominates over other conventional techniques like chromatography. However, the non-redox electrochemical nature of ACh demands the modification of electrodes with suitable enzymes like acetylcholine esterase (AChE) and choline oxidase (ChO) for the effective electrochemical detection.[] In the absence of enzymes, the reaction is possible only in the presence of highly conductive materials with exceptional catalytic activity. In a particular non-enzymatic reaction, ACh is hydrolyzed in an alkaline solution to produce acetic acid and choline as given in Scheme #elsa202000024-fig-0014#6, followed by the oxidation of the alcoholic group of choline to the carboxylic acid by TMOs derivative mediators.[]

5862_elsa202000024-fig-0008.jpg

FIGURE 8Open in figure viewerPowerPointTMOs modified glutamic acid (GA) sensors. (A) Fabrication and proposed mechanism for the detection of L-GA and uric acid (UA) by the Co3O4 nanosheets. (B) Current responses of control experiment (bare, particles, and nanosheet-coated electrodes). Reproduced from Ref. 218 with permission from The Royal Society of Chemistry. (C) Schematic illustration of the biosensor design based on mixed ceria and titania nanoparticles and the glutamic acid (GluA) detection principle. (D) Cyclic voltammograms of biosensors with and without ceria–titania oxides in the presence and absence of 1mM GluA in 0.1 M PBS (pH 7.4) and scan rate of 0.05 V/s. (E) Amperometric responses to successive additions of 5 mM GluA and calibration curve (inset) of the GluA biosensor in hypoxic conditions in CSF. Reprinted from Biosens. Bioelectron., 52, Özel, R. E.; Ispas, C.; Ganesana, M.; Leiter, J. C.; Andreescu, S., Glutamate Oxidase Biosensor Based on Mixed Ceria and Titania Nanoparticles for the Detection of Glutamate in Hypoxic Environments, 397–402, Copyright (2013), with permission from Elsevier

The aforementioned deadly effects of the aberrant level of Glu on the human body have raised the importance of its proper detection and quantification. As for the detection of other neurotransmitters, electrochemical detection is one of the effective techniques for the selective, sensitive, and real-time analysis of Glu. But the non-electroactive characteristic of Glu had created a barrier in its straightforward detection, and hence either the modification of the electrode using a highly conductive material or immobilization of an enzyme is imperative for its efficient analysis.[] The next generation glutamate sensors, which is an efficient and cost-effective method that does not even require an enzyme, rather use a metal or metal oxide for mediating the redox reaction of the analyte.[] A typical example is reported by Hussain et al. where L-glutamic acid is detected non-enzymatically and the oxidation of L-glutamic acid to pent-2-enedioic acid occurs via Glu intermediate (Scheme #elsa202000024-fig-0015#7).[] Thus, more studies have been conducted for the introduction of novel materials with high electrochemical activity.

22957_cctc202200013-fig-0001.jpg

Figure 1Open in figure viewerPowerPointIRu-corrected chronopotentiograms for the oxidation (10 mA cm−2, 23±2 °C) of 0.1 M H2SO4(aq.) with FTO electrodes functionalised by galvanostatic oxidation (120 mA cm−2, 23±2 °C, 2 h) of 0.1 M H2SO4(aq.) containing (a) different metal precursor combinations: 5 mM Bi3+ (grey), 5 mM Bi3++5 mM Fe3+ (orange), 5 mM Bi3++5 mM Ni2+ (green), 5 mM Bi3++5 mM Mn2+ (red) and 5 mM Bi3++5 mM Co2+ (blue); and (b) different Bi3+ and Co2+ combinations: 5 mM Co2+ (solid black), 5 mM Bi3+ (solid grey), 5 mM Bi3++10 mM Co2+ (dash-dotted), 5 mM Bi3++0.5 mM Co2+ (dotted), 0.5 mM Bi3++5 mM Co2+ (dashed) and 5 mM Bi3++5 mM Co2+ (solid blue). Deposition solutions with Bi3+ contained 0.3 M (5 mM Bi3+) and 0.03 M (0.5 mM Bi3+) ethylene glycol.

When tested in nominally pure 0.1 M H2SO4, all functionalised electrodes demonstrated catalytic activity for water electrooxidation, which improved during galvanostatic tests at 10 mA cm−2 (Figure #cctc202200013-fig-0001#1a). Except for Fe3+, inclusion of the transition metals examined herein in the electrodeposition solutions increased the catalytic performance of the electrodes after 18–20 h of operation as compared to those modified in the presence of Bi3+ only. Similar to the behaviour observed during electrodeposition (Figure S3e), the cobalt-bismuth-based system demonstrated a rapid improvement in the activity during the initial 2 h of tests and slow improvement afterwards (Figure #cctc202200013-fig-0001#1a). Importantly, online analysis of O2 evolved during experiments with the Co−Bi-based electrodes (Supplementary Video) confirmed 100 % faradaic efficiency for the oxygen evolution reaction (OER) (Figure S4). Higher performance of the cobalt-bismuth system as compared to combinations based on other transition metals examined here agrees well with observations for the conceptually similar OER catalysts based on PbO2 and SbOx. Interestingly, electrodes modified with BiOx only did not produce intense gas evolution, indicating that other processes, possibility H2O oxidation to H2O2, substantially contribute to the measured currents (Supplementary Video). Based on these results, further investigation focused on the best performing Co+Bi combination. Contrasting the voltammetric results, galvanostatic deposition again revealed that the combination of 5 mM Bi3+ and 5 mM Co2+ required the least positive potential during deposition at 120 mA cm−2 (Figure S5e–h) and for the OER in nominally pure 0.1 M H2SO4(aq.) at 10 mA cm−2 (Figure #cctc202200013-fig-0001#1b). Investigation of the effect of the electrodeposition current density on the performance of the Co−Bi-based catalytic layers within the 10–200 mA cm−2 range confirmed that the best results are obtained at 120 mA cm−2 (Figure S6). Thus, the use of the 5 mM Co2++5 mM Bi3+ precursor composition and relatively high deposition current density enabled the best OER performance, which corresponds to the reaction rate of 10 mA cm−2 at approximately 2.0 V vs. RHE at ambient temperature. Importantly and quite unexpectedly, this performance slowly but continuously improved during operation rather than degraded (Figure #cctc202200013-fig-0001#1). Overall, the results discussed above suggest that the catalytic materials electrodeposited onto FTO substrates from the acidic solutions of Bi3+ and Co2+ precursors are not purely Co−Bi-based, but also contain tin oxides. We also hypothesise that the incorporation of tin species and the formation of the true catalytically active state of the material occurs during the initial period of “activation” observed both during the electrodeposition (Figures S3 and S5–S7) and testing (Figures #cctc202200013-fig-0001#1–#cctc202200013-fig-0002#2). Contaminants that could unintentionally improve the catalytic activity and stability, specifically Ag and Pb, were below the limits of detection both in the electrolyte solution and on the electrode surface. The only exception was the electrode analysed after the extended 48 h durability test, which contained a very minor, less than 0.1 at.% with respect to Bi, amount of lead, which likely originated as contamination from the 0.1 M H2SO4 electrolyte. The contaminant that we could not avoid on the electrode surface, even with our best efforts to preclean all the components of the system, was iron. In fact, Fe was even found on the surface of an unmodified FTO electrode that was precleaned with aqua regia. At the same time, iron was below the detection limit in all electrolyte solutions examined, thus indicating its very low amounts in the system. Importantly, initial tests have demonstrated that Fe produces a significantly less active water oxidation electrocatalyst when combined with Bi (Figure #cctc202200013-fig-0001#1a), further supporting the hypothesis that Co is the major catalytically active element of the investigated multicomponent material.

22957_cctc202200013-fig-0002.jpg

Figure 2Open in figure viewerPowerPointIRu-corrected chronopotentiograms for the oxidation (10 mA cm−2, 23±2 °C) of 0.1 M H2SO4(aq.) using (a) Pt−Ti and (b) FTO electrodes modified by galvanostatic oxidation (120 mA cm−2, 23±2 °C, 2 h) of 0.1 M H2SO4(aq.) containing 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+ with no Sn2+ added (FTO – blue; Pt−Ti – dark blue) and in the presence of 0.005 (Pt−Ti – dark red), 0.1 (FTO – light green; Pt−Ti – green) and 1 mM Sn2+ (FTO – orange).

Further, we aimed to investigate the possibility of formation of the Co−Bi-based water oxidation catalyst on a platinised titanium (Pt−Ti) substrate, which is chemically very distinct from FTO. Cyclic voltammetry and chronopotentiograms recorded with Pt−Ti in the electrodeposition solution demonstrated significantly higher current densities at less positive potentials than FTO, which is associated with the catalytic activity of Pt for the WOR and higher conductivity of this substrate (Figure S7). When used for water electrooxidation in 0.1 M H2SO4 at 10 mA cm−2, the Co−Bi-functionalised Pt−Ti electrodes exhibited lower activity with no rapid activation during the initial 2 h as was observed for the FTO electrodes modified in the presence of 5 mM Bi3++5 mM Co2+ precursors in the same manner (Figure #cctc202200013-fig-0002#2a). Although kinetics of electrodeposition processes as well as composition and morphology of electrodeposited materials are most commonly affected by the chemical nature of the substrate, we hypothesised that another factor may be at play here. Although rapid activation of the Pt−Ti electrodes functionalised in this manner was still not observed during tests in 0.1 M H2SO4, their activity progressively improved during water oxidation at 10 mA cm−2 and approached that of the FTO electrodes modified in the 5 mM Bi3++5 mM Co2+ solution with no intentionally added Sn2+ after 16 h of operation (Figure #cctc202200013-fig-0002#2a). Interestingly, the use of the very low concentration (0.005 mM) of Sn2+ provided the best result for the Pt−Ti substrate, suggesting that very low amounts of this precursor, as expected for the corrosion of FTO under conditions employed, are required for improved activity. This is further circumstantially confirmed by the lack of any positive effects of the introduction of additional Sn2+ to the deposition solution on the mode of the catalyst deposition onto FTO (Figure S7). Moreover, although the initial activity of the FTO-supported catalysts was slightly improved by the presence of Sn2+ during catalyst formation, the quasi-stabilised activity was either unaffected (with 0.1 mM Sn2+) or even degraded when excessive amount (1 mM) of the tin precursor was added (Figure #cctc202200013-fig-0006#6b). Overall, the results discussed above suggest that the catalytic materials electrodeposited onto FTO substrates from the acidic solutions of Bi3+ and Co2+ precursors are not purely Co−Bi-based, but also contain tin oxides. We also hypothesise that the incorporation of tin species and the formation of the true catalytically active state of the material occurs during the initial period of “activation” observed both during the electrodeposition (Figures S3 and S5–S7) and testing (Figures #cctc202200013-fig-0001#1–#cctc202200013-fig-0002#2).

22957_cctc202200013-fig-0003.jpg

Figure 3Open in figure viewerPowerPointEvolution of the FTO electrode functionalised by galvanostatic oxidation (120 mA cm−2, 2 h, 23±2 °C) of 0.1 M H2SO4(aq.) containing 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+ during the water oxidation test in 0.1 M H2SO4 at 10 mA cm−2 and 23±2 °C: (a) IRu-corrected chronopotentiogram highlighting the “activation” (green background) and “stability” regions (pink background); (b) UV-vis absorption spectra (corrected for the background FTO absorption), (c–f) photographs along with top- and side-view scanning electron micrographs of (c) a blank unmodified FTO, and FTO electrodes after (d) electrodeposition, (e) 2 h “activation” and (f) 48 h stability test. Yellow dashed circles in panel (e) highlight aggregates developing on the electrode surface during operation.

To further understand transformations of the investigated FTO-supported Co−Bi−Sn-based electrocatalysts, longer-term water oxidation tests accompanied by the characterisation of the electrodes were undertaken. As already discussed above, galvanostatic oxidation of 0.1 M H2SO4 using such electrodes can be tentatively separated into two stages: (i) initial activation (ca 2 h) followed by (ii) stable operation for at least 46 h (Figure #cctc202200013-fig-0003#3a). Electrodes immediately after electrodeposition, as well as after the “activation” and “stability” stages were analysed by a set of physical characterisation techniques. Changes in the UV-vis absorption spectra of the electrodes (Figure #cctc202200013-fig-0003#3b) were consistent with the visually observed intensification of the dark yellow colouration of the catalyst films during water electrooxidation (Figure #cctc202200013-fig-0003#3c–f, Supplementary Video). Although identification of the nature of the multiple absorption bands detected is challenging, the UV-vis data indicate that the absorption profiles of the electrode at different stages of the experiments are qualitatively similar and differ in the absorption intensity only. In other words, the same material progressively develops on the electrode during the electrodeposition and subsequent water electrooxidation tests. Scanning electron microscopic analysis identified a faint coating of the original FTO grains (Figure #cctc202200013-fig-0003#3c) with small particles after electrodeposition (Figure #cctc202200013-fig-0003#3d). After the 2 h “activation” stage, these particles were still present but were accompanied by occasional larger aggregates of ca 100–500 nm in width (Figure #cctc202200013-fig-0003#3e). More significant were the morphological changes induced by the 48 h stability tests. Electrodes derived from such experiments were largely coated with crude particles with the width of ca 500 nm, although the thickness of the coating remained low (below ca 50 nm), notwithstanding enhanced colouration (Figure #cctc202200013-fig-0003#3f). Possibly, the colour intensification was largely associated with the morphological changes to the catalytic layer rather than deposition of additional material, the source of which is not obvious in a nominally pure 0.1 M H2SO4 (note that no deposition occurs onto a blank FTO under the same conditions; Figure S2). Another noteworthy finding was the formation of a very thick layer after deposition from the solution containing 5 mM Bi3+ only (Figure S10), indicating that the presence of Co2+ suppresses the excessive formation of the bismuth-based oxide material. Even after 48 h of the stability test, materials formed on the FTO surface could not be detected by X-ray diffraction, which exhibited only tin(IV) oxide reflections from the FTO substrate (Figure S11). This observation suggests a generally amorphous nature of the deposited films. Energy dispersive X-ray spectroscopic (EDS) analysis of the Co−Bi-functionalised FTO confirmed the presence of bismuth on the surface, but not cobalt, which was likely beyond the detection limit of the method (Figures S12–S14). More sensitive analysis by inductively-coupled plasma mass spectroscopy (ICP-MS) enabled the detection of cobalt, which was present on the electrode surface in the amount of few at.% with respect to bismuth (Table S1). We emphasise that this ICP-MS analysis should be considered semiquantitative rather than truly quantitative due to the very low quantities of the overall material deposited. During initial operation for 2 h, no leaching of Co into the electrolyte solution was detected, but some was dissolved after 48 h of operation. The amount of Bi released to the 0.1 M H2SO4 electrolyte solution during the water oxidation test increased very slowly between 2 and 48 h of tests. When considered together with the potentiometric data (Figure #cctc202200013-fig-0003#3a), these results suggest that the system has achieved a dynamic equilibrium after ca 2 h of operation and did not undergo significant compositional changes further.

22957_cctc202200013-fig-0004.jpg

Figure 4Open in figure viewerPowerPointCo 2p, Fe 2p, Sn 3d and Bi 4f X-ray photoelectron spectra of FTO electrodes functionalised by galvanostatic oxidation (120 mA cm−2, 23±2 °C, 2 h) of 0.1 M H2SO4(aq.) containing 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+ before (blue) and after 2 (green) or 48 h (red) tests for the electrooxidation (10 mA cm−2, 23±2 °C) of 0.1 M H2SO4(aq.). All spectra are normalised to the intensity of the Bi 4f7/2 peak.

Further insights into the composition of the near-surface, most relevant to catalysis, layer of the electrochemically generated Co−Bi-based films, were derived from the X-ray photoelectron spectroscopic analysis (Figure #cctc202200013-fig-0004#4 and Figure S15). Bi 4f spectra of the catalyst at all stages of the experiment demonstrated a single set of signals with 4f7/2 and 4f5/2 peaks at ca 159.0 and 164.5 eV, which are typically reported for Bi2O3. Iron could not be detected in the as-deposited material, and was present at the noise level only at the surface of the Co−Bi-based sample after short and longer-term electrocatalytic tests, attesting that Fe is majorly confined within the bulk of the film, not at the surface, and therefore is unlikely to affect the catalytic activity to any significant extent. After electrodeposition, both cobalt and tin also produced very weak XPS signals, which however notably enhanced after 2 h and especially 48 h of tests for water electrooxidation. The binding energies for SnII and SnIV oxides are reported to be very similar, which complicates the assignment of the major oxidation state of tin in the material. Oxidation state of Co, though always highly ambiguous to determine from XPS, is likely to be 3+ as concluded from the comparisons of the spectral profile detected herein with published data. The Co:Sn:Bi surface composition of the catalyst evolved from the initial ca 0 : 1 : 99 to approximately 1 : 2 : 97 after 2 h of the “activation” period (Figure #cctc202200013-fig-0004#4) followed by further enhancement of the portion of cobalt to ca 2 : 2 : 96 after 48 h of operation. Given that the rate of the oxygen evolution reaction catalysed by the cobalt-bismuth-tin-based material after the “activation” step did not change to a very significant extent, one might conclude that the very small amounts of cobalt exposed to the electrolyte solution at this stage are the major catalytically active species, while further apparent development of Co on the surface during the “stability” stage does not induce strong improvements in the performance. Partial loss of electroactive Co species from the electrode surface observed during the voltammetric experiment in Figure #cctc202200013-fig-0007#7a might be ascribed to the stripping of the additional cobalt oxides formed after 48 h of the durability test, which do not significantly contribute to the catalytic activity (see Figure #cctc202200013-fig-0004#4 and discussions above). Extending the negative reverse potential to the less positive values of 1 and 0.65 V vs. RHE resulted in deactivation of the electrode, most likely due to the reductive stripping of the active species, as concluded from the significantly deteriorated water oxidation rates in the forward sweeps up to ca 2.15 V vs. RHE (Figure S18). However, application of more positive potentials initiated the deposition of the catalyst, which was able to sustain electrooxidation current densities in the reverse sweeps that were only slightly lower than in voltammograms recorded within a “safe” negative switching potential of 1.45 V vs. RHE (Figure S18). Such behaviour strongly resembled that reported for self-healing water oxidation electrocatalysts operating at different pH in the presence of mM-level concentrations of the metal precursors. In contrast, the Co−Bi−Sn-based system reported herein could be electrooxidatively reformed from negligibly small quantities of the precursors leached into the solution from the very thin catalytic film.

22957_cctc202200013-fig-0005.jpg

Figure 5Open in figure viewerPowerPoint(a, b) Transmission electron microscopic characterisation, (c) high-angle annular dark-field image and (d-h) EDS elemental mapping of the material removed from the surface of the Co−Bi-functionalised FTO electrode that was tested for water electrooxidation (10 mA cm−2, 48 h) in 0.1 M H2SO4 at 23±2 °C. Catalyst films were prepared by galvanostatic oxidation (120 mA cm−2, 23±2 °C, 2 h) of 0.1 M H2SO4(aq.) containing 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+.

To understand the microstructural arrangement of tin, cobalt and bismuth oxides within the investigated catalyst, transmission electron microscopic (TEM) analysis of the materials tested for the OER (10 mA cm−2) over 48 h and then removed by sonication from the FTO surface was undertaken. The catalyst was found to largely present densely packed aggregates of nanoparticles of variable size (Figure #cctc202200013-fig-0005#5a). Higher magnification imaging suggests that the catalytic film is only comprised of an amorphous material, which would explain the lack of any XRD signals from the film (Figure #cctc202200013-fig-0005#5b–c and Figure S11). Energy dispersive X-ray spectroscopic mapping in scanning TEM revealed that cobalt, tin and bismuth are uniformly distributed throughout the material, i. e. are intermixed at the nanoscale with no segregation of separate phases (Figure #cctc202200013-fig-0005#5d–h). Thus, XAS analysis shows that 48 h water electrooxidation induces further electrooxidation of all metals present in the catalyst films, with the Co K-, Sn K- and Bi L3-edge XANES shifting closer to Co3+, Sn4+ and Bi3+ references, respectively (Figure #cctc202200013-fig-0006#6a–c). It is also noteworthy that all EXAFS datasets collected for the Co−Bi−Sn catalyst (except for the Sn K-edge data contaminated with the FTO support) are similar in terms of the absence of high frequency contributions which are associated with Co−Co, Co−Bi, or Co−Sn interactions (Figure #cctc202200013-fig-0006#6d–f). First, this indicates that all metal oxide phases are highly disordered (consistent with the lack of any XRD reflections; Figure S11). Second, this suggests that cobalt and tin species can be either randomly adsorbed onto or substitutionally doped into the major BiOx phase (Table S1), resulting in highly uniform distribution through the catalyst material (Figure #cctc202200013-fig-0005#5).

22957_cctc202200013-fig-0006.jpg

Figure 6Open in figure viewerPowerPoint(a–c) XANES, (d1-f1) EXAFS and (d2-f2) FT EXAFS data collected at (a, d) Co K-, (b, e) Sn K-and (c, f) Bi L3-edge for the Co−Bi−Sn-based electrocatalysts after electrodeposition (light blue) and after 48 h water electrooxidation tests (blue). Electrodeposition was undertaken by galvanostatic oxidation (120 mA cm−2, 23±2 °C, 2 h) of 0.1 M H2SO4(aq.) containing 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+ using FTO electrodes. Electrocatalytic tests were undertaken in 0.1 M H2SO4(aq.) at 10 mA cm−2 at 23±2 °C. For the Sn K-edge analysis, catalytic material after the stability test was removed from the FTO surface by scratching off (as-deposited) and exfoliation with a Kapton tape (after 48 h test).

Although rapid activation of the Pt−Ti electrodes functionalised in this manner was still not observed during tests in 0.1 M H2SO4, their activity progressively improved during water oxidation at 10 mA cm−2 and approached that of the FTO electrodes modified in the 5 mM Bi3++5 mM Co2+ solution with no intentionally added Sn2+ after 16 h of operation (Figure #cctc202200013-fig-0002#2a). Interestingly, the use of the very low concentration (0.005 mM) of Sn2+ provided the best result for the Pt−Ti substrate, suggesting that very low amounts of this precursor, as expected for the corrosion of FTO under conditions employed, are required for improved activity. This is further circumstantially confirmed by the lack of any positive effects of the introduction of additional Sn2+ to the deposition solution on the mode of the catalyst deposition onto FTO (Figure S7). Moreover, although the initial activity of the FTO-supported catalysts was slightly improved by the presence of Sn2+ during catalyst formation, the quasi-stabilised activity was either unaffected (with 0.1 mM Sn2+) or even degraded when excessive amount (1 mM) of the tin precursor was added (Figure #cctc202200013-fig-0006#6b). Finally, X-ray absorption spectroscopic (XAS) analysis was undertaken to obtain insights into the structural features of the investigated Co−Bi−Sn-based catalytic materials after electrodeposition and 48 h stability tests (Figure #cctc202200013-fig-0006#6). The X-ray absorption near edge structure (XANES) collected at the Co K-edge for the as-deposited material appeared in between the spectra for the CoIIO and CoIIIOOH. The XANES shifts to higher energy and the pre-edge becomes very similar to that of CoIIIOOH after electrocatalytic tests consistent with further oxidation of the metal and cobalt(III) becoming the dominating state. The first peak in the Co K-edge extended X-ray absorption fine structure (EXAFS) data for the Co−Bi−Sn catalyst shifts to lower energy (see arrow in Figure #cctc202200013-fig-0006#6d1), which indicates a contraction of the first coordination sphere bond lengths and is consistent with an oxidation event (Figure #cctc202200013-fig-0006#6a). Sn K-edge XANES data were challenging to collect due to the interference from the FTO support. For the as-deposited sample, this background contribution was higher as it was scratched off the electrode surface; this is best seen in the Fourier transformed (FT) EXAFS data (Figure #cctc202200013-fig-0006#6e2). The tested film was more carefully exfoliated using a Kapton tape, which minimised the contamination with F-doped SnO2 and exhibited distinct EXAFS pattern (Figure #cctc202200013-fig-0006#6e2). Nevertheless, signals different from those of the FTO substrate could be detected in both cases. XANES data are well fit by a combination of SnIVO2 and SnIICl2 reference materials (Figure #cctc202200013-fig-0006#6b), which indicates the presence of the tin species different from F-doped SnO2 and therefore further confirms the incorporation of this metal into the catalyst material (Figure #cctc202200013-fig-0006#6b). The stabilised sample after 48 h electrocatalytic test contained approximately equimolar amount of SnII and SnIV according to XANES. Bi L3-edge XANES in the as-deposited catalyst films has a distinct shape from the Bi2O3 reference, but the spectrum of the tested material strongly resembles that of the BiIII oxide standard (Figure #cctc202200013-fig-0006#6c). The FT of the EXAFS indicates that the sample has on average longer bond lengths before than after the electrocatalytic measurements (Figure #cctc202200013-fig-0006#6f2), consistent with a higher oxidation state of the tested catalyst. Thus, XAS analysis shows that 48 h water electrooxidation induces further electrooxidation of all metals present in the catalyst films, with the Co K-, Sn K- and Bi L3-edge XANES shifting closer to Co3+, Sn4+ and Bi3+ references, respectively (Figure #cctc202200013-fig-0006#6a–c). It is also noteworthy that all EXAFS datasets collected for the Co−Bi−Sn catalyst (except for the Sn K-edge data contaminated with the FTO support) are similar in terms of the absence of high frequency contributions which are associated with Co−Co, Co−Bi, or Co−Sn interactions (Figure #cctc202200013-fig-0006#6d–f). First, this indicates that all metal oxide phases are highly disordered (consistent with the lack of any XRD reflections; Figure S11). Second, this suggests that cobalt and tin species can be either randomly adsorbed onto or substitutionally doped into the major BiOx phase (Table S1), resulting in highly uniform distribution through the catalyst material (Figure #cctc202200013-fig-0005#5).

22957_cctc202200013-fig-0007.jpg

Figure 7Open in figure viewerPowerPoint(a) Not IRu-corrected cyclic voltammetry (scan rate, v=0.020 V s−1; 10 cycles; black arrows show the scan direction; red arrows in the inset show the evolution of the precatalytic features) and (b) IRu-corrected quasi-steady-state polarisation plots derived from short-term chronopotentiograms in 0.1 M H2SO4(aq.) at 23±2 °C (Figure S16) for the FTO electrode functionalised with the Co−Bi−Sn-based catalyst and used for a 48 h water electrooxidation test at 10 mA cm−2 in the same electrolyte solution at 23±2 °C. Catalytic film was obtained by oxidation (120 mA cm−2, 23±2 °C, 2 h) of 0.1 M H2SO4(aq.) containing 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+. Currents are normalised to the geometric surface area of the electrode.

Well-defined pre-catalytic features that can be attributed to the redox transformations of the surface-confined cobalt species emerged in the initial cycles down to 1.45 V vs. RHE, but partially diminished with cycling (Figure #cctc202200013-fig-0007#7a). However, this did not significantly affect the catalytic water electrooxidation current wave at potentials more positive than ca 1.8 V vs. RHE indicating integrity of the key catalytic species. This stability was further confirmed by short-term chronopotentiometric analysis at current densities varied within the 4–20 mA cm−2 range (Figure S16). The semi-logarithmic polarisation plot constructed using the quasi-stabilised potentials at the examined current densities was close to linear within the investigated narrow range with a corresponding E vs. log j (Tafel) slope of ca 0.08 V dec−1 (Figure #cctc202200013-fig-0007#7b), which is close to the results previously reported for other OER catalysts operating in electrolyte solutions with low pH. Tests at very high current densities up to 400 mA cm−2 revealed a notable increase in the E vs. log j slope, possibly indicating charge-transfer limitations within the modified electrode. More importantly, the catalyst did not show any signs of degradation even under these highly challenging conditions (consider very low material loading and flat electrode morphology), but again demonstrated improvements in the performance (Figure S17). Partial loss of electroactive Co species from the electrode surface observed during the voltammetric experiment in Figure #cctc202200013-fig-0007#7a might be ascribed to the stripping of the additional cobalt oxides formed after 48 h of the durability test, which do not significantly contribute to the catalytic activity (see Figure #cctc202200013-fig-0004#4 and discussions above). Extending the negative reverse potential to the less positive values of 1 and 0.65 V vs. RHE resulted in deactivation of the electrode, most likely due to the reductive stripping of the active species, as concluded from the significantly deteriorated water oxidation rates in the forward sweeps up to ca 2.15 V vs. RHE (Figure S18). However, application of more positive potentials initiated the deposition of the catalyst, which was able to sustain electrooxidation current densities in the reverse sweeps that were only slightly lower than in voltammograms recorded within a “safe” negative switching potential of 1.45 V vs. RHE (Figure S18). Such behaviour strongly resembled that reported for self-healing water oxidation electrocatalysts operating at different pH in the presence of mM-level concentrations of the metal precursors. In contrast, the Co−Bi−Sn-based system reported herein could be electrooxidatively reformed from negligibly small quantities of the precursors leached into the solution from the very thin catalytic film.

22957_cctc202200013-fig-0008.jpg

Figure 8Open in figure viewerPowerPointExtended stability tests of the (a–b) FTO and (c) Pt−Ti electrodes functionalised with the Co−Bi−Sn-based catalyst during chronopotentiometric (10 mA cm−2; “ON” and “OFF” indicate when the current was applied and stopped, respectively) oxidation of aqueous H2SO4 solutions under (a) interrupted (0.1 M H2SO4, 23±2 °C) and (b–c) continuous mode. In panel (b–c) experiments were undertaken in 0.1 M H2SO4 at 23±2 °C (blue), 0.5 M H2SO4 at 23±2 °C (red) and 0.1 M H2SO4 at 60 °C (orange); blue line and light blue shading show mean and standard deviation values for n=4 independent measurements. Electrodeposition was accomplished by galvanostatic oxidation (120 mA cm−2, 23±2 °C, 2 h) of 5 mM Bi3+ (+0.3 M ethylene glycol)+5 mM Co2+ in 0.1 M H2SO4(aq.). Potentials are corrected for IRu losses.

To further highlight the ability of the Co−Bi−Sn-based catalyst to self-repair, i. e. reform after idle periods with no positive potential applied, longer-term galvanostatic oxidation of water was undertaken at 10 mA cm−2 over a 9 day period with periodic interruptions in the applied current density after the initial 100 h of the test (Figure #cctc202200013-fig-0008#8a). During the “current off” periods, the potential of the working electrode dropped to values more negative than ca 0.8 V vs. RHE, which according to the voltammetric analysis should have resulted in the loss of the active components (Figure S18). However, the catalyst essentially immediately restored its activity as soon as electrooxidation was reinitiated after several hours of resting under open circuit conditions. Moreover, at the end of the 9-day-long test, the potential required to sustain the water oxidation rate of 10 mA cm−2 by the Co−Bi−Sn-based catalyst was approximately 1.9 V vs. RHE, which compares well to many noble-metal-free thin-film catalysts (Table S2) operating under similar low-pH conditions at ambient temperature.

4944_chem202200148-fig-0001.jpg

Figure 1Open in figure viewerPowerPoint(top) Representation of different conformers of a 6-membered ring molecule (chair – C, boat – B, envelope – E and, half-chair – H, following the IUPAC notation22). Mercator projection of the Cremer and Pople's puckering coordinates’ sphere for a 6-membered ring. (bottom) −1 sugar moieties in PDB 2WW1 (red dots), 2WW3 (green dots) and 2WZS (purple triangles) are depicted over the Mercator conformational surface. The experiments in Ref. [8] describe the 4C1→4H5/1S5→1S5 conformational itinerary (green arrow), while the expected (yellow arrow) and alternative (blue arrow) pathways are OS2→B2,5→1S5 and 3S1→3H4→1C4, respectively. The blue line represents an analytical function from Ref. [15] connected to the TS mimic conformational space.

Since we talk about the ring mimicry, the conformational study of 6-membered rings (6-rings) is the base to classify and quantify the enzymatic itineraries of GHs. In 1975, Cremer and Pople developed a mathematical expression based on puckering coordinate(s) allowing the graphical representation of the conformational space of a ring. As depicted in Figure #chem202200148-fig-0001#1, a Mercator representation projects the puckering space in a two-dimensional surface where all the conformations of a pyranose ring are presented. In this work, we will use the IUPAC notation of pyranose conformations. By the principle of least nuclear motion, the conformations of the reactants (MC) and products (PC) must surround the TS, following an ideal linear pathway. In α-mannosidases, we observe two main itineraries: OS2→B2,5→1S5 (GH38, 76, and 125; the expected pathway for GH92, Figure #chem202200148-fig-0001#1 – yellow arrow) and 3S1→3H4→1C4 (GH47; the alternative pathway, Figure #chem202200148-fig-0001#1 – blue arrow). Due to the significant double-bond character between the pyranic oxygen and the anomeric carbon, the C2-C1-O5-C5 torsion angle tends to zero, and the conformational space of the TS mimics was recently found around a linear region separating the Mercator surface (Figure #chem202200148-fig-0001#1 – blue line). In the present work we focus on the inverting Ca2+-dependent exo-α-1,2-mannosidase, whose catalytic residues are a proton donor (Glu) and an assistant base (Asp). In Ref. [8], Zhu et al. characterized and crystallized Bacteroides thetaiotaomicron GH92 (BT3990, BtGH92) complex with α-1,2-S-mannobiose 2 (Figure #chem202200148-fig-0002#2) in the presence and in the absence of Ca2+ (MC mimic), and mannoimidazole 4 (Figure #chem202200148-fig-0002#2 – MVL) in the presence of Ca2+ (TS mimic). The MC mimic shows an undistorted −1 sugar moiety (4C1), both in the presence and in the absence of Ca2+. The TS mimic shows a 4H5/1S5 distorted conformation. Connecting both regions (Figure #chem202200148-fig-0001#1, green arrow), the experiments show an unexpected 4C1→4H5/1S5→1S5 pathway. Furthermore, analyzing the available MC mimic complexes (PDB 2WW1 and 2WW3), the catalytic water is not kept in a position allowing the nucleophilic attack (>4 Å), while the TS mimic (PDB 2WZS) shows a water molecule at a distance of ∼3 Å from the anomeric center (Figure #chem202200148-fig-0003#3). This fact and the resulting pathway disrespecting the least nuclear motion principle indicate a possible conformational mismatch between the S-glycoside 3 and the natural substrate 1 (as observed in GH125).

4944_chem202200148-fig-0002.jpg

Figure 2Open in figure viewerPowerPointSchematic representation of the molecular structures of the substrates used to study the hydrolytic mechanism of GH92 α-mannosidases (each ring moiety occupies either the −1 or the +1 subsite as noted in this representation).

In the present work we focus on the inverting Ca2+-dependent exo-α-1,2-mannosidase, whose catalytic residues are a proton donor (Glu) and an assistant base (Asp). In Ref. [8], Zhu et al. characterized and crystallized Bacteroides thetaiotaomicron GH92 (BT3990, BtGH92) complex with α-1,2-S-mannobiose 2 (Figure #chem202200148-fig-0002#2) in the presence and in the absence of Ca2+ (MC mimic), and mannoimidazole 4 (Figure #chem202200148-fig-0002#2 – MVL) in the presence of Ca2+ (TS mimic). The MC mimic shows an undistorted −1 sugar moiety (4C1), both in the presence and in the absence of Ca2+. The TS mimic shows a 4H5/1S5 distorted conformation. Connecting both regions (Figure #chem202200148-fig-0001#1, green arrow), the experiments show an unexpected 4C1→4H5/1S5→1S5 pathway. Furthermore, analyzing the available MC mimic complexes (PDB 2WW1 and 2WW3), the catalytic water is not kept in a position allowing the nucleophilic attack (>4 Å), while the TS mimic (PDB 2WZS) shows a water molecule at a distance of ∼3 Å from the anomeric center (Figure #chem202200148-fig-0003#3). This fact and the resulting pathway disrespecting the least nuclear motion principle indicate a possible conformational mismatch between the S-glycoside 3 and the natural substrate 1 (as observed in GH125). C-analogue of α-1,2-mannobiose 2 (Figure #chem202200148-fig-0002#2, compound 2) was received via previously published procedure using Ni-catalyzed Suzuki-Miyaura cross-coupling reaction followed by stereoselective transformations of obtained C-pseudodisaccharide. Furthermore, we performed the cloning and expression of the wild-type (WT) and an acid catalyst mutant (E494Q) of EfGH92 (see Supporting Information for details). Once obtained the different catalysts, we crystalized EfGH92 (WT) with C-disaccharide 2 and E494Q (mutant) with the natural α-1,2-mannobiose 1, and determined their crystal structures at high resolutions. Further synthetic details, conformational analysis of C-disaccharide 2 in water solution, and results of the experiments are presented in the Supporting Information.

4944_chem202200148-fig-0003.jpg

Figure 3Open in figure viewerPowerPointSchematic diagrams of BT3990 (BtGH92) complexed with α-1,2-S-mannobiose 3 and mannoimidazole 4.8

In the present work we focus on the inverting Ca2+-dependent exo-α-1,2-mannosidase, whose catalytic residues are a proton donor (Glu) and an assistant base (Asp). In Ref. [8], Zhu et al. characterized and crystallized Bacteroides thetaiotaomicron GH92 (BT3990, BtGH92) complex with α-1,2-S-mannobiose 2 (Figure #chem202200148-fig-0002#2) in the presence and in the absence of Ca2+ (MC mimic), and mannoimidazole 4 (Figure #chem202200148-fig-0002#2 – MVL) in the presence of Ca2+ (TS mimic). The MC mimic shows an undistorted −1 sugar moiety (4C1), both in the presence and in the absence of Ca2+. The TS mimic shows a 4H5/1S5 distorted conformation. Connecting both regions (Figure #chem202200148-fig-0001#1, green arrow), the experiments show an unexpected 4C1→4H5/1S5→1S5 pathway. Furthermore, analyzing the available MC mimic complexes (PDB 2WW1 and 2WW3), the catalytic water is not kept in a position allowing the nucleophilic attack (>4 Å), while the TS mimic (PDB 2WZS) shows a water molecule at a distance of ∼3 Å from the anomeric center (Figure #chem202200148-fig-0003#3). This fact and the resulting pathway disrespecting the least nuclear motion principle indicate a possible conformational mismatch between the S-glycoside 3 and the natural substrate 1 (as observed in GH125). While the hydrolytic mechanism of GH92 enzymes remains locked, our goal is to decipher the conformational changes occurring over the −1 sugar from the formation of the Michaelis complex to the delivery of the inverted product. Our most burning question in this work solved the structural differences between the S-glycoside 3 and MVL 4 complexes (Figure #chem202200148-fig-0003#3), and why is the catalytic water not kept close to the S-glycoside. The answer could be related to the abrupt conformational change of the −1 sugar. After the inspection of the structure, we observed the most relevant difference in the vicinity of the Ca2+ cation. In the presence of MVL 4, the catalytic water is coordinated to the calcium cation. In the presence of S-glycoside 3, the hydroxyl group (OH) in carbon 2 (C2) is occupying the position of the catalytic water. With MVL 4, the hydroxyl at C2 interacts with the proton donor (E533).

4944_chem202200148-fig-0004.jpg

Figure 4Open in figure viewerPowerPoint(top) Simplified representative structures of MC, TS and PC of the QM-optimized cluster model of BtGH92 with 1, (middle) the proposed mechanism for GH92 enzymes and (bottom) the chain A of the EfGH92 enzyme in complex with the C-disaccharide 2 (MC mimic), and a zoom view of the puckering coordinates space of the observed −1 sugar moieties in the available experiments and calculations (further details in the Supporting Information).

The optimized cluster model (depicted in Figure #chem202200148-fig-0004#4 as MC) presents a distorted E5/B2,5 conformation for the −1 sugar ring. This result differs significantly from the 4C1 conformation observed in the S-glycoside 3. In our scan calculation, the proton in E533 (equivalent to E494 of EfGH92) was transferred to the glycosidic oxygen, the cleavage of the glycosidic bond, and the nucleophilic attack of the catalytic water were activated by constraining the involved distances. The optimized structures of MC, TS (“TS-like”) and PC are presented in Figure #chem202200148-fig-0004#4. We obtained the potential energy barriers of ΔE≠=14.1 kcal ⋅ mol−1 and ΔE0=−2.7 kcal ⋅ mol−1. This result is in good agreement with the available kinetics parameters. X-ray crystallography showed that WT and the mutant form a tetramer where ligands and Ca2+ ions are allocated in each active site cavity (−1 and +1 subsites). On the one hand, three of the four C-glycosides 2 attached to WT exist in a conformation corresponding to the QM-optimized cluster model (E5/B2,5), while one outlier presents a 1S5/1,4B structure (Figure #chem202200148-fig-0004#4 – green dots). On the other hand, the natural substrate 1 attached to the mutant presents one of the four structures in the E5 region, two of them in the E5/4H5 region, and the last one in the 4H5/1S5 space (Figure #chem202200148-fig-0004#4 – purple dots). This evidence together with the conformations found in the BtGH92-MVL complex 4 indicate a simplified conformational itinerary crossing the E5→1S5 region (Figure #chem202200148-fig-0004#4 – green arrow). Following the −1 sugar conformation along the reaction pathway (Figure #chem202200148-fig-0004#4 – red dots and black dashed line), we can conclude that the itinerary in our cluster model is E5/B2,5→B2,5/E5→1S5/B2,5. Being aware of the subtle conformational changes observed in the −1 sugar both experimentally and computationally, and the dynamic nature of the biochemical systems, the accuracy of results is limited. The experimental results lead to an MC in the vicinity of E5 and a TS crossing 4H5/1S5. While the calculations suggest an MC in E5/B2,5 and a TS in B2,5/E5. In any case, investigated changes are only minor. Nevertheless, when we slightly simplify the nomenclature, we can conclude that GH92 α-1,2-mannosidase follows an E5→B2,5/1S5→1S5 conformational itinerary (Figure #chem202200148-fig-0004#4, middle). In conclusion, computations in this study have newly been a powerful tool to decrypt the reaction mechanism of an enzyme-substrate biosystem. Both experimental results and calculations confirm that the catalytic mechanism of GH92 α-1,2-mannosidases follows the E5→B2,5/1S5→1S5 itinerary (Figure #chem202200148-fig-0004#4). As observed in GH125 enzymes, S-glycosides may not act as good MC mimics, due to different interaction patterns with the active site residues, and, in this case, also, with a metal cation. The conformational difference between the C-glycoside and the S-glycoside allows the approach of the catalytic water to the anomeric carbon, in the first case, where the water is coordinated with the Ca2+ cation. We identify a new class of MC mimic through the substitution of glycosidic oxygen by the CH2 group. This chemical change keeps the conformation of the sugar, but the glycosidic position becomes hydrophobic, suppressing thus the interaction with the catalytic residue.

4944_chem202200148-fig-0005.jpg

Figure 5Open in figure viewerPowerPointActive sites and observed electron densities (2Fo-Fc, 1σ) of (A) WT EfGH92 in complex with C-disaccharide 2, and (B) E494Q EfGH92 in complex with α-1,2-mannobiose. (C) Structural superposition of the active sites of BtGH92 in complex with α-1,2-S-mannobiose 3 (PDB 2WW3, grey and orange), and EfGH92 in complex with 2 (this work, cyan and red). The OH in C2 of the S-glycoside occupies the same position as the catalytic water in presence of the C-disaccharide. (D) Main hydrogen bond interactions observed in the active site of E494Q EfGH92 in complex with α-1,2-mannobiose. The OH in C2 interacts with the free oxygen of the mutated catalytic acid Q494 residue.

The analysis of crystal structures of active sites revealed (Figure #chem202200148-fig-0005#5) that the distorted α-d-mannopyranosyl moiety occupies the −1 subsite, while the +1 subsite is occupied by an undistorted 2-CH2- and 2-O-mannosyl leaving group. Comparing the EfGH92-C-glycoside complex and the PDB 2WW3 (thioglycoside) structure, we observe that the hydroxyl group at C2 of the S-glycoside occupies the same position as the catalytic water in the C-disaccharide complex. Due to the hydrophobic nature of the pseudo-glycosidic methylene group, the E494 catalytic acid residue changes its orientation, and a water molecule interacts with the OH group at C2 of the C-glycoside. The closest water to the anomeric carbon is further than 4 Å in case of the S-glycoside, while a well-oriented catalytic water is coordinated to Ca2+ at ∼3 Å from the anomeric center of the C-glycoside (Figure #chem202200148-fig-0005#5 – C). In Figure #chem202200148-fig-0005#5 – D, we also depict the main hydrogen bond interactions present in the E494Q EfGH92-mannobiose complex. As observed in the QM-optimized BtGH92 cluster model, the OH group at C2 interacts with the oxygen of the mutated glutamine residue Q494. The NH2 group of the mutated residue interacts with the glycosidic oxygen, properly oriented for a hypothetic proton transfer. The hydroxyls at C3 and C4 interact with aspartic acid D313. The catalytic water interacts with two aspartic acid residues, D602 and D604. Finally, the hydroxymethyl arm interacts with a water molecule and serine S66. These observations are in good agreement with the interactions observed in our calculations, strengthening the viability of the model.

4178_elsa202100136-fig-0001.jpg

FIGURE 1Open in figure viewerPowerPoint(A) Schematic diagram of 3D printed electrodes in the horizontal and vertical orientations using ABS/carbon black material. (B) Pictures of vertical printed (VP), horizontal printed smooth surface (HPSS) and horizontal printed rough surface (HPRS) electrodes. Cyclic voltammetric recordings at the different surfaces in the presence of (C) 1 mmol/L ferrocene carboxylic acid in 0.1 mol/L NaOH and (D) 1 mmol/L serotonin hydrochloric acid in tris buffered saline (0.05 mol/L tris and 0.15 mol/L NaCl). (E) Nyquist representations of the impedance spectra (EIS) in the determination of charge transfer (Rct) for VP, HPRS, and HPSS 3D printed electrodes. (F) displays a comparison of the Rct, estimated by EIS fitting analysis. Reprinted with permission from reference 10, Copyright (2018), Springer US

Patel's research group has been working to demonstrate that by varying the printing parameters on a 3D printer, it is possible to enhance the performance of electrochemical sensors without the need of surface posttreatments. To date, two articles have been published providing an in-depth study on this subject. The first work, proposed by Hamzah and co-authors, evaluated the 3D-printing orientation of electrodes composed by ABS/carbon black under horizontal and vertical directions (See Figure #elsa202100136-fig-0001#1A and B). Cyclic voltammograms were recorded using both redox probes ferrocene carboxylic acid and serotonin hydrochloric acid (Figure #elsa202100136-fig-0001#1C and D, respectively). From such figures, it is possible to see that the vertical direction 3D-printed electrode (blue line) provided the better electrochemical performance (higher peak currents and well-defined peaks) for both molecules. A study by electrochemical impedance spectroscopy (EIS) was performed and is shown in Figure #elsa202100136-fig-0001#1E. From the Nyquist plot it is possible to note that the vertical direction 3D-printed electrode (VP) provided the smallest semi-circle which indicate a reduced charge transfer resistance (Rct) (Figure #elsa202100136-fig-0001#1F), corroborating with the cyclic voltammetry measurements (Figures #elsa202100136-fig-0001#1C and D). The authors attributed the achieved findings to the conductive pathway's orientation from electrical connection to solution interface, i.e., the electrode's material internal structure. Although it is possible to conclude that the VP electrode improves the electrochemical performance, it is worth mentioning (from Figure #elsa202100136-fig-0001#1F) that such surface presented an Rct value around 40,000 Ω. Comparing this value with some others achieved in carbon-black-based 3D-printed electrodes that had their surfaces treated, it is possible to conclude that the VP electrode provided the higher Rct value. For instance, Rocha et al. employing a chemical/electrochemical treatment upon the 3D-printed electrode surface (+1.4V followed by -1.0V, each potential held for 200 s in NaOH media), found a very low or negligible Rct value (through EIS measurements) since a semicircle on the Nyquist plot was not observed. Such result is a strong evidence that the chemical/electrochemically-treated electrode surface provides a faster electron transfer compared to the pristine electrode. In the same sense, Wirth and coworkers developed a surface posttreatment applying the electrolysis of water to produce hydroxide ions aiming the insulating thermoplastic (polylactic acid) remotion by saponification. The authors carried out EIS measurements for Proto-Pasta-based electrode and achieved and Rct value of 12,868 Ω (value around three-fold lower than that estimated onto VP electrode). Rocha et al. employing an eco-friendly laser-based treatment acquired an Rct value of 2.3 Ω (very low Rct value).

15977_fig1.jpg

Figure 1Open in figure viewerPowerPointThe refined multiple sequence alignment of the human adrenergic β2 receptor (adrb2, 2RH1) and the dopamine D1 (drd1) and D2 (drd2) receptors. The adrb2 (DSC) bars indicate the transmembrane (TM) helix regions and the second extracellular loop helix (EC2 Helix) in the adrb2 structure. The lysozyme in adrb2 and the third intracellular loop (IC3) in drd1 and drd2 between TM5 and TM6 were excised; this is indicated with a dashed line. The strikethrough amino acid stretch WYRAT was cut out in the template structure. The green ring at the N terminus of TM5 in the drd2 sequence indicates the gap caused by the smaller number of amino acids between the cysteine bridge (EC2-SS-TM3) and TM5. Amino acids marked in dark blue indicate fully conserved positions, medium blue residues have highly similar physicochemical character, and light blue residues have less similar physicochemical character. The conserved cysteine bridge between TM3 and EC2 (EC2-SS-TM3) is indicated. The most conserved residue in each helix is marked with the index 50.

15977_fig2.jpg

Figure 2Open in figure viewerPowerPointSchematic view of the interactions between the full agonist SKF89626 and the dopamine D1 receptor homology model. The typical catecholamine agonist–receptor key interactions with Asp1033.32, Ser1985.42, and Ser2025.46 are shown. The meta-hydroxy group of SKF89626 interacts via hydrogen bonding with Ser1985.42, and the para-hydroxy group interacts with Ser2025.46. In addition, the para-hydroxy also forms a hydrogen bond to Thr1083.37. Phe2896.52 forms a face-to-edge π–π interaction with the agonist, and a methyl–π interaction with Ile1043.33 is formed as well. Polar residues are shown in purple, whereas hydrophobic residues are in green. Blue shades indicate ligand–receptor solvent accessibility.

15977_fig3.jpg

Figure 3Open in figure viewerPowerPointSelected full and partial D1 receptor agonists and structurally similar inactives screened against the new protein structure based pharmacophore model. For a more detailed account of the set, see reference 43.

15977_fig4.jpg

Figure 4Open in figure viewerPowerPointTop (left) and side view (right) of the new receptor-based pharmacophore model superimposed onto the D1 structure model. The transmembrane helix 6 (TM6) and the hydrogen atoms of the interacting amino acids, together with the corresponding excluded volumes, are not shown. The conformation of SKF89626 is taken from the ligand–receptor model complex, while the relative positions of the pharmacophore features are tuned to generate the best hit rate.

15977_fig5.jpg

Figure 5Open in figure viewerPowerPointThe potent partial D1 agonist A70108 (left) together with its enantiomer, the full but less potent agonist A70360. DHX and its three aza analogues 1–3 are shown at right.

15977_fig6.jpg

Figure 6Open in figure viewerPowerPointThe agonist dinapsoline together with the octahydrobenzo[h]isoquinoline analogues 4 and 5; compound 4 is a potent full D1 agonist, and 5 is inactive.

15977_fig7.jpg

Figure 7Open in figure viewerPowerPointTwo orthogonal views of the dopamine D1 (blue) and D2 (yellow) receptor models together with the corresponding full agonists (R)-2-OH-NPA (blue) and SKF89626 (yellow) present in their binding sites. The typical monoaminergic key interacting amino acid residues are shown explicitly. The structures differ particularly in the second and the third extracellular loops (EC2 and EC3), but also in the transmembrane (TM) region, where important interacting amino acids are positioned.

15977_fig8.jpg

Figure 8Open in figure viewerPowerPointSide view of the superposed dopamine D1 (blue) and D2 (yellow) receptor models. Transmembrane helix 6 (TM6) is cut out. The conserved tryptophan residue in TM6 that differs in conformation, together with the non-conserved amino acids in TM3, TM6, and TM7, are included and colored by corresponding receptor. Together with the amino acids in TM2, TM3, and TM7, Trp6.48 forms the D2-characteristic propyl pocket, which is a major contributor to D1/D2 selectivity. Tyr7.43 in drd2 interacts with Asp3.32, whereas the corresponding residue (Trp7.43) is unable to make that bond and is instead rotated toward TM6 and Trp6.48. His6.55 interacts with the meta-hydroxy group of the D2 agonist (R)-2-OH-NPA (yellow), while the corresponding Asn6.55 forms a hydrogen bond with Ser5.43. The D1 agonist SKF89626 (blue) binds deeper in the binding crevice and makes interactions with both Ser3.36 and Asp3.32.

15977_fig9.jpg

Figure 9Open in figure viewerPowerPointRepresentation of the solvent-accessible surface of the D2 (left) and D1 (right) receptors, as viewed from the binding pocket in the direction of the D2-characteristic propyl pocket region. The N-propyl functional group of (R)-2-OH-NPA is included to illustrate the shape difference between the receptors.

31771_open201900103-fig-0001.jpg

Figure 1Open in figure viewerPowerPointThe syn conformers of o-methoxyheteroarenes. The standard nomenclature is given at first. For convenience, the corresponding abbreviation is employed to represent the nomenclature and structure of each compound, where the first digit marks the position of OCH3 on the ring and the second digit in square brackets, [5] or [6], marks the five- or six-membered ring, followed by digits and symbols corresponding to the position of heteroatoms on the ring. Finally, the re-defined abbreviations setting the position of OCH3 as ortho are displayed in italic type.

Figure #open201900103-fig-0001#1 depicts the syn conformers of all kinds of the monocyclic o-methoxyheteroarenes having ortho-heteroatom effect (compounds 1–36), which are calculated at ωB97XD/cc-pVTZ level with Gaussian 09. Note that the syn or anti conformers are defined according to the orientation of methyl group with respect to the in-plane lone pair electrons on the ortho-heteroatom. For those containing more than one ortho-heteroatoms, according to the conventional rules, the priority has the largest atom (higher atomic number) in determining the substituent position. However, since the effects of ortho-heteroatom on conformational preferences have the order N>O or S, the ortho-position of each compound is re-defined accordingly. Table 1 lists the calculated energy differences between syn and anti conformers of o-methoxyheteroarenes having ortho-heteroatom effect. It can be seen that, for all the ethers having an ortho-N heteroatom 1–28, the syn conformers have lower potential energy and are more stable than the corresponding anti ones, suggesting the syn preferences. This means that the ortho-N heteroatom has a strong syn-preferring effect, which is also verified by performing resonance enhanced two-photon ionization experiments on 2-[6]-1N 1 (Figure #open201900103-fig-0002#2) and the cationic spectroscopy. On the other hand, for ethers having an ortho-O or ortho-S but no ortho-N (ethers 29–36), most have anti preferences (ethers 29–34, 36), except for the 2-[5]1O4N5N 35 (with very weak syn conformational preference). This means that the ortho-O or ortho-S heteroatom has a relatively weak anti-preferring effect, which may be affected by other factors. The o-heteroatoms play an imperative role in determining the conformational preferences, while the heteroatoms on other positions have slight effects. So, based on the conformational preferences, all the monocyclic o-heteroaromatic ethers can be divided into two groups. The ethers 1–28 having one ortho-N heteroatom belong to group I, which prefer syn conformers. The ethers 29–36 having only one heteroatom (O or S) at the double ortho-positions belong to group II, which mainly prefer anti conformers. Chein and Corey observed the syn preferring ortho-N and the anti preferring ortho-O or ortho-S heteroatom effects in studying ethers 1, 3, 12, 13, 29 and 32, but, by checking all the o-methoxyheteroarenes, we found an exception, the syn preference of 2-[5]1O4N5N 35. For the o-methoxyheteroarenes, the ortho-heteroatom effects on conformational preferences derived by complete induction method here can be considered as a kind of propensity regularity, which is convenient for application.

31771_open201900103-fig-0002.jpg

Figure 2Open in figure viewerPowerPointThe one-color resonance enhanced two-photon ionization (1C−R2PI) spectrum of 2-[6]-1N 1. The S1←S0 electronic transition energy (E1) of syn conformer is set as 0 cm−1. The E1 of anti conformer is predicted by CIS/cc-pVTZ calculations (the electronic energy corrected by EOM-CCSD/cc-pVTZ) at about 107 cm−1 lower than that of the syn conformer, which is not observed in the 1C-R2PI spectrum, indicating too low population to be detected. The observed bands are assigned based on the cationic spectroscopy.18

Figure #open201900103-fig-0001#1 depicts the syn conformers of all kinds of the monocyclic o-methoxyheteroarenes having ortho-heteroatom effect (compounds 1–36), which are calculated at ωB97XD/cc-pVTZ level with Gaussian 09. Note that the syn or anti conformers are defined according to the orientation of methyl group with respect to the in-plane lone pair electrons on the ortho-heteroatom. For those containing more than one ortho-heteroatoms, according to the conventional rules, the priority has the largest atom (higher atomic number) in determining the substituent position. However, since the effects of ortho-heteroatom on conformational preferences have the order N>O or S, the ortho-position of each compound is re-defined accordingly. Table 1 lists the calculated energy differences between syn and anti conformers of o-methoxyheteroarenes having ortho-heteroatom effect. It can be seen that, for all the ethers having an ortho-N heteroatom 1–28, the syn conformers have lower potential energy and are more stable than the corresponding anti ones, suggesting the syn preferences. This means that the ortho-N heteroatom has a strong syn-preferring effect, which is also verified by performing resonance enhanced two-photon ionization experiments on 2-[6]-1N 1 (Figure #open201900103-fig-0002#2) and the cationic spectroscopy. On the other hand, for ethers having an ortho-O or ortho-S but no ortho-N (ethers 29–36), most have anti preferences (ethers 29–34, 36), except for the 2-[5]1O4N5N 35 (with very weak syn conformational preference). This means that the ortho-O or ortho-S heteroatom has a relatively weak anti-preferring effect, which may be affected by other factors. The o-heteroatoms play an imperative role in determining the conformational preferences, while the heteroatoms on other positions have slight effects. So, based on the conformational preferences, all the monocyclic o-heteroaromatic ethers can be divided into two groups. The ethers 1–28 having one ortho-N heteroatom belong to group I, which prefer syn conformers. The ethers 29–36 having only one heteroatom (O or S) at the double ortho-positions belong to group II, which mainly prefer anti conformers. Chein and Corey observed the syn preferring ortho-N and the anti preferring ortho-O or ortho-S heteroatom effects in studying ethers 1, 3, 12, 13, 29 and 32, but, by checking all the o-methoxyheteroarenes, we found an exception, the syn preference of 2-[5]1O4N5N 35. For the o-methoxyheteroarenes, the ortho-heteroatom effects on conformational preferences derived by complete induction method here can be considered as a kind of propensity regularity, which is convenient for application.

31771_open201900103-fig-0003.jpg

Figure 3Open in figure viewerPowerPointThe switching of conformational preference of (a) 2-[6]1N 1 upon protonation, (b) 2-[5]1O 29 upon ionization, and (c) 2-[5]1S 30 upon ionization.

Conformational modulation is usually based on conformational switching. It is reported that the conformational preferences can be affected by protonation. Figure #open201900103-fig-0003#3 (a) displays the potential curves of 2-[6]1N 1 and its conjugate acid, 2-[6]1NH+ 1 a. For the neutral 2-[6]1N 1, the energy of anti conformer is 4.08 kcal/mol higher than that of the syn one, indicating a syn preference. While for 2-[6]1NH+ 1 a, the energy difference between the two conformers is determined to be −3.23 kcal/mol and a strong anti preference is observed. Even for the ether having only one stable conformer (ethers 50–53, Figure S2), the anti preferences can also be found after ortho-N protonation (Table S2). The potential curves for other ethers 2–28 in group I are given in the Figures S3 and S4. It is demonstrated that, although the neutral or anionic ortho-N remarkably prefer syn conformers, the protonated ortho-N prefers the anti ones. This means that, for the ethers containing ortho-N heteroatom, the conformational preferences can be modulated through protonation or deprotonation on the ortho-N heteroatom. For the ethers of group II, as shown in Table 1, the anti preferences of ethers only having ortho-O or S heteroatom can be observed before and after the protonation, and the syn conformers of the protonated species are nonplanar and relatively unstable. So, it can be concluded that, for all the monocyclic o-heteroaromatic ethers, the protonation on o-heteroatom gives anti preferences. Additionally, the anti preferences of ethers in group II are enhanced by protonation (data for compounds 2-[5]1OH4N5N+ 35 a and 2-[5]1SH4N5N+ 36 a are unavailable due to the failed convergence in optimization steps), and most of protonated ethers in group I & II have strong anti preferences (>2.0 kcal/mol), which might be useful for molecular design. As demonstrated in our previous study, ionization can trigger the conformational switching of some aromatic compounds. The ionization/oxidation effect on the conformational preferences of o-methoxyheteroarenes is investigated, as tabulated in Table 1. Firstly, for the compounds of group I, syn preferences exist before and after the ionization (data for the ionized state of compounds 2-[5]1N4O5N 22 and 2-[6]1N4N5N6N 26 are unavailable due to the failed convergence of optimization steps). Ionization can either enhance or decrease the conformational preferences, but, for application in molecular design, we can find the strong syn preference (>2 kcal/mol) at least in one of two states (before and after ionization) for most ethers of group I, except for the 2-[5]1N4O5N 22, which has a weak syn preference (0.84 kcal/mol) and is unstable in ionized state. For the ethers with only one ortho-O or S heteroatom 29–36 (goup II), the anti preferences can be changed to syn preferences upon ionization, suggesting an ionization-induced conformational switching (as shown in Figure S5). It is to say, although these ethers have anti preferences before ionization, they, similar to ethers of group I, have syn preferences after ionization. For example, in the ionized state, the syn 2-[5]1O+ 29 b and syn 2-[5]1S+ 30 b are more stable than the anti conformers (Figures #open201900103-fig-0003#3(b) and (c)). Marked with conformational distributions (data for compounds 2-[5]1O4N5N+ 35 b and 2-[5]1S4N5N+ 36 b are not available), the ionization reactions are exhibited in Figure S6. So, it can be concluded that, for all the studied ethers, the ionized or oxidized ones have syn preferences. Additionally, enhancement of conformational preferences upon ionization (Figure S7) is observed for those ethers having weak syn preferences (<1 kcal/mol), except for 2-[5]1N4O5N 22 that is unstable in the ionized state.

31771_open201900103-fig-0004.jpg

Figure 4Open in figure viewerPowerPointThe syn conformers of o-heteroaromatic amides containing one or two heteroatoms.

For other heteroaromatic compounds like amides, some of which have recently been explored as drugs, the effect of protonation and oxidation on their conformational preferences are similar to that of heteroaromatic ethers. For the amides containing one or two heteroatoms (Figure #open201900103-fig-0004#4), all the protonated ones are anti-preferring, while all the ionized ones are syn-preferring, as shown in Table 2. Interestingly, although both protonation and ionization introduce a positive charge, they have such different effects on conformational preferences. Note that, for the amides, the syn-preference of ortho-N and anti- preference of ortho-S are similar to those of ethers, however, unlike the anti-preference of ethers containing ortho-O, the syn-preference is found for amides containing ortho-O. Additionally, similar to the ethers, there are also several anomalies among the data for the amides in Table 2, such as the small values (<1 kcal/mol) for energy differences, and the unavailable data due to the failed convergence during the self-consistent field (SCF) or geometrical optimization processes. To confirm the anomalies in Tables 1 and 2 obtained at ωB97XD/cc-pVTZ level, two more computational methods are employed and the results are tabulated in Table 3. Firstly, the calculations are performed at ωB97XD/6-311++g(d,p) level to investigate whether the anomalies are induced by the basis set. As seen from Table 3, it is found that most of the values are similar to the results calculated with cc-pVTZ basis set. For compounds 27 b and 31, although the sign of these values are opposite to those results calculated at ωB97XD/cc-pVTZ level, all these values are very small and the conformational populations Nsyn : Nanti approximate 1 : 1. The conformational energy differences are also calculated by using mp2/6-311++g(d,p), which reveals that the values obtained from mp2 calculations are very similar to those from DFT calculations.

31771_open201900103-fig-0005.jpg

Figure 5Open in figure viewerPowerPointThe potential effects of protonation and ionization on the conformations of drugs. (a) the C3a agonist 54 and the protonation effect on conformational preference of 1,3-thiazole-4-carboxamide 11′; (b) the C3a antagonist 55 and the ionization induced conformational switching of 2-thiophenylcarboxamide 30′; (c) the EGFR inhibitor 56 and the pH-triggered conformational switching of ether 57 (substructure of compound 56), (d) Hsp70, (e) factor Xa, (f) MAP kinase inhibitor.

This simple finding enables easy prediction of conformational preferences under acidification or oxidizing conditions. The pH- and redox-triggered conformational switch can alter the biological function of drugs by dramatically influencing complementarity between drug and receptor. Understanding key factors determining conformation will help to select heteroaromatic templates for elaboration in drug discovery and performance. For the recently developed C3a agonist/antagonist (compounds 54 and 55) from different conformers or several kinds of inhibitors (EGFR 56, Hsp70 58, factor Xa 59 and MAP kinase inhibitor 60) containing structures similar to o-heteroaromatic ethers or amides, the pH or redox effect should be taken into account in their design or function (Figure #open201900103-fig-0005#5). Additionally, the conformational fixation of amide has been used as a new synthetic method, in the ring-closing metathesis to yield medium-sized lactams.

31771_open201900103-fig-0006.jpg

Figure 6Open in figure viewerPowerPointElectrostatic surfaces of 2-[6]1N 1/2-[6]1NH+ 1 a conformers and adenine 61.

The protonation induced conformational switching can also be applied in the experiments of molecular recognition. The combination of 2-[6]1N 1 and adenine 61 was found to occur at low pH, as the electrostatic surfaces of 2-[6]1N 1/2-[6]1NH+ 1 a conformers and adenine shown in Figure #open201900103-fig-0006#6, the preferred anti conformer of 2-[6]1NH+ 1 a is more suitable for combination with adenine 61 through two hydrogen bonds to simulate the nucleic acid base pair (A−T) than the preferred syn conformer of 2-[6]1N 1.

31771_open201900103-fig-0007.jpg

Figure 7Open in figure viewerPowerPointPartial NPA charges on OCH3 and ortho-positions of the syn conformers of (a) 2-[6]1N 1, (b) 2-[5]1N− 2, (c) 2-[5]1O 29 and (d) 2-[5]1S 30 together with their protonated and ionized states.

Figure #open201900103-fig-0007#7 displays the partial natural population analysis (NPA) charges on atoms of OCH3 group and two ortho positions. As shown in Figure #open201900103-fig-0007#7(a), the positive charges on the Hmethyl atoms in compounds 2-[6]1N 1 and 2-[6]1NH+ 1 a are +0.179 and +0.189, respectively. Correspondingly, the ortho-Haryl atoms are also positively charged (+0.221 for 2-[6]1N 1 and +0.262 for 2-[6]1NH+ 1 a). On the other side of OCH3, the charge on ortho-N heteroatom is determined to be −0.509 in 2-[6]1N 1 and −0.485 in 2-[6]1NH+ 1 a, while the charge on the introduced proton is +0.438 in 2-[6]1NH+ 1 a, which shows that, upon protonation, the introduced positive charge is mainly populated on the proton connecting to the ortho-N heteroatom. For the protonated ether 2-[6]1NH+ 1 a, the electrostatic repulsion between the introduced positive proton (+0.438) and the positive Hmethyl atoms of OCH3 group (+0.189) will significantly make the syn conformers unstable. Similar results are also found in Figures #open201900103-fig-0007#7(b)–(c). For 2-[5]1SH+ 30 a shown in Figure #open201900103-fig-0007#7(d), the ortho-S heteroatom becomes more positive upon protonation, indicating that the introduced positive charge is mainly populated on ortho-S instead of the proton itself, which will also destabilize the syn conformer. Additionally, another indispensable factor supporting the anti preferences of protonated ethers is the spatial or steric effects, due to the increased crowdedness caused by the protonation on the o-heteroatoms. However, the quantitative analysis below indicates that the steric effect seems not the determinant. The other reaction introducing a positive charge is ionization. The positive charges on the Hmethyl (+0.217 for 2-[6]1N+ 1 b) and ortho-Haryl (+0.255 for 2-[6]1N+ 1 b) atoms are increased upon ionization, and the increased electrostatic repulsion between them will destabilize the anti conformers. While the negative charge on ortho-C is remarkably decreased upon ionization, and the electrostatic attraction between it and the positive Hmethyl will be weakened, which also destabilize the anti conformers. For the ortho-heteroatoms, as shown in Figures #open201900103-fig-0007#7(a)-(c), these o-heteroatoms become slightly less negative upon ionization (−0.458 for 2-[6]1N+ 1 b, −0.583 for 2-[5]1N. 2 b and −0.418 for 2-[5]1O+ 29 b), and the decreased attractive interactions with Hmethyl are still the supporting factor for the syn preferences observed in these ionized compounds. However, as shown in Figure #open201900103-fig-0007#7(d), the positive charge on ortho-S is increased upon ionization, and the repulsive interaction with Hmethyl is against the syn preferences of 2-[5]1S+ 30 b. The quantitative analysis below shows that, due to the decreased negative charge on ether-O atom, the delocalization effects dominate the syn preferences of 2-[5]1S+ 30 b. As shown in Figure #open201900103-fig-0007#7(a)-(d), the negative charge on ether-O atom is remarkably decreased upon ionization. It seems that the population of the introduced positive charge in a molecule can to some extent determine the conformational preferences. For the protonated compounds, the introduced positive charge is mainly located on the proton or the ortho-heteroatom, due to the increased repulsion between Hmethyl and proton, the protonated molecule will be anti-preferring. While for the ionized compounds, the population of the introduced positive charge on the ether-O, Hmethyl and ortho-Haryl may play important roles in determining the syn preference. These findings rationalize the effects of protonation and ionization on conformational preferences. However, it seems that, especially for few neutral compounds containing ortho-O heteroatom, the conformational preferences cannot be interpreted only by the electrostatic interactions. For instance, the attraction between the positive Hmethyl (+0.174) and the negative hetero-O atom (-0.446) does not support the anti preference of 2-[5]1O 29. While the electronic repulsion between lone-pairs of electrons of the ether-O atom (-0.484) and hetero-O atom also destabilize the anti conformer of 2-[5]1O 29. Similarly, as shown in Figure #open201900103-fig-0007#7(d), the strong repulsion between the positive Hmethyl (+0.208) and the positive hetero-S atom (+0.439) is contradictory to the syn preference of 2-[5]1S+ 30 b. In fact, besides of the electrostatic factor, it has been found that the repulsion and steric effects within molecules act in similar way but cause quite opposite effects. So, the steric and quantum factors also need to be considered.

31771_open201900103-fig-5001.jpg

Scheme 1Open in figure viewerPowerPointDefinitions of the gIGM,inter and ginter functions.

Figure #open201900103-fig-0008#8 displays the results of the IGM analysis for some typical compounds 1, 30, 1′ and 30′ together with their corresponding protonated and ionized forms. As shown in graph (a) of Figure #open201900103-fig-0008#8, for the neutral anti conformers, the isosurfaces are filled by green to red, revealing that the interactions in these regions are mainly repulsive. While for the syn conformers, the properties of the interactions in neutral compounds 1 and 1′ are remarkably different from those in the compounds 30 and 30′. Particularly, the interfragment interactions between substituent (OCH3 or CONH2) and pyridine are explicitly attractive. While in the case of compounds 30 and 30′, the isosurfaces in these regions are mainly colored by green, revealing a very weak attractions. These differences support the results listed in Tables 1 and 2, i. e., the compounds 1 and 1′ are syn preferred while the two thiophene derivatives (compounds 30 and 30′) are slightly anti preferred. For protonation, since our attention are mainly focused on the protonation on ortho-N heteroatom and only some estimated values (without ZPCs) are available for compound 30 a, here only the results of compounds 1 a and 1′ a are exhibited in Figure #open201900103-fig-0008#8(b). For the anti conformers, there are still moderate repulsions between the substituent and the ring. Comparing with the unprotonated syn conformers in Figure #open201900103-fig-0008#8(a), the blue regions in syn 2-[6]1NH+ 1 a and 2′-[6]1NH+ 1′ a are remarkably smaller than those in the neutral 2-[6]1N 1 and 2′-[6]1N 1′, indicating the weaker attractions in these protonated syn conformers. While the red areas are increased in these two protonated syn conformers, which coincides with the protonation-induced conformational switching as shown in Tables 1 and 2. As shown in Figure #open201900103-fig-0008#8(c), the interactions in anti 2-[6]1N+ 1 b are slightly more repulsive than that in anti 2-[6]1N 1, while the strength of attractive interactions in syn 2-[6]1N+ 1 b is similar to the neutral form, revealing the enhanced syn preference upon ionization. In the case of compound 30, the red or orange isosurface between OCH3 and ortho-S heteroatom in syn 2-[5]1S 30 is slightly decreased upon ionization, indicating a decreased repulsion. These changes are not remarkable in the amide 30′, which coincides with the weak conformational preferences listed in Table 2. For the 2′-[6]1N+ 1′ b cation, it is found that the attractive interaction region (blue region between NH2 and ortho-N heteroatom) shrinks remarkably. This shrinking of attraction area may be attributed to the decreasing of negative charge on the ortho-N heteroatom upon ionization, which will weaken the N−H⋅⋅⋅N hydrogen bond. By using Espinosa's model (based on QTAIM data), the energy of NH⋅⋅⋅N hydrogen bond in the molecule 1′ and 1′ b are determined to be −1.89 kcal/mol and −0.57 kcal/mol, respectively, revealing that upon ionization, the NH…N hydrogen bond has been weakened. This supports the above IGM analysis shown in Figure #open201900103-fig-0008#8.

31771_open201900103-fig-0008.jpg

Figure 8Open in figure viewerPowerPointThe plotting for the δginter isosurfaces of interfragment interactions between the OCH3 and the heteroaromatic ring of (a) the neutral compounds, (b) the protonated forms and (c) the ionized forms. The isosurfaces for each conformers are color-filled by the blue-green-red (BGR) color scale method.

30491_ejic201600394-fig-0001.jpg

Figure 1Open in figure viewerPowerPointa) The Zr6O4(OH)4(RCO2)12 secondary building unit, derivatives of which link the majority of Zr MOFs. b) Representation of the solid-state structure of [Zr6O4(OH)4(bpdc)6]n, where bpdc = biphenyl-4,4′-dicarboxylate, commonly known as UiO-67. c) Representation of the solid-state structure of [Zr6O4(OH)4(btb)6(OH)6(H2O)6]n, where btb = benzene-1,3,5-tribenzoate. d) Representation of the solid-state structure of [Zr6(OH)8(FeCl-TCPP)2]n, where TCPP = tetrakis(4-carboxyphenyl)porphyrin, commonly known as PCN-222-Fe. Redrawn from CCDC depositions 1441659, 1000802 and 893545, in turn.

Since the discovery of the UiO-66 series of MOFs by Lillerud et al. in 2008, wherein Zr6O4(OH)4 octahedral secondary building units (SBUs) link twelve linear dicarboxylate linkers each in three dimensions to form a highly porous network, a number of further Zr MOFs have been characterised, with most based on this Zr6 SBU (Figure #ejic201600394-fig-0001#1a and b)., Interpenetrated analogues of the UiO-66 series are known, while the MIL-140 series of MOFs are also linked by dicarboxylates, but connected by one-dimensional Zr-oxide chains to form structures with one-dimensional porous channels. Trigonal tricarboxylate ligands have been shown to link lower connectivity Zr6 SBUs capped by solvents and/or monocarboxylates into two and three dimensional frameworks (Figure #ejic201600394-fig-0001#1c). Planar tetracarboxylates can also be linked by derivatives of the Zr6 SBU into highly porous frameworks, for example, tetrakis(4-carboxyphenyl)porphyrin (PCN-222, MOF-545, Figure #ejic201600394-fig-0001#1d) and 1,3,6,8-tetrakis(p-benzoate)pyrene (NU-1000), while a twelve-connected Zr8O6 SBU has also been observed in a porphyrin-based system. Tetrahedral tetracarboxylates have also been linked by Zr6 SBUs of varying connectivities into three dimensional networks.[], Recently, Zr MOFs have been prepared using 1,2,3-trioxobenzene units as coordinating components in linear ligands, for example MIL-163, with one dimensional Zr–O chains linking the organic moieties. Considerable attention has been directed towards the metalation of free-base porphyrin containing Zr MOFs.[] Two Zr MOFs containing porphyrin units were synthesised using the tetratopic H4-TCPP-H2 [tetrakis(4-carboxyphenyl)porphyrin] ligand, with the traditional 12-connected Zr6O4(OH)4 cluster observed in MOF-525 and an 8-connected Zr6O8(H2O)8 cluster observed in MOF-545, also described as PCN-222 (Figure #ejic201600394-fig-0001#1d shows the solid-state structure of PCN-222-Fe).[] The resulting structures were highly porous with BET surface areas of 2620 m2 g–1 and 2260 m2 g–1 for MOF-525 and MOF-545, respectively, however the high porosity of the structures did not negatively impact their stability, with both MOFs observed to be stable in acidic and aqueous environments. Unlike other porphyrin containing MOFs, the porphyrin units are not metalated as ZrIV ions are not coordinated during synthesis. The presence of both permanent porosity and free porphyrin sites enables opportunities for postsynthetic metalation, with MOF-525 able to be quantitatively metalated with iron chloride to form MOF-525-Fe. The benefits of postsynthetic metalation is realised as MOF-525-Fe could not be obtained by direct synthesis with the pre-metalated ligand (H4-TCPP-FeCl).

30491_ejic201600394-fig-0002.jpg

Figure 2Open in figure viewerPowerPointExamples of covalent postsynthetic modifications at the pendant amino units of UiO-66-NH2.

The vast majority of reported postsynthetic modification protocols are based upon the reactivity of pendant functional moieties on the organic linker of the MOF, with amine groups often exploited due to the wealth of chemical transformations possible (Figure #ejic201600394-fig-0002#2). The first amine containing Zr MOF, UiO-66-NH2, with chemical composition [Zr6O4(OH)4(NH2-bdc)6]n (where NH2-bdc is 2-amino-1,4-benzenedicarboxylate) was described independently by Tilset[] and Cohen in 2010. Both groups subsequently demonstrated the postsynthetic reactivity of the pre-installed amine handle through reactions with a variety of acid anhydrides, resulting in amide functionalised pores. The liquid phase reactions (chlorinated solvents, varying times) were followed by a number of experimental techniques, including FT-IR and 1H NMR spectroscopy, which provide evidence of the percentage conversion, whilst PXRD was used to demonstrate retention of crystallinity during the modification, highlighting the high chemical stability of Zr MOFs. It was observed that the greatest conversion was achieved with smaller anhydrides, likely due to pore size based restrictions.

30491_ejic201600394-fig-0003.jpg

Figure 3Open in figure viewerPowerPointa) Postsynthetic polymerisation (PSP) of UiO-66-NH2 functionalised with methacrylic anhydride. b) Representation of the PSP process used in the construction of stand-alone membranes, starting from UiO-66-NH2. Reproduced (modified) with permission from ref.30 Copyright (2015) Wiley-VCH.

UiO-66-NH2 has been subjected to PSM with methacrylic anhydride, resulting in 67 % conversion, evidenced from 1H NMR spectra of HF digests of the MOF, which was also in good agreement with elemental analysis results. The methacrylate functionality was then used in further transformations (Figure #ejic201600394-fig-0003#3a), where the modified MOF nanoparticles were mixed with butyl methacrylate (BMA) and the photo initiator phenylbis(2,4,6-trimethylbenzoyl)phosphine oxide, and then irradiated with UV light for several minutes. This photoinduced postsynthetic polymerisation (PSP) process results in a square shaped polymer which can be peeled from the Teflon surface as a stand-alone membrane (Figure #ejic201600394-fig-0003#3b).

30491_ejic201600394-fig-0004.jpg

Figure 4Open in figure viewerPowerPointSchematic illustration of a) Diels–Alder cycloaddition, and b) Cu-catalysed azide-alkyne cycloaddition, both carried out postsynthetically on substituted interpenetrated Zr MOFs (PIZOFs).

Zr MOFs containing substituted derivatives of the extended 4,4′-[1,4-phenylenebis(ethyne-2,1-diyl)]-dibenzoate ligand form interpenetrated UiO-66 analogues commonly known as porous interpenetrated zirconium organic frameworks (PIZOFs). One such MOF containing pendant furan moieties was functionalised by Diels–Alder cycloadditions with a variety of substrates (Figure #ejic201600394-fig-0004#4a); reaction with maleimide, N-methylmaleimide or N-phenylmaleimide resulted in conversions of 98 %, 99 % and 89 %, in turn, whilst modest exo/endo selectivities of 24 %, 16 % and 17 % were observed. The authors also report the copper-catalysed azide-alkyne cycloaddition (CuAAC) of another member of the series bearing pendant propargyl moieties. The alkyne units were treated with 4-methylbenzyl azide, using typical CuI-catalysed conditions, resulting in 98 % formation of the triazole product at room temperature (Figure #ejic201600394-fig-0004#4b). This methodology can be viewed as an attractive route for the functionalisation of MOF pores under mild conditions, provided that the CuI can be efficiently removed after modification. Pendant alkyne functionalities have also been incorporated within a UiO-68 analogue containing 2′,5′-diethynyl-p-terphenyl-4,4′′-dicarboxylate ligands. The high porosity of the MOF enables the CuAAC reaction to proceed quantitatively with a number of substrates, namely azidoethane, azidoacetate and azidomethylbenzene, resulting in triazole functionalised MOFs with maintained porosity and crystallinity.

30491_ejic201600394-fig-0005.jpg

Figure 5Open in figure viewerPowerPointa) Tetra-azido modified Zr MOFs undergoing Cu-catalysed azide-alkyne cycloaddition with a variety of substrates. b) Functionalisation of UiO-67-N3 with propargylamine to produce a catalyst for the Knoevenagel condensation between benzaldehyde and ethyl cyanoacetate.

In a similar manner, other groups have focused on the synthesis of Zr MOFs containing pendant azide groups, which are likewise available for PSM by CuAAC. Zhou et al. reported the single crystal structures of a series of UiO-68 type MOFs containing p-terphenyl-4,4′′-dicarboxylate ligands, bearing either two or four pendant methyl or azidomethyl functional groups on the central benzene ring. By judiciously adjusting the molar ratios of the ligands in the synthetic mixture, mixed-ligand MOFs with azide loadings of 0, 25, 50, 75 and 100 % were obtained. The MOFs containing various azide loadings were treated with propargyl alcohol, as well as a variety of other alkyne substrates (Figure #ejic201600394-fig-0005#5a). The introduction of nucleophilic hydroxyl groups results in improved CO2/N2 selectivity performances when compared to the parent azide containing MOFs. In another report, the UiO-68 type MOF containing two pendant azide groups was postsynthetically modified with a tetra-acetylene crosslinker under typical CuAAC conditions. The acetylene containing compound was observed to bridge the organic ligands of the MOF, resulting in the formation of a cross linked MOF (CLM) whilst retaining crystallinity. The authors then selectively destroyed the metal coordination bonds within the CLM by reacting with acidic HCl solutions. The resulting polymer gels were insoluble and retained the shape of the original crystals, hence the MOFs were able to act as a template towards polymer gels with pre-designed architectures. UiO-67-N3, which contains 2-azidobiphenyl-4,4′-dicarboxylate, has been prepared, although the reaction temperature had to be reduced to 80 °C as conventional synthetic methods (140 °C) resulted in thermal cyclisation of the ligand to 9H-carbazole-2,7-dicarboxylate in situ. This phenomenon also results in a poor thermal stability of the MOF, as thermal cyclisation results in a bent ligand geometry that cannot be accommodated by the framework, resulting in collapse of the overall network structure. Low chemical stability was observed, with crystallinity lost when the MOF was immersed in common organic solvents, however framework integrity was retained in DMF, allowing CuAAC reactions with alkyne substrates to be performed. Quantitative conversion was observed when the MOF was modified with methyl propiolate, 3-butyn-1-ol and propargylamine, and interestingly the modified products demonstrated improved chemical and thermal stabilities when compared with the parent framework. The pendant amino group of the propargylamine modified framework was an efficient catalyst for the Knoevenagel condensation between benzaldehyde and ethyl cyanoacetate (Figure #ejic201600394-fig-0005#5b), with the MOF sufficiently stable to be reused for multiple catalytic cycles.

30491_ejic201600394-fig-0006.jpg

Figure 6Open in figure viewerPowerPointSchematic illustration of the thiol-ene radical addition of ethanethiol to UiO-68-allyl, with the single crystal images highlighting that the crystals remain intact during the transformation (scale bar = 200 µm). Reproduced (modified) with permission from ref.37 Copyright (2015) Elsevier.

UiO-68-allyl, comprising 2′,5′-bis(allyloxy)-p-terphenyl-4,4′′-dicarboxylate ligands, was synthesised and subject to postsynthetic modification by thiol-ene radical addition, through near quantitative reaction of the allyl moiety with ethanethiol under UV light in the presence of 2,2-dimethoxy-2-phenylacetophenone as a photoinitiator (Figure #ejic201600394-fig-0006#6). Thiol-ene and thiol-yne additions represent very useful routes to access functionalised MOFs, as they allow the mild insertion of a variety of functional groups across unsaturated double and triple bonds, although success thus far has been limited to pendant groups.

30491_ejic201600394-fig-0007.jpg

Figure 7Open in figure viewerPowerPointa) Schematic illustration of the postsynthetic oxidation of thiol units to form sulfonic acid groups. b) An Arrhenius plot of UiO-66-(SO3H)2 at 90 % relative humidity, revealing an activation of energy of 0.32 eV for proton conduction. Reproduced (modified) with permission from ref.41 Copyright (2015) Wiley-VCH.

MOFs bearing pendant carboxylic acid or sulfonic acid groups are of broad interest as they have been shown to result in improved characteristics including enhanced CO2 selectivity and proton conductivity. These additional acidic groups provide further sites for attachment to metal ions, which in many cases disrupts direct MOF formation, and so introduction of acidic groups onto the backbone of the MOF has been investigated by a postsynthetic oxidation approach. To generate sulfonic acid moieties, UiO-66-(SH)2 was synthesised directly from 1,4-dicarboxybenzene-2,5-dithiol, then the MOF was subjected to harsh chemical conditions: an aqueous 30 % H2O2 solution for oxidation of the thiol groups, followed by an acidic solution for protonation, resulting in quantitative conversion to UiO-66-(SO3H)2 (Figure #ejic201600394-fig-0007#7a). The postsynthetic process was completed in as little as 90 min, resulting in quantitative conversion, while PXRD was used to prove the retention of crystallinity of the MOF. Introduction of the acidic sulfonic groups greatly improved the hydrophilicity of the MOF, which was beneficial for proton conduction. UiO-66-(SO3H)2 demonstrated an excellent superprotonic conductivity of 8.4 × 10–2 Scm–1 at 80 °C and 90 % relative humidity, and an activation energy for proton transfer of 0.32 eV (Figure #ejic201600394-fig-0007#7b). This is the highest reported value of protonic conductivity of MOFs and is comparable to the performance of Nafion, a perfluorinated polymer membrane that is the most effective polymer electrolyte known, and as such this material could potentially be incorporated into proton-exchange membrane fuel cells as an alternative energy technology.

30491_ejic201600394-fig-0008.jpg

Figure 8Open in figure viewerPowerPointSchematic illustration of the redox switching behaviour of UiO-68-(OH)2 to the quinone form, which was unable to be synthesised directly (scale bar = 50 µm). Reproduced (modified) with permission from ref.43 Copyright (2015) American Chemical Society.

Redox processes have also been observed to occur postsynthetically in a single-crystal to single-crystal (SCSC) manner in UiO-68-(OH)2, which contains 2′,5′-dihydroxy-p-terphenyl-4,4′′-dicarboxylate ligands. The ligand within the MOF could be oxidised in the solid phase from the diol to the quinone form, resulting in altered spectroscopic properties (Figure #ejic201600394-fig-0008#8). The transformation was fully reversible, and it was observed that the transformation occurs on the outmost surface of the crystals and gradually diffuses inwards. Interestingly, the MOF could not be directly synthesised with the quinone based ligand, however incorporating these organic-based molecular switches into the solid phase may prove useful for applications such as memory storage or redox-based electronic devices.

30491_ejic201600394-fig-0009.jpg

Figure 9Open in figure viewerPowerPointConditions for postsynthetic hydroxylation and chlorination directly on the aromatic rings of UiO-66 and UiO-66-NH2 respectively. Exposure of UiO-66-NH2 to excess chlorine gas eventually results in its decomposition.

There are considerably fewer examples of postsynthetic modification of MOFs at integral linker sites rather than on pendant functional groups, likely as a consequence of the generally low reactivity of the largely structural components. The chemical stability of Zr MOFs means they can tolerate harsh reaction conditions, facilitating reactions directly on unfunctionalised aromatic rings. Bradshaw et al. showed that UiO-66 was stable in the presence of hydroxyl radicals in water to selectively form UiO-66-OH by PSM with no loss of crystallinity. The·OH radicals can be generated by UV-A irradiation of the superparamagnetic photocatalytic composite material γ-Fe2O3@SiO2@TiO2 or UV-C irradiation of aqueous H2O2 solutions, with the former method resulting in a higher hydroxylation yield (77 % vs. 41 % conversion, Figure #ejic201600394-fig-0009#9), and both higher than achievable by postsynthetic exchange (see Section 6). Introduction of free hydroxyl units to MOFs by PSM is valuable, as they are often coordinated to metals during direct synthesis. Similar chemical stability of UiO-66 in the presence of Cl2 gas was demonstrated by DeCoste and Peterson, who showed that UiO-66-NH2 underwent electrophilic aromatic substitution in the presence of chlorine to form UiO-66-NH2-Cl. Further exposure to chlorine resulted in eventual degradation of the MOF and pore collapse (Figure #ejic201600394-fig-0009#9), but UiO-66-NH2 exhibits a Cl2 storage capacity of 124 % w/w, and is a candidate material for air purification.

30491_ejic201600394-fig-0010.jpg

Figure 10Open in figure viewerPowerPointa) Schematic of Zr MOFs that have been halogenated at unsaturated C–C bonds. Crystal structures of b) trans-edb–I2–H2, c) trans,trans-bdb–I4–H2, and d) trans,trans-peb–I4–H2 isolated from iodinated Zr MOFs showing exclusive trans stereochemistry. Redrawn from CCDC depositions 1400977, 1443200 and 1443201, in turn. Single-crystal to single-crystal transformation of e) [Zr6O4(OH)4(edb)6]n to f) [Zr6O4(OH)4(edb-Br2)6]n, redrawn from CCDC depositions 1062508 and 1062510. g) Iodine uptake of the MOFs described in part a). Reproduced with permission from ref.48 Copyright (2016) Wiley-VCH. h) Comparison of elastic moduli vs. indentation depth for [Zr6O4(OH)4(edb)6]n and [Zr6O4(OH)4(edb-Br2)6]n measured by nanoindentation. Reproduced with permission from ref.[10] Copyright (2016) Royal Society of Chemistry.

The efficacy of reactions such as the Sonogashira and Heck couplings to link aromatic units means that many MOF linkers contain integral carbon–carbon double and triple bonds in their backbones. Forgan et al. have demonstrated that Zr MOFs possess the requisite mechanical stability[] to allow halogenation of integral unsaturated bonds of a series of materials, where the change in hybridisation of linker carbon atoms results in an overall mechanical contraction. UiO-66 topology MOFs containing integral alkene, alkyne and butadiyne units (Figure #ejic201600394-fig-0010#10a) were quantitatively brominated by immersing them in chloroform solutions containing Br2, with no decrease in porosity or crystallinity. As the linkers have restricted dynamic motion when bound within the MOFs, the transformations are stereoselective, yielding exclusively trans-dihaloalkene and meso-dihaloalkane units (Figure #ejic201600394-fig-0010#10b–d). The MOFs have sufficient mechanical stability to allow single-crystal to single-crystal transformations to occur, with full structural characterisation of the halogenated products possible (Figure #ejic201600394-fig-0010#10e,f), while their chemical stability towards both halogens and water means that bromohydrination – the installation of one bromine and one hydroxyl unit across an unsaturated bond – can also be carried out. Halogenation was extended to chemisorption of iodine from the vapour phase by addition across multiple bonds, which could have great significance in the sequestration of radioactive iodine from the nuclear industry. The interpenetrated Zr MOF [Zr6O4(OH)4(peb)6]n, where peb is the bis-alkyne ligand 4,4′-[1,4-phenylenebis(ethyne-2,1-diyl)]-dibenzoate, can adsorb 279 % w/w of iodine by a combination of chemisorption across the alkyne units and physisorption in the large pores (Figure #ejic201600394-fig-0010#10g). The change in hybridisation of the carbon atoms in the linkers on postsynthetic halogenation also results in a change in mechanical compliance, which can be measured by single crystal nanoindentation. For example, the MOF [Zr6O4(OH)4(edb)6]n, a UiO-66 topology material where edb = 4,4′-ethynylenedibenzoate, has a Young's modulus of 15.1 (±0.8) GPa, which decreases on bromination to 9.3 (±0.6) GPa as a consequence of the increased flexibility of the linker (Figure #ejic201600394-fig-0010#10h).[]

30491_ejic201600394-fig-0011.jpg

Figure 11Open in figure viewerPowerPointa) Schematic illustrating temperature control of pore openings in PNIPAM-modified UiO-66-NH2. b) The postsynthetic surface modification procedure to install PNIPAM on the surface of UiO-66-NH2. Reproduced with permission from ref.51 Copyright (2015) Royal Society of Chemistry.

Limiting postsynthetic modification to the outer surfaces of MOF particles has been shown to induce a number of attractive properties related to both stability and application. A small group of surface modified Zr MOFs have been prepared largely from amino-tagged UiO-66 derivatives, using modifying agents that are too large to penetrate the pores of the MOF and so limit PSM to the particle surfaces. UiO-66-NH2 can be surface coated with a porous organic polymer by carrying out a Sonogashira coupling-based polymerisation in the presence of the MOF. The resultant materials exhibited considerable hydrophobicity and highly efficient removal of organic compounds from water. The mode of attachment of the polymer to the MOF is presumably noncovalent – attempts to functionalise UiO-66-I in the same way were unsuccessful – so this may be a case of the MOF acting simply as a template for polymer formation. Sada and Kokado attached poly(N-isopropylacrylamide) (PNIPAM) chains terminated with N-hydroxysuccinimide (NHS) activated esters to UiO-66-NH2 by amide coupling (Figure #ejic201600394-fig-0011#11). The temperature responsive mechanical behaviour of PNIPAM – it collapses into a globule formation at higher temperatures – allowed the researchers to control access to the pores, resulting in stimuli-responsive release of drug molecules from UiO-66-PNIPAM when temperatures were lowered.

30491_ejic201600394-fig-0012.jpg

Figure 12Open in figure viewerPowerPointa) Schematic showing the surface modification of UiO-66-N3 nanoparticles with DNA by b) postsynthetic strain-promoted alkyne-azide cycloaddition with dibenzocyclooctyne (DBCO) functionalised oligonucleotide strands. Reproduced (modified) with permission from ref.55 Copyright (2015) American Chemical Society.

Mirkin also showed that oligonucleotides could be covalently attached to the surface of UiO-66-N3, utilising strain-promoted alkyne-azide cycloaddition (SPAAC) to “click” dibenzylcyclooctyne-substituted DNA strands of 20 nucleobases in length to the MOF (Figure #ejic201600394-fig-0012#12). Radiolabelling confirmed modification occurred at the surface only, and the DNA modified MOF particles showed enhanced stability and cellular uptake.

30491_ejic201600394-fig-0013.jpg

Figure 13Open in figure viewerPowerPointa) Representations of the crystal structures of UiO-67(bipy) which can be metalated to UiO-67(bipy)CuCl2 in a single-crystal to single-crystal manner. Redrawn from CCDC depositions 968930 and 1042965, respectively. b) Catalytic arene borylation by UiO-67(bipy)Ir. c) Catalytic dehydrogenation of cyclohexanones by UiO-67(bipy)Pd.

Zr MOFs possessing metal chelation sites have received appreciable amounts of interest for a number of applications, with one of the most widely studied materials being UiO-67(bipy), also known as MOF-867, composed of 2,2′-bipyridine-5,5′-dicarboxylate bridging ligands. The structure of UiO-67(bipy) is very similar to that of the well-known UiO-67, with the addition of the accessible bipy sites lining the MOF pores, as the hard ZrIV cations are not coordinated during synthesis. The high chemical stability of Zr MOFs enables UiO-67(bipy) to undergo metalation in a single-crystal to single-crystal manner (Figure #ejic201600394-fig-0013#13a). Postsynthetic metalation can be achieved with a range of substrates, allowing the insertion of CuCl, CuCl2, CoCl2, FeBr2 and Cr(CO)6 within the framework by exposing the MOF to either solutions or vapour (in a closed system) containing the transition metal source. Postsynthetic metalation of the framework occurs almost quantitatively and interestingly the symmetry of UiO-67(bipy) is altered during the process, with the space group changing from Fm3m to Pa3, as a direct result of the ordering of the metalated linkers. Metalation with [Ir(COD)2]BF4 (COD = 1,5-cyclooctadiene) results in less than 10 % occupancy of Ir at the bipyridine sites, however the resulting material was found to be an efficient and recyclable catalyst for borylation of a number of arene precursors (Figure #ejic201600394-fig-0013#13b). Several other reports have focused on the postsynthetic metalation of UiO-67(bipy) for catalytic applications. In 2014, Lin et al. postsynthetically metalated UiO-67(bipy) by soaking the MOF in either a tetrahydrofuran or dimethyl sulfoxide solution of [Ir(COD)(OMe)]2 or [Pd(CH3CN)4][BF4] respectively at room temperature, resulting in an Ir loading of 30 % and a Pd loading of 24 %. The metalated frameworks were again proven to be highly efficient single site heterogeneous catalysts; C–H borylation of arenes and intramolecular ortho-silylation of benzylic silyl ethers to benzoxasiloles using UiO-67(bipy)Ir, and dehydrogenation of substituted cyclohexanones to phenols using UiO-67(bipy)Pd (Figure #ejic201600394-fig-0013#13c) were demonstrated. The catalytic activities of the metalated MOFs were compared to the homogeneous catalysts, and in the silylation reaction, UiO-67(bipy)Ir was found to be at least 1250 times more active. These high activities coupled with the recyclability of the MOF catalysts, due to their high stabilities, makes them prime candidates for use in catalytic technologies.

30491_ejic201600394-fig-0014.jpg

Figure 14Open in figure viewerPowerPointPostsynthetically metalated UiO-67 type MOFs containing either photocatalytically active a) ruthenium or b) manganese centres that were studied for the oxidative hydroxylation of arylboronic acids or for the reduction of CO2 to formate respectively.

Cohen et al. investigated the catalytic activity of a mixed ligand UiO-67 type MOF containing both biphenyl-4,4′-dicarboxylate (bpdc) and 2,2′-bipyridine-5,5′-dicarboxylate (bpydc) ligands in a 3:1 ratio. Postsynthetic metalation of the mixed-ligand MOF with Ru(bipy)2Cl2 resulted in a Ru loading of ca. 10 %, although this could be controlled by alteration of the reaction time. The resulting MOF was found to be an efficient and recyclable photocatalyst for the aerobic oxidative hydroxylation of arylboronic acids (Figure #ejic201600394-fig-0014#14a). Another mixed ligand UiO-67 type MOF, this time containing a 1:1 ratio of the bpdc and bpydc ligands, was metalated with bromopentacarbonylmanganese(I) [Mn(CO)5Br]. Inductively coupled plasma-optical emission spectroscopy (ICP-OES) confirmed that 76 % of the bpydc ligands were successfully metalated with the MnI photocatalyst, while direct synthetic attempts under comparatively harsh conditions were unsuccessful due to decomposition of the MnI complex. The photocatalytic reduction of CO2 to formate was investigated using the MnI incorporated MOF, in the presence of [Ru(dmb)3]2+ (dmb = 4,4′-dimethyl-2,2′-bipyridine) as a redox photosensitiser and 1-benzyl-1,4-dihydronicotinamide as a sacrificial reducing agent, and it was found that the overall turnover number and selectivity was higher than the homogeneous catalytic systems (Figure #ejic201600394-fig-0014#14b).

30491_ejic201600394-fig-0015.jpg

Figure 15Open in figure viewerPowerPointSchematic illustration highlighting metalations of UiO-66-CAT with Cr and UiO-66-TCAT with Pd, yielding excellent catalysts for oxidation and C–H activation, respectively.

Zr MOFs containing alternative metal chelation sites have been studied, with a catechol functionalised UiO-66 MOF recently synthesised and investigated for postsynthetic metalation and subsequently heterogeneous catalytic activity. UiO-66-CAT could not be prepared directly, rather the catechol bearing ligand, 2,3-dihydroxy-1,4-benzenedicarboxylate, was introduced postsynthetically either via postsynthetic deprotection (PSD) or postsynthetic exchange (PSE). Briefly, PSE is a process where the solid phase MOF undergoes ligand exchange, resulting in incorporation of secondary ligands by substitution of the parent ligand (described in detail in Section 6) while PSD describes the cleavage of protecting groups using external stimuli to reveal the desired functionality. Subsequently, postsynthetic metalation with either iron, in the form of Fe(ClO4)3 or Fe(CF3SO3)3, or chromium (K2CrO4) yielded unprecedented metal-monochelato species containing coordinatively unsaturated metal centres. UiO-66-CAT-Cr was found to be a highly efficient catalyst for the oxidation of alcohols to ketones even with very low chromium loadings (0.5–1.0 mol-%, Figure #ejic201600394-fig-0015#15). The versatility of the catalyst was proven during the oxidation of eight different substrates, which was observed to occur almost quantitatively in many cases. The high chemical stability of UiO-66-CAT-Cr allows the MOF to be recycled for a number of catalytic cycles, further increasing its potential for use in catalytic applications. In a further study, UiO-66-CAT was postsynthetically metalated with gallium [aqueous Ga(NO3)3(H2O)x solution] to afford UiO-66-CAT-Ga. The three metalated monochelato MOFs (UiO-66-CAT-M, M = CrIII, FeIII or GaIII) were examined for their catalytic activities during the photocatalytic reduction of CO2 to formate, however the Fe material did not demonstrate any catalytic activity as the redox potential for FeIII (0.77 V vs. standard hydrogen electrode) is not suitable for CO2 reduction. Of the remaining two metalated materials, it was found that UiO-66-CAT-Cr demonstrated greater catalytic activity, with turnover numbers calculated as 11.22 and 6.14 for UiO-66-CAT-Cr and UiO-66-CAT-Ga, respectively. A very similar analogue, UiO-66-TCAT, was synthesised by PSE of UiO-66 with 1,4-dicarboxybenzene-2,3-dithiol to afford a mixed ligand MOF containing open sulfur metal chelating sites. UiO-66-TCAT was postsynthetically metalated with Pd(OAc)2 (CH2Cl2 solution) to afford a dark brown solid containing accessible and coordinatively unsaturated Pd centres. Strong Pd–S bonds enable this MOF to be an efficient and recyclable catalyst for regioselective sp2 C–H oxidation, with conversion to ether or aryl halide functionalities demonstrated (Figure #ejic201600394-fig-0015#15).

30491_ejic201600394-fig-0016.jpg

Figure 16Open in figure viewerPowerPointa) Representation of the salicylaldimine containing MOF, sal-M-MOF, with the metal binding site (both Co and Fe were investigated) highlighted, alongside b) summarised conditions for the catalytic hydrogenation of 1-octene using the sal-M-MOFs.

A salicylaldimine based Zr MOF (sal-MOF) containing terphenyl based bridging ligands with pendant salicylaldimine functionality similar to UiO-68, was synthesised and postsynthetically metalated with tetrahydrofuran solutions containing cobalt (CoCl2) or iron (FeCl2·4H2O) to afford active single site solid catalysts for olefin hydrogenation, christened sal-Co-MOF and sal-Fe-MOF, respectively (Figure #ejic201600394-fig-0016#16). The spatial arrangement of the metal-salicylaldimine units within the MOF prevents oligomerisation, in contrast to homogeneous analogues that oligomerise, causing them to be inactive olefin hydrogenation catalysts. Impressively, during the hydrogenation of 1-octene under optimised conditions, sal-Fe-MOF had a turnover number of 1.45 × 105. The high catalytic activity of the sal-Fe-MOF is also recognised through the successful hydrogenation of a range of chemically and structurally varied terminal alkenes, while functional groups such as aldehydes, ketones and esters can be tolerated.

30491_ejic201600394-fig-0017.jpg

Figure 17Open in figure viewerPowerPointa) Schematic illustration of mPT-MOF-Ir highlighting both ligands incorporated in the MOF, alongside catalytic conditions investigated during the preparation of b) benzoxasiloles and c) azasilolanes.

Lin and co-workers have utilised extended UiO-type MOFs containing longer ligands with potential metal docking sites, resulting in increased pore openings which offers opportunities for a wider range of substrates for catalytic investigations. An extended dicarboxylate ligand containing phenanthroline units was incorporated into a mixed ligand Zr MOF, alongside 4,4′-bis(carboxyphenyl)-2-nitro-1,1′-biphenyl (≈ 1:2 ratio) which offers reduced steric bulk, with the MOF named mPT-MOF (Figure #ejic201600394-fig-0017#17a). Postsynthetic metalation with [Ir(COD)(OMe)]2 successfully produced mPT-MOF-Ir, a highly active catalyst for three different C–H activation reactions, specifically tandem hydrosilylation/ortho-silylation of aryl ketones and aldehydes (Figure #ejic201600394-fig-0017#17b), tandem dehydrocoupling/ortho-silylation reactions of N-methylbenzylamines (Figure #ejic201600394-fig-0017#17c), and borylations of aromatic C–H bonds. It was found that the mixed ligand MOFs were the most efficient catalysts compared to those comprising only the chelating ligands, likely as a result of the large open channels present due to the incorporation of less bulky secondary ligands.

30491_ejic201600394-fig-0018.jpg

Figure 18Open in figure viewerPowerPointa) Schematic illustration of the chiral BINAP-MOF alongside the catalytic hydrogenation of β-keto esters that was studied for the Ru metalated species. b) Schematic illustration of the chiral MOF containing pendant chiral diene moieties (E2-MOF) with the catalytic activity of the MOF for 1,4-additions of arylboronic acids to α,β-unsaturated ketones exemplified.

Efficient Zr MOF based enantioselective heterogeneous catalysts have been synthesised by incorporating extended chiral BINAP [2,2′-bis(diphenylphosphanyl)-1,1′-binaphthyl] dicarboxylate ligands into UiO-type MOFs. This so-called BINAP-MOF with general formula [Zr6O4(OH)4(Ligand)6]n (see part a of Figure #ejic201600394-fig-0018#18 for the extended chiral BINAP based ligand) adopts the inherent chirality of the bridging ligands and crystallises in the F23 space group, with the extended ligands resulting in a highly porous structure with a solvent accessible void space of 76.3 %. Postsynthetic metalation was successfully demonstrated with either Rh [bis(norbornadiene)rhodium(I) tetrafluoroborate] or Ru [bis(2-methylallyl)(1,5-cyclooctadiene)ruthenium(II)], enabling a broad scope of catalytic activities to be achieved by judicious choice of the incorporated active metal. Rh-BINAP-MOF (that is the rhodium metalated BINAP-MOF) was observed to be an efficient catalyst for the conjugate addition of arylboronic acids to 2-cyclohexenone with enantiomeric excess values in excess of 99 %, whilst being 3 times more active than the homogeneous control. In contrast, Ru-BINAP-MOF (that is the ruthenium metalated BINAP-MOF) was found to be highly active for the hydrogenation of β-keto esters (Figure #ejic201600394-fig-0018#18a) and substituted alkenes. A later study on Rh-BINAP-MOF showed that by using a mixed ligand strategy, where the second ligand does not contain metal binding sites and contains considerably less steric bulk, a catalytic framework with more open and accessible channels results. The mixed ligand MOF was found to be an active catalyst for sterically demanding substrates when the conventional BINAP-MOF was no longer active. An alternative approach to obtain an enantioselective Zr MOF catalyst was to incorporate pendant chiral diene moieties. The chiral diene containing UiO-type MOF, with general formula [Zr6O4(OH)4(Ligand)6]n (see part b of Figure #ejic201600394-fig-0018#18 for the chiral diene based ligand), termed E2-MOF, was proposed to be more efficient than the BINAP-MOF due to the decrease in steric demand, which should result in improved access of the reagents and consequently transport of the product from the active site. E2-MOF was metalated with either [RhCl(C2H4)2]2 or Rh(acac)(C2H4)2 and the resultant materials were investigated as heterogeneous catalysts for the asymmetric 1,4-addition of arylboronic acids to α,β-unsaturated ketones (Figure #ejic201600394-fig-0018#18b) or asymmetric 1,2-addition of arylboronic acids to aldimines respectively. The potential of E2-MOF·Rh(acac) is realised as better yields and enantioselectivities were obtained when compared with the homogeneous control.

30491_ejic201600394-fig-0019.jpg

Figure 19Open in figure viewerPowerPointRepresentation of the crystal structure of FJI-H6(Cu), which is generated by postsynthetic metalation of the freebase porphyrin MOF (FJI-H6) and catalyses the cycloaddition of CO2 and chloropropylene oxide. Redrawn from CCDC deposition 1043281.

An extended MOF-525 analogue, termed FJI-H6 and containing H2TBPP {H2TBPP = 4′,4′′′,4′′′′′,4′′′′′′′-(porphyrin-5,10,15,20-tetra-yl)tetrakis[(1,1′-biphenyl)-4-carboxylate]} ligands was reported, resulting in a highly porous structure containing 2.5 nm pores and a BET surface area of 5033 m2 g–1. Access to single crystals of FJI-H6 unambiguously confirmed that postsynthetic metalation of the porphyrin units is possible, with a single-crystal to single-crystal transformation observed upon treatment with Cu(NO3)2 in DMF at 85 °C, resulting in FJI-H6(Cu) (Figure #ejic201600394-fig-0019#19). The CuII ions occupy the square planar N4 coordination site in the porphyrin units as expected, resulting in axial positions that are uncoordinated. Both FJI-H6 and FJI-H6(Cu) were investigated as catalysts for the cycloaddition of CO2 with chloropropylene oxide to form the corresponding cyclic carbonate. Under the conditions examined, FJI-H6(Cu) outperformed the parent MOF by ca. 9 %, and interestingly retained its crystallinity, which was not the case for the parent MOF.

30491_ejic201600394-fig-0020.jpg

Figure 20Open in figure viewerPowerPointPostsynthetic exchange in UiO-66 series MOFs. a) Postsynthetic linker exchange of Br-bdc for NH2-bdc in UiO-66-Br. b) Postsynthetic metal exchange of Zr for Ti in UiO-66.

Postsynthetic exchange of Zr MOFs was reported in 2012 by Cohen et al., using aerosol time of flight mass spectrometry (ATOFMS) to detect exchange on individual particles. Samples of UiO-66-Br and UiO-66-NH2 were suspended in water for 5 days, with ATOFMS revealing that greater than 50 % of the MOF particles underwent exchange. The exchange process was found to be dependent on the solvent, with highest levels of exchange observed in water, followed by DMF, methanol and chloroform. An alternative exchange procedure was investigated, wherein UiO-66-Br was submersed in an aqueous solution containing NH2-bdc, and it was found that 76 % of the Br-bdc ligands could be replaced by NH2-bdc (Figure #ejic201600394-fig-0020#20). Interestingly, the converse exchange of Br-bdc into UiO-66-NH2 under similar conditions was found to occur to a much lesser degree, likely due to a number of different factors, which could reasonably include donor abilities, solubility and steric effects. Incorporation of TiIV into UiO-66 was first reported in 2012 by Cohen et al., whereby as-synthesised UiO-66 was immersed in DMF solutions containing TiIV sources.[] Using TiCl4(THF)2 as the TiIV source, loadings as high as 37.9 mol-% Ti were obtained (93 % of particles showed exchange by ATOFMS) with crystallinity retained during the transformation (Figure #ejic201600394-fig-0020#20b). N2 adsorption isotherms revealed that porosity was also retained, suggesting that metal ion substitution had occurred and that TiIV ions were not located within the pores. Titanium-exchanged UiO-66 was later investigated as a catalyst for the oxidation of cyclohexene. Catalytic performance improved at 70 °C although this was still not as efficient as UiO-66 containing TiIV supported at the nodes (TiIV anchored onto the capping hydroxyl groups), due to the tetrahedral coordination environment of the TiIV supported material (in comparison to the 7- or eight-coordinate TiIV ions in the transmetalated species which is dependent upon the extent of framework hydroxylation), as revealed from XPS (X-ray photoelectron spectroscopy) analysis.

30491_ejic201600394-fig-0021.jpg

Figure 21Open in figure viewerPowerPointa) The [FeFe] hydrogenase mimic incorporated into UiO-66 by postsynthetic exchange. b) Comparison of photocatalytic H2 evolution showing that UiO-66-[FeFe](dcbdt)(CO)6 outperformed both the unfunctionalised MOF and the homogeneous catalytic system. Reproduced (modified) with permission from ref.78 Copyright (2013) American Chemical Society.

During a later study, UiO-66 was postsynthetically exchanged with [FeFe](dcbdt)(CO)6 (dcbdt = 1,4-dicarboxybenzene-2,3-dithiolate) (Figure #ejic201600394-fig-0021#21a) resulting in substitution of 14 % of the bdc ligands for the thermally unstable [FeFe] complex. The structural similarity of the [FeFe] subunit with [FeFe] hydrogenases prompted the examination of the material for photocatalytic hydrogen production, with the [FeFe] containing MOF found to be an efficient catalyst, outperforming the homogeneous reference system both in terms of reaction rate and hydrogen production (Figure #ejic201600394-fig-0021#21b). This is due to the improved stability of the [FeFe] complex when contained within the MOF, whilst site localization also hinders charge recombination.

30491_ejic201600394-fig-0022.jpg

Figure 22Open in figure viewerPowerPointa) A terphenyl-dicarboxylate N-heterocyclic carbene linker which can be postsynthetically exchanged into UiO-68-(CH3)2, resulting in a MOF that can b) catalyse the isomerisation of 1-octen-3-ol to 3-octanone.

Postsynthetic exchange has also been performed on extended Zr MOFs, such as UiO-68-(CH3)2, containing 2′,5′-dimethyl-p-terphenyl-4,4′′-dicarboxylate ligands, where an iridium N-heterocyclic carbene (NHC) metallolinker was incorporated into the framework (Figure #ejic201600394-fig-0022#22a). Direct synthesis of the mixed ligand MOF was also considered, however, fine-tuning of the PSE conditions (methanolic ligand solution, 60 °C, 24 h) resulted in higher loadings of the metalloligand within the framework (17 % vs. 29 % as per 1H NMR spectroscopic analysis). Isomerisation of 1-octen-3-ol to 3-octanone was investigated in the presence of the Ir-containing MOFs (obtained via direct synthesis and PSE), however leaching of Ir was detected from the directly synthesised material, although this was not significant for the exchanged MOF. The Ir-containing exchanged UiO-68 type MOF proved to be an efficient catalyst, resulting in 99 % conversion without a loss of activity over three catalytic cycles (Figure #ejic201600394-fig-0022#22b).

30491_ejic201600394-fig-0023.jpg

Figure 23Open in figure viewerPowerPointa) Increasing isosteric heats of adsorption for CO2 in UiO-66 as more Zr is exchanged for Ti. Reproduced (modified) with permission from ref.88 Copyright (2013) Royal Society of Chemistry. b) Photocatalytic reduction of CO2 to formate by UiO-66-NH2 is also enhanced by exchange of Zr for Ti. The samples NH2-UiO-66(Zr/Ti)-100-4 and NH2-UiO-66(Zr/Ti)-120-16 have 34 % and 57 % of Zr cations exchanged for Ti, respectively. Reproduced (modified) with permission from ref.90 Copyright (2015) Royal Society of Chemistry.

Transmetalated UiO-66 samples containing TiIV were subsequently investigated for their ability to improve CO2 uptake capacities. UiO-66 samples containing TiIV loadings as high as 56 mol-% were obtained [denoted UiO-66(Ti56)] by allowing the transmetalation process to occur over a longer period of time. The CO2 uptake capacity at 273 K was observed to increase from 2.2 mmol g–1 for UiO-66 to 4 mmol g–1 for UiO-66(Ti56), representing an enhancement of 81 %. The authors suggest that the decrease in the octahedral pore diameter upon substitution with TiIV results in an increased isosteric heat of adsorption (Qst) for CO2 (Figure #ejic201600394-fig-0023#23a) and this coupled with the decreased density of the framework is responsible for the increased uptake. This methodology was investigated for exchanging TiIV into UiO-66 mixed matrix membranes, with the resulting membranes demonstrating dramatically improved permeabilities for CO2 compared to parent UiO-66 membranes (as high as 153 % improvement) without affecting selectivity. Following the interest surrounding improving CO2 uptake capacities, Ti-doped UiO-type MOFs have been investigated for photocatalytic CO2 reduction. Transmetalation of UiO-66-NH2 in DMF solutions containing TiCl4(THF)2 at different incubation temperatures and durations revealed that up to 57 mol-% TiIV could be incorporated [120 °C, 16 d, denoted as NH2-UiO-66(Zr/Ti)-120-16 in Figure #ejic201600394-fig-0023#23b]. Photocatalytic reduction of CO2 to formate was investigated using both pristine UiO-66-NH2 and transmetalated materials. Under the conditions investigated, 57 % Ti-exchanged UiO-66-NH2 resulted in the production of 1.7 times more formate than parent UiO-66-NH2. The improved performance were rationalised by a combination of low temperature electron spin resonance experiments and DFT calculations, which suggest that incorporation of TiIV into the cluster facilitates the transfer of electrons from the NH2-bdc ligands upon photo-excitation to the metal ion cluster. The reduced TiIII is then able to act as an electron donor as the ZrIV is reduced to ZrIII, resulting in improved electron transfer and ultimately improved catalytic abilities. A similar study extended the range of transmetalated UiO-type MOFs containing TiIV, to a UiO-66 analogue containing mixed ligands. Mixed ligand UiO-66-NH2, containing NH2-bdc as well as 2,5-diamino-1,4-benzenedicarboxylate (14 %) was transmetalated, resulting in a material with hexametallic nodes of average formula ≈ Zr4.3Ti1.7. The mixed ligand MOF was also investigated for the photocatalytic reduction of CO2, alongside Ti substituted UiO-66-NH2, and it was realised that the mixed ligand MOF was a more superior catalyst due to the introduction of new energy bands which facilitate improved light absorption and charge transfer.

30491_ejic201600394-fig-0024.jpg

Figure 24Open in figure viewerPowerPointa) Schematic showing thermal removal of TFA from defects in UiO-66 to create a Lewis acid catalyst for citronellal cyclisation. b) Comparison of conversion vs. time for different samples, with the most defective MOF being the most catalytic. Reproduced (modified) with permission from ref.98 Copyright (2013) American Chemical Society.

An interesting study conducted by Vos et al. reported the synthesis of UiO-66 using a combination of hydrochloric (HCl) and trifluoroacetic acid (TFA), resulting in highly crystalline materials that were subsequently investigated as catalysts for the “ene”-type cyclisation of citronellal to isopulegol (Figure #ejic201600394-fig-0024#24a). UiO-66-10HCl, that is UiO-66 synthesised in the presence of 10 equiv. of TFA and 1 equiv. of HCl, was observed to contain both physisorbed and cluster bound TFA (trifluoroacetate had partially replaced bdc) from thermal analysis and 19F solid-state NMR spectroscopy. Thermal treatment of UiO-66-10HCl was followed by in-situ IR spectroscopy, revealing that dehydroxylation of the Zr cluster begins prior to removal of trifluoroacetate, although the two processes occur simultaneously at higher temperatures. Postsynthetic thermal treatment results in Zr6O8(bdc)4 – a highly defective material containing Zr6 clusters surrounded by 8 carboxylates rather than the usual 12 – which has a high number of Lewis acid (ZrIV) sites and increased pore dimensions, resulting in dramatically improved catalytic performances. UiO-66-10HCl is considerably more active in the cyclisation of citronellal to isopulegol (Figure #ejic201600394-fig-0024#24b), while in the Meerwein reduction of 4-tert-butylcyclohexanone with 2-propanol, conversions of 7 % vs. 93 % are achieved for UiO-66-NO2 and UiO-66-NO2-10HCl, respectively.

30491_ejic201600394-fig-0025.jpg

Figure 25Open in figure viewerPowerPointCrystal structures of a) PCN-700, which contains vacancies, or “pockets” between 8-connected nodes that can be filled by postsynthetic addition of 1,4-benzenedicarboxylate, to give b) PCN-701, and 2′,5′-dimethylterphenyl-4,4′′-dicarboxylate, to give c) PCN-702. Both pockets can be filled to give d) PCN-703, but only if the smaller pocket is occupied by 1,4-benzenedicarboxylate first. Reproduced with permission from ref.101 Copyright (2015) American Chemical Society.

An alternative “repair” strategy has been applied to single crystals of PCN-700, resulting in the first Zr MOF bearing ligands that not only present different chemical functionality but are also of different lengths, resulting in what are commonly known as multivariate MOFs. The staggered arrangement of the phenyl rings in the 2,2′-dimethylbiphenyl-4,4′-dicarboxylate ligands of PCN-700 results in a bcu topology containing 8 connected nodes (Figure #ejic201600394-fig-0025#25a), in contrast to the 12 connected nodes of the UiO-66 series with fcu topology. The structure contains two “pockets” of different lengths (7.0 and 16.4 Å) which can accommodate linear dicarboxylate ligands, and as such the authors examined the postsynthetic installation of bdc (Figure #ejic201600394-fig-0025#25b) and tpdc-Me2 (tpdc-Me2 = 2′,5′-dimethyl-p-terphenyl-4,4′′-dicarboxylate, Figure #ejic201600394-fig-0025#25c). The postsynthetic order of addition of the bridging ligands is important, as upon insertion of the first ligand the size of the second pocket is altered (Figure #ejic201600394-fig-0025#25d), although by careful considerations a multivariate MOF containing 3 ligands was isolated and unambiguously characterised using single-crystal X-ray diffraction, whilst functionalised ligands such as NH2-bdc can also be incorporated, highlighting the versatility of this technique.

30491_ejic201600394-fig-0026.jpg

Figure 26Open in figure viewerPowerPointa) Schematic illustration of the PSE of oxalic acid into UiO-66, highlighting covalent attachment to the Zr6 node at defect sites. Reproduced (modified) with permission from ref.103 Copyright (2015) Centre National de la Recherche Scientifique (CNRS) and The Royal Society of Chemistry. b) UiO-66-ea breakthrough curves for 10:90 CO2/N2 (v/v) mixtures at 313 K and 1 atm with capacities at 0 % and 82 % relative humidity shown. Co and Ci are the concentrations of each gas at the outlet and inlet respectively. Reproduced (modified) with permission from ref.104 Copyright (2015) Royal Society of Chemistry.

The presence of either µ3-OH or terminal OH or H2O moieties at Zr nodes, whose concentration depends on both the defect concentration and topology of the MOF, results in a range of possible sites for functionalisation of the Zr6 clusters, and the Brønsted acidity of these groups has been calculated for a range of Zr MOFs using potentiometric titrations. Alternatively, defect sites within Zr MOFs present attachment sites for secondary ligands that contain potential coordinating sites, such as carboxylates or phosphonates. For example, defect-free and intentionally defective UiO-66 were synthesised, and then exposed to DMF solutions containing oxalic acid at room temperature for 2 hours. Oxalic acid was successfully incorporated into defective UiO-66 (Figure #ejic201600394-fig-0026#26a) however, no incorporation was observed for the defect-free material, and hence postsynthetic linker exchange can be excluded as a method of incorporation. During breakthrough experiments with a number of toxic compounds, UiO-66-ox (UiO-66 containing cluster bound oxalic acid) greatly outperformed both the defective and defect-free UiO-66 samples in terms of storage capacities for SO2 and NO2, whilst only a modest improvement was observed for NH3. An alternative strategy was employed for the installation of ethanolamine onto the Zr6 clusters of UiO-66, by exposing dehydrated UiO-66 (UiO-66 heated at 300 °C for 1 hour under an N2 atmosphere) to an anhydrous toluene solution containing ethanolamine (ea). Similar to UiO-66-ox, UiO-66-ea [Zr6O4(OH)2(ea)2(bdc)6]n contains pendant amino moieties (ethanolamine is grafted onto Zr6 clusters at the triangular windows formed between three ZrIV ions, partially replacing bridging µ3-OH moieties) which were expected to improve the material's CO2 adsorption capacity. Breakthrough experiments under simulated flue gas conditions (10:90 CO2/N2, 313 K) revealed that the purification capacity of UiO-66-ea is ca. 18 times that of non-modified UiO-66, and interestingly this capacity is maintained even under 82 % relative humidity (Figure #ejic201600394-fig-0026#26b).

30491_ejic201600394-fig-0027.jpg

Figure 27Open in figure viewerPowerPointa) Schematic of Solvent Assisted Linker Incorporation (SALI) at the 8-connected node of NU-1000 to give fluorinated MOFs which b) offer improved isosteric heats of adsorption for CO2. Reproduced (modified) with permission from ref.105 Copyright (2013) American Chemical Society.

The availability of terminal OH and H2O ligands on the Zr6 clusters of NU-1000 {[Zr6(µ3-O)4(µ3-OH)4(OH)4(OH2)4(TBAPy)2]n, where TBAPy = 1,3,6,8-tetrakis(p-benzoate)pyrene} to undergo postsynthetic exchange has resulted in extensive research where this phenomenon has been exploited in a process the authors call SALI (Solvent Assisted Ligand Incorporation). In 2013, Farha et al. reported the incorporation of a series of perfluorinated alkanes of varying length onto the nodes of NU-1000. Microcrystalline samples of NU-1000 were placed in contact with DMF solutions containing the desired perfluorinated aliphatic carboxylic acid and left to react at 60 °C for 18–24 hours (Figure #ejic201600394-fig-0027#27a). The maximum uptake of the perfluorinated alkane ligands per Zr6 node is 4, which would result in coordinatively saturated Zr6(µ3-O)4(µ3-OH)4(RCO2)12 clusters, similar to those found in UiO-type MOFs. The resultant materials, SALI-n, where n corresponds to the length of the perfluorinated alkane chain, could be obtained in quantitative yields for the lower alkane species, which dropped slightly as the chain length increased presumably due to steric effects. CO2 adsorption experiments revealed that, despite the additional functionality occupying pore volume, CO2 uptake capacities increased across the SALI-n series, with the Qst0 value for SALI-9 being twice that of unmodified NU-1000 (Figure #ejic201600394-fig-0027#27b). It was later found that NU-1000 materials containing perfluorinated alkanes (obtained via SALI) demonstrate enhanced water stabilities. The node functionalisation removes the polar terminal OH and H2O ligands, which interact strongly with water, while the incorporated fluorinated ligands prevent water from accessing the Zr6 clusters. Perfluorinated NU-1000 materials were subsequently loaded with Pd nanoparticles (confined within the pores), and investigated as catalysts for the C–H arylation of indoles in H2O.

30491_ejic201600394-fig-0028.jpg

Figure 28Open in figure viewerPowerPointa) Schematic illustration for grafting of AuMe(PMe3) onto the nodes of UiO-66, with the associated release of methane. b) Dehydration of the [Zr6(µ3-O)4(µ3-OH)4] clusters of UiO-66, forming [Zr6O6] clusters which are found to c) contain bound tert-butoxide anions upon treatment with lithium tert-butoxide.

Metalation of isolated Zr6O4(OH)4(RCOO)12 clusters has been studied, hence it is no surprise that efforts have investigated this route for the incorporation of secondary metals into Zr MOFs. Reaction of UiO-67 powder with an ether solution containing [AuMe(PMe3)] releases methane when the gold complex is successfully grafted onto the Zr-cluster (Figure #ejic201600394-fig-0028#28a), allowing the extent of postsynthetic modification to be quantitatively measured. Spectroscopic techniques and elemental analysis were also used, providing evidence that [Zr6O4(OH)3(OAuPMe3)(bpdc)6] was obtained. Under different reaction conditions, Long and co-workers successfully grafted lithium tert-butoxide onto the clusters of UiO-66 by first dehydrating the framework to form Zr6O6 clusters (Figure #ejic201600394-fig-0028#28b). Upon interaction with lithium tert-butoxide, which is expected to replace the now vacant µ3-OH sites by a similar binding motif, 25 % of the available sites were found to be occupied by tert-butoxide anions (Figure #ejic201600394-fig-0028#28c). Ionic conductivity measurements revealed that the lithium tert-butoxide grafted material behaves as a solid Li+ electrolyte with a conductivity of 1.8 × 10–5 S cm–1 at 293 K.

30491_ejic201600394-fig-0029.jpg

Figure 29Open in figure viewerPowerPointa) The 8-connected Zr6O4(OH)8(H2O)4 cluster of PCN-700 and b) the bimetallic [Zr6Ni4O8(OH)8(H2O)8] cluster that is formed upon metalation (terminal water ligands on Ni cations not shown for clarity). c) Crystallographic snapshots of the incorporation and migration of Ni cations at the cluster of PCN-700. Reproduced with permission from ref.123 Copyright (2015) Wiley-VCH.

Definitive evidence of the metalation process has been revealed by single-crystal X-ray diffraction, which has been used to provide snapshots at set time intervals (2, 4, 6, 12, and 24 h) of the metalation of single crystals of an 8-connected Zr MOF, linked by 2,2′-dimethylbiphenyl-4,4′-dicarboxylate ligands (PCN-700), with either Ni or Co. The metal source (CoII or NiII) was introduced as a DMF solution to single crystals of PCN-700 at 85 °C for 48 h, and it was observed that cluster metalation is accompanied by ligand migration, whereby the ligand dissociates from the Zr ion and attaches to the postsynthetically introduced metal ion. Overall, during metalation the 8-connected [Zr6O4(OH)8(H2O)4] cluster (Figure #ejic201600394-fig-0029#29a) transforms to become a bimetallic [Zr6M4O8(OH)8(H2O)8] (M = Co or Ni, Figure #ejic201600394-fig-0029#29b) cluster, whilst maintaining crystallinity. The flexibility of the material is key, as the ligands are required to withstand torsional angle changes (13.2°), ultimately enabling the postsynthetic ligand migration and metalation to occur and crystallographic snapshots of the process to be collected (Figure #ejic201600394-fig-0029#29c).

30491_ejic201600394-fig-0030.jpg

Figure 30Open in figure viewerPowerPointa) The solid state packing structure of NU-1000 is shown (right) and broken down to show both the TBAPy ligands and the Zr6 clusters. Their connectivity is shown, highlighting the presence of pore directed terminal OH ligands. b) Schematic illustration of AIM of NU-1000 with AlMe3 through reaction with the terminal OH ligands. Reproduced (modified) with permission from ref.18 Copyright (2013) American Chemical Society.

Gas phase metalation has emerged as an alternative method for the functionalisation of NU-1000, again taking advantage of the terminal OH and H2O ligands at its 8-connected clusters (Figure #ejic201600394-fig-0030#30a), and has been termed Atomic Layer Deposition (ALD) in MOFs, shortened to AIM (Figure #ejic201600394-fig-0030#30b). Microcrystalline samples of NU-1000 were placed in an ALD reactor and exposed to Zn(Et)2 or Al(Me)3 at 140 °C or 120 °C respectively. ICP-OES revealed that metalation of NU-1000 was successful, suggesting that 0.5 Zn or 1.4 Al atoms were incorporated per Zr atom using very short experiment times (minutes), whilst only slightly better metalation yields were obtained for the corresponding liquid phase reactions performed over several hours.

11978_cssc201903236-fig-5001.jpg

Scheme 1Open in figure viewerPowerPointPhosphorylation of glycerol (1 a/b) catalyzed by non-specific acid phosphatases (NSAPs) or phytase using pyrophosphate (PPi) or monophosphate (Pi), respectively.

11520_chem202102082-fig-0001.jpg

Figure 1Open in figure viewerPowerPointType-II clathrate crystal structure of Na24-δGe136 (space group Fdm) with the four-bonded Ge framework atoms represented as blue and the Na guest atoms as larger grey spheres. The pentagon-dodecahedral Ge20 cages (512) centered by Na1 atoms (site 16c) appear orange and the hexakaidecahedral Ge28 cages (51262) centered by Na2 atoms (site 8b) appear yellowish. The bonds between the Ge framework atoms are drawn as black, unit-cell edges (origin choice 2) as white lines.

The symmetric signal a1 with an isotropic shift of about 1200 ppm showed strong quadrupole coupling with a coupling constant of 308 kHz, as extracted from a signal fit. In agreement with that, the respective signal obtained at MAS conditions featured rotational side bands, which were visible as broad humps particularly towards larger shift (Figure #chem202102082-fig-0007#7b). Signal a1 therefore is assigned to Na1 atoms at site 16c (Fdm) centering the dodecahedral cages (Figure #chem202102082-fig-0001#1) with .m trigonal point symmetry, for which quadrupole coupling may be expected. The broadened and more asymmetric signal a2 with an isotropic shift of about 1500 ppm thus should arise from Na2 atoms in the hexakaidecahedral cages. The lack of quadrupole coupling for that signal is in line with the 3m cubic point symmetry of site 8b in the cage center. However, as the signal is broad and asymmetric, contributions of slightly different local environments are probable. Both a reduced occupancy of the Na2 site, but also partial off-center positions of Na2 atoms in the cages may be responsible. Weak characteristics of quadrupole coupling resulting from a corresponding non-cubic local symmetry for a part of the Na2 atoms might be unresolved.

11520_chem202102082-fig-0002.jpg

Figure 2Open in figure viewerPowerPointPreparation of Na24-δGe136: XRPD patterns (Cu Kα1) of product I and of the samples after its heat treatment at different conditions (black dots). For product II, also the calculated pattern obtained by Rietveld refinement (red line), the intensity residuals (black curve below) and the calculated reflection positions for α-Ge (upper tick row) and Na22Ge136 (bottom tick row) are plotted.

Since the exothermic effect at 410 °C was associated to the formation of crystalline Na24-δGe136 and α-Ge on dynamic heating, another DTA-TG investigation to the lower Tmax=350 °C, with a subsequent annealing time of 4 h at that temperature, was performed (Figure S5). As expected, α-Ge and Na24-δGe136 were observed in the specimen by XRPD afterwards. Also a small contribution of Na2GeO3 was revealed (Figure #chem202102082-fig-0004#4). Annealing of product I at 350 °C should thus enable the preparation of Na24-δGe136. However, the preparation of bulk samples at 350 °C was found challenging. By principle, the temperature control on larger samples and by using a tube furnace is less precise than in a thermobalance. With only a short reaction time needed at 350 °C, the crystallinity and the composition of the product were, therefore, hardly reproducible. On the other hand, a too low temperature of T=320 °C did not lead to a crystalline product, even after prolonged annealing for 66 h. By XRPD, only a small amount of α-Ge, similar to the already present one in product I, was observed. Oxides or Na12Ge17 were not revealed in the specimen, however (Figure #chem202102082-fig-0002#2). The changes in the measured background already hinted on some structural reorganization on the nanometer scale. After heat treatment for 66 h, but at T=330 °C, the presence of crystalline Na24-δGe136 and α-Ge was confirmed (Figure #chem202102082-fig-0002#2). But, a large contribution of α-Ge and only a small fraction of crystalline Na24-δGe136 were found, while about half of the sample still was X-ray amorphous. A further increase of the annealing temperature to 340 °C, and annealing for likewise 66 h, resulted in a widely crystalline product. Nevertheless, only a similarly small fraction of Na24-δGe136 was found (Figure #chem202102082-fig-0002#2). The product thus consisted mainly of crystalline α-Ge. Comprehensibly, a too long annealing time may lead to a widely crystalline product, but to proceeding decomposition and thus to lower yield of the metastable Na24-δGe136 phase. When the annealing time was shortened to 18 h while keeping the annealing temperature at 340 °C, the content of crystalline Na24-δGe136 distinctly increased. At these conditions, the clathrate phase was reproducibly found in an about 1 : 1 mass-ratio with α-Ge, while only a small fraction of the sample (typically below 10 mass-%, see below) remained undetected by XRPD. For a selective preparation of the clathrate phase, an annealing temperature of 340 °C and an annealing time of about 18 h is thus considered most favorable. The XRPD pattern of a typical product obtained by this annealing procedure is shown in Figure #chem202102082-fig-0002#2. This sample was extensively characterized and will be referred to as product II in the following. Na24-δGe136 crystallizes in the space group Fdm (no. 227). The structure solution from the XRPD data (Figure #chem202102082-fig-0002#2) exactly matched the original structure model for a type-II clathrate (Table 2). This makes an appropriate description with the Na atoms localized in the cage centers, i. e. on Wyckoff sites 16c and, particularly, on 8b probable, the latter of which had been questioned for the related type-II clathrate Na24Si136. Initially, the structure refinement by the Rietveld method included the scale factor, the lattice parameter, the positional parameters of Ge2 and Ge3, the isotropic thermal displacement parameters (Uiso) for all sites, as well as the phase fraction, the lattice parameter and the displacement parameter of the Ge1 atom for the second phase α-Ge. The refined positional parameters yielded expectable interatomic distances for the Ge framework (Tables S1 and S2). The displacement parameters of the framework atoms had equal values within the estimated standard deviation, and approached the one refined for α-Ge. This result indicated a reasonable scale factor as well as meaningful phase fractions of 45 mass-% for the clathrate phase and 55 mass-% for α-Ge. On the other hand, the displacement parameter of Na2 was peculiarly large (Uiso(Na2)>0.2 Å2). As an off-center split atomic position for Na2 in the large Ge28 cages could not be revealed, the result suggested a reduced occupancy. Due to the low scattering power of Na as compared to Ge and the limitations by the XRPD data, a correlation of the occupancy factor for Na2 with the displacement parameters occurred, and a lowered, but inappropriately uncertain Uiso(Na2) was obtained (Table S1). Finally, two refinements were performed, for which either the displacement parameter or the occupancy factor of the Na2 site were fixed and incrementally altered, while the respective other one was refined together with all other parameters. The final result was each indicated by the same minimum in the Bragg residual value RB (Table S1). For both restrained refinements, the obtained occ(Na2) and Uiso(Na2) approached the same values within the estimated standard deviation (e.s.d). The finally revealed displacement parameter of Na2, furthermore, had a similarly large value as it had been observed for the related Na24Si136, if described with the same type-II clathrate structure model with non-split Na2 position. Concerning the Na1 site, the occupancy factor slightly decreased on unfixing, but remained 1 within three times the e.s.d. Moreover, the displacement parameter got negative on reduced occupancy of Na1, hinting on a mismatch of actually observed and assigned electron density due to correlating parameters. In conclusion, the Na1 site was considered fully occupied by Na, and its occupancy factor was fixed to 1 for the final refinement cycle (Tables 2, S1). The calculated or assigned phase compositions from the refinement results, Na21.4(7)Ge136 and Na22.14Ge136, were practically equal within one e.s.d., so that the clathrate composition in product II is considered to be about Na22Ge136. This composition actually is consistent with the observed lattice parameter of a=15.4355(4) Å, which is somewhat smaller than 15.4412(7) Å of the recently reported slightly sodium-richer composition Na23.0(5)Ge136. However, while the difference in the lattice parameter between both clathrate products is significant, the composition difference is actually close to the resolution limit on basis of the XRPD data used in both cases. By using LaB6 as an internal standard for reflection intensities in Rietveld refinement, Na22Ge136 and α-Ge were found to actually represent almost the entire mass of the sample (Table 2). Product II was thus almost completely crystalline, and only a small mass-fraction of the sample (5 to 10 mass-%) was not detected by XRPD. By investigating several specimens of product II with LaB6 standard in that way, the 45 : 55 relative mass ratio of Na22Ge136 and α-Ge as well as the concluded mass fraction of not detected phases remained unchanged. Only in one exceptional case, about 25 mass-% of the specimen was estimated not to contribute to Bragg diffraction, showing that product II may feature inhomogeneities with respect to the degree of crystallinity. Also, small reflections hardly rising above the background, were occasionally observed in the XRPD patterns of the product II specimens, hinting on traces of crystalline impurities. However, differently to the specimens investigated after DTA-TG measurements, an unambiguous assignment in product II was not possible due to the very low intensity of the reflections.

11520_chem202102082-fig-0003.jpg

Figure 3Open in figure viewerPowerPointDTA-TG investigation of product I on heating at 10 K/min.

To elucidate the thermal and the crystallization behavior of product I, DTA-TG experiments followed by XRPD analyses of the used specimens were performed. Up to 700 °C, only a small mass-loss of less than 1 % was observed, but several broad exothermic effects were detected (Figure #chem202102082-fig-0003#3). On measurements of several specimens at the same heating rate, but to different maximum temperature, these effects were qualitatively reproducible (Figure S5). While the two broad exothermic effects at 160 °C and 260 °C came along with discernable, likewise smeared, steps in the TG curve, the sharper effect at 410 °C and the two weaker effects at 460 °C and 510 °C did not show such a coincidence. Further distinct mass loss was observed only above 600 °C.

11520_chem202102082-fig-0004.jpg

Figure 4Open in figure viewerPowerPoint(top) Experimental XRPD patterns (Cu Kα1) of the specimens after the DTA-TG measurements of product I to the given maximum temperature Tmax and an optional annealing for a time t (the color code corresponds to the one in Figures 3 and S5); (bottom) Calculated XRPD patterns for α-Ge,39 the type-II clathrate phase Na24-δGe136,2 Na2GeO3,40 Na4GeO441 and Na12Ge17.42

The XRPD investigations after the DTA-TG measurements showed that the exothermal effects below 300 °C should be associated to the crystallization of small amounts of Na12Ge17 and Na4GeO4. Besides the small amount of α-Ge already present in product I, both phases were identified by their strongest characteristic reflections in the specimen after the measurement to Tmax=300 °C (Figure #chem202102082-fig-0004#4). The stronger exothermic effect at 410 °C was found to be associated to the crystallization of α-Ge and Na24-δGe136. After heating to Tmax=430 °C, both phases were detected in the respective specimen, besides a small contribution of Na4GeO4. Na12Ge17 was not detected, however. Finally, the exothermic signals at 460 °C and 510 °C were found to be associated to the decomposition of the clathrate phase at the applied heating rate, and also to a transformation of Na4GeO4 into Na2GeO3. After the measurement to Tmax=700 °C, the specimen contained mainly crystalline α-Ge, besides a small fraction of Na2GeO3 (Figure #chem202102082-fig-0004#4). Details of the mechanisms still have to be investigated, but evidently, an overall mass-loss of the specimen is neither involved with the formation nor with the decomposition of the clathrate phase and, as well, not with the transformation of the ortho-germanate. Moreover, Na24-δGe136 is likely to decompose exothermally, so that its metastability may be concluded at the investigated conditions. Since the exothermic effect at 410 °C was associated to the formation of crystalline Na24-δGe136 and α-Ge on dynamic heating, another DTA-TG investigation to the lower Tmax=350 °C, with a subsequent annealing time of 4 h at that temperature, was performed (Figure S5). As expected, α-Ge and Na24-δGe136 were observed in the specimen by XRPD afterwards. Also a small contribution of Na2GeO3 was revealed (Figure #chem202102082-fig-0004#4). Annealing of product I at 350 °C should thus enable the preparation of Na24-δGe136. However, the preparation of bulk samples at 350 °C was found challenging. By principle, the temperature control on larger samples and by using a tube furnace is less precise than in a thermobalance. With only a short reaction time needed at 350 °C, the crystallinity and the composition of the product were, therefore, hardly reproducible. On the other hand, a too low temperature of T=320 °C did not lead to a crystalline product, even after prolonged annealing for 66 h. By XRPD, only a small amount of α-Ge, similar to the already present one in product I, was observed. Oxides or Na12Ge17 were not revealed in the specimen, however (Figure #chem202102082-fig-0002#2). The changes in the measured background already hinted on some structural reorganization on the nanometer scale. After heat treatment for 66 h, but at T=330 °C, the presence of crystalline Na24-δGe136 and α-Ge was confirmed (Figure #chem202102082-fig-0002#2). But, a large contribution of α-Ge and only a small fraction of crystalline Na24-δGe136 were found, while about half of the sample still was X-ray amorphous. A further increase of the annealing temperature to 340 °C, and annealing for likewise 66 h, resulted in a widely crystalline product. Nevertheless, only a similarly small fraction of Na24-δGe136 was found (Figure #chem202102082-fig-0002#2). The product thus consisted mainly of crystalline α-Ge. Comprehensibly, a too long annealing time may lead to a widely crystalline product, but to proceeding decomposition and thus to lower yield of the metastable Na24-δGe136 phase. When the annealing time was shortened to 18 h while keeping the annealing temperature at 340 °C, the content of crystalline Na24-δGe136 distinctly increased. At these conditions, the clathrate phase was reproducibly found in an about 1 : 1 mass-ratio with α-Ge, while only a small fraction of the sample (typically below 10 mass-%, see below) remained undetected by XRPD. For a selective preparation of the clathrate phase, an annealing temperature of 340 °C and an annealing time of about 18 h is thus considered most favorable. The XRPD pattern of a typical product obtained by this annealing procedure is shown in Figure #chem202102082-fig-0002#2. This sample was extensively characterized and will be referred to as product II in the following.

11520_chem202102082-fig-0005.jpg

Figure 5Open in figure viewerPowerPoint(top) Selected region of X-ray diffraction patterns of product II measured after storage on air or in an argon glovebox for the given times. If not otherwise mentioned, the specimens were stored at room temperature for the whole time. (bottom) Calculated XRPD patterns for LaB6,49 α-Ge39 and a type-II clathrate phase2 at sodium-rich composition Na22Ge136 (a=15.4355 Å) and NaxGe136 (x→0, a=15.26 Å).

Na24-δGe136 in product II was found to be subject to distinct changes at room temperature on time scales of hours to months. The aging on air and in argon atmosphere was traced by XRPD (Figure #chem202102082-fig-0005#5). Although the changes were found slower on storage in argon atmosphere, the principle behavior remained the same as on storage in air (Figure #chem202102082-fig-0006#6). A degradation solely due to reaction with moisture or oxygen may thus be excluded. Moreover, also a temperature increase to 70 °C additionally investigated in argon atmosphere led to acceleration of the changes, which further evidences the solely kinetic stability of the phase, and the low activation barrier for the observed aging process. The eye-catching changes of the clathrate phase on aging were a decrease of the lattice parameter (Figure #chem202102082-fig-0006#6), coming along with distinct, angle-dependent broadening of the Bragg reflections (Figure #chem202102082-fig-0005#5, Figure S1). Moreover, after longer storage time, a second isotypic clathrate phase with distinctly smaller lattice parameter formed. In XRPD, α-Ge had sharp and symmetric reflections (Figure #chem202102082-fig-0005#5), and the lattice parameter of a=6.5551(2) Å (Table 2) determined against LaB6 internal standard was close to that reported for pure α-Ge (a=5.65748(4) Å). Na as a part of the crystal structure of α-Ge in relevant concentration may thus be excluded as an explanation for the large Na content of product II. This seems unlikely anyway, as a large solubility of Na in diamond-type α-Ge has never been reported. As well, from Rietveld refinement there was no indication for a Na occupancy in the cages of the type-II clathrate larger than the ideal Na24Ge136. So far, such a case has been reported only for the related Na24+δSi136 obtained at high-pressure conditions. Framework substitution of the type-II clathrate by Na atoms similar to other alkali-metal clathrates like the Li-containing Na16Cs8LixGe136-x or the framework-substituted type-III clathrate Cs30(Na, Sn)172 may be excluded as well, particularly, if considering the large Na content required in such a case to reasonably explain the result of chemical analysis. Therefore, it is evident that the small mass fraction of product II not detected by XRPD (Table 2) contains a large portion of the total Na content of the sample.

11520_chem202102082-fig-0006.jpg

Figure 6Open in figure viewerPowerPointDevelopment of the lattice parameter of Na24-δGe136 on aging in air or argon atmosphere. Open symbols represent the values for the sodium-deficient phase NaxGe136 forming after longer time periods. The dataset for 97 days was obtained from the sample stored in argon for 90 days and subsequently annealed at 70 °C for another 7 days.

Na24-δGe136 in product II was found to be subject to distinct changes at room temperature on time scales of hours to months. The aging on air and in argon atmosphere was traced by XRPD (Figure #chem202102082-fig-0005#5). Although the changes were found slower on storage in argon atmosphere, the principle behavior remained the same as on storage in air (Figure #chem202102082-fig-0006#6). A degradation solely due to reaction with moisture or oxygen may thus be excluded. Moreover, also a temperature increase to 70 °C additionally investigated in argon atmosphere led to acceleration of the changes, which further evidences the solely kinetic stability of the phase, and the low activation barrier for the observed aging process. The eye-catching changes of the clathrate phase on aging were a decrease of the lattice parameter (Figure #chem202102082-fig-0006#6), coming along with distinct, angle-dependent broadening of the Bragg reflections (Figure #chem202102082-fig-0005#5, Figure S1). Moreover, after longer storage time, a second isotypic clathrate phase with distinctly smaller lattice parameter formed. Besides the structure model for Na22Ge136, also the sample composition deduced from XRPD and chemical analysis is widely supported by the NMR spectrum of product II. Actually, the third signal a3 at close to 0 ppm indicates the presence of a considerable fraction of cationic Na+ species in the sample. Even more, the slightly asymmetric signal a3 at static conditions (Figure #chem202102082-fig-0007#7a) was revealed to actually comprise the contributions of, at least, two chemically different Na species by the spectrum obtained at MAS conditions (Figure #chem202102082-fig-0007#7b). The larger signal component at a shift of 0 ppm was assigned to salt-like components in product II. Among them might be compounds such as sodium germanates, the presence of which has been indicated by XRPD after DTA-TG investigation and chemical analysis of product II. The smaller contribution to signal a3, having an isotropic shift of 47 ppm as revealed from the respective signal at MAS conditions, might originate from Na12Ge17 or a related cluster compound, the presence of which seems plausible with the above investigations as well. The reference spectrum measured for Na12Ge17 showed a relatively sharp and symmetric signal at the identical isotropic shift (Figure #chem202102082-fig-0006#6c). The shift was found close to the one observed for sodium atoms in the chemically related compound Na4Si4. Although Na12Ge17 has a complex crystal structure, the pseudo-hexagonal arrangement of the cluster entities and sodium atoms seems to provide similar coordination environments for all crystallographically different sodium atoms, so that their 23Na NMR signals actually merge to the observed sharp total signal.

11520_chem202102082-fig-0007.jpg

Figure 7Open in figure viewerPowerPoint23Na NMR spectroscopy. (a) Spectrum of product II measured at static conditions (black curve) with a fit (red curve) considering the three individual signals a1, a2 and a3 and leaving the residuals (grey curve) shown below. (b) Spectrum of product II measured at MAS conditions with a rotation frequency of 20 kHz revealing two signals close to 0 ppm, for which the central transitions and the rotational sidebands are marked with larger and smaller tick lines or asterisks, respectively. The comparably broad central transition and rotational side bands of signal a1 are marked by red triangles. (c) Spectrum of Na12Ge17 measured at static conditions as a reference. Dashed horizontal lines represent zero intensity for each graph.

As expected for a sample containing a type-II clathrate phase with almost completely Na-filled cages, the 23Na spectra of product II revealed two distinctly downfield-shifted signals. Together with the short relaxation times, which were estimated to be in the range of only 100 ms, this hints on strong Knight shifts due to conduction electrons. The signals are denoted with a1 and a2 in the static spectrum (Figure #chem202102082-fig-0007#7). Additionally, a third, in comparison to a1 and a2, much sharper and slightly asymmetric signal a3 was observed in the static spectrum, which is assigned to non-metallic secondary phases and will be discussed later in more detail. The symmetric signal a1 with an isotropic shift of about 1200 ppm showed strong quadrupole coupling with a coupling constant of 308 kHz, as extracted from a signal fit. In agreement with that, the respective signal obtained at MAS conditions featured rotational side bands, which were visible as broad humps particularly towards larger shift (Figure #chem202102082-fig-0007#7b). Signal a1 therefore is assigned to Na1 atoms at site 16c (Fdm) centering the dodecahedral cages (Figure #chem202102082-fig-0001#1) with .m trigonal point symmetry, for which quadrupole coupling may be expected. The broadened and more asymmetric signal a2 with an isotropic shift of about 1500 ppm thus should arise from Na2 atoms in the hexakaidecahedral cages. The lack of quadrupole coupling for that signal is in line with the 3m cubic point symmetry of site 8b in the cage center. However, as the signal is broad and asymmetric, contributions of slightly different local environments are probable. Both a reduced occupancy of the Na2 site, but also partial off-center positions of Na2 atoms in the cages may be responsible. Weak characteristics of quadrupole coupling resulting from a corresponding non-cubic local symmetry for a part of the Na2 atoms might be unresolved. Besides the structure model for Na22Ge136, also the sample composition deduced from XRPD and chemical analysis is widely supported by the NMR spectrum of product II. Actually, the third signal a3 at close to 0 ppm indicates the presence of a considerable fraction of cationic Na+ species in the sample. Even more, the slightly asymmetric signal a3 at static conditions (Figure #chem202102082-fig-0007#7a) was revealed to actually comprise the contributions of, at least, two chemically different Na species by the spectrum obtained at MAS conditions (Figure #chem202102082-fig-0007#7b). The larger signal component at a shift of 0 ppm was assigned to salt-like components in product II. Among them might be compounds such as sodium germanates, the presence of which has been indicated by XRPD after DTA-TG investigation and chemical analysis of product II. The smaller contribution to signal a3, having an isotropic shift of 47 ppm as revealed from the respective signal at MAS conditions, might originate from Na12Ge17 or a related cluster compound, the presence of which seems plausible with the above investigations as well. The reference spectrum measured for Na12Ge17 showed a relatively sharp and symmetric signal at the identical isotropic shift (Figure #chem202102082-fig-0006#6c). The shift was found close to the one observed for sodium atoms in the chemically related compound Na4Si4. Although Na12Ge17 has a complex crystal structure, the pseudo-hexagonal arrangement of the cluster entities and sodium atoms seems to provide similar coordination environments for all crystallographically different sodium atoms, so that their 23Na NMR signals actually merge to the observed sharp total signal. The relative quantification of Na contained in the clathrate phase and that in the non-metallic by-products of product II based on 23Na NMR was complicated by the different relaxation times and the large differences in the resonance frequencies of the different species. As pulse frequency, pulse length, and the delay times were optimized for the clathrate signals a1 and a2, the a3 signals of the by-products in the depicted spectra (Figure #chem202102082-fig-0007#7) actually appeared with reduced intensity. Also, the visible phase mismatch for the signals close to 0 ppm in the MAS spectrum (Figure #chem202102082-fig-0007#7b) stems from the non-ideal recording conditions for these signals. A pulse optimization series revealed that the absolute intensity for signal a3 in the spectrum at static conditions (Figure #chem202102082-fig-0007#7a) actually represents less than 50 % of the absolute intensity at optimal conditions for this signal. Therefore, the integrated signal intensity ratio of Na atoms of the clathrate phase to those of the by-products I(a1+a2)/I(a3)=78/22 extracted from the signal fit (Figure #chem202102082-fig-0007#7a, Table 3) actually translates into a molar ratio of Na atoms in the clathrate phase to Na atoms in the by-products approaching 1 : 1. The results of 23Na NMR thus also support the conclusion, that about half of the Na content found for product II by chemical analysis actually was contained in by-products, which were not assigned as crystalline phases in XRPD. The overall composition of product II is thus also reasonably reflected in the 23Na solid state NMR spectra. Only the relative amount of Na in Na12Ge17 or other cluster compounds and the signal at 0 ppm seems to disagree. However, as Na12Ge17 is very sensitive to moisture and air and the NMR experiments could not be performed in a totally inert environment, a part of sensitive germanides like Na12Ge17 might have reacted already to salt-like products such as NaOH or sodium germanates during the initial recording of the static spectrum and, possibly at a higher rate, on MAS conditions, which may lead to a slight temperature increase of the specimen.

11520_chem202102082-fig-0008.jpg

Figure 8Open in figure viewerPowerPointSpecific magnetic susceptibility of product II extrapolated to infinite external field vs. absolute temperature (red dots). The hump at T≈40 K likely originates from residual oxygen in the measuring system. A fit according to Equation (4) for data T≥80 K (solid black line) yielded the temperature-independent term χs,0 (dashed black line); the sum of diamagnetic increments calculated for the sample composition as obtained by chemical analysis (blue dashed line, Table 1) or for an assumed phase mixture of 45 mass-% Na22Ge136 and 55 mass-% α-Ge (green dashed line) are plotted for comparison.

The observed value of χ0,s=−1.90×10−7 emu g−1 indicates a weaker diamagnetism than expected from the sum of the diamagnetic increments calculated for a sample composition of 45 mass-% Na22Ge136 and 55 mass-% α-Ge by using the molar increments for Na+[53] and elemental four-bonded Ge0 in α-Ge (Figure #chem202102082-fig-0008#8). If also the contribution of the impurities is accounted for by using the nominal sample composition from chemical analysis (Table 1) and, as an approximation, besides the above increments for Na+ and (4b)Ge0 thus also the ones for O2−, C4+, N5+, S4+[53] are considered, a somewhat larger deviation is estimated (Figure #chem202102082-fig-0008#8). Irrespective of such uncertainty, by using the diamagnetic increment value for α-Ge, the increment of (4b)Ge0 in the clathrate should, by principle, be underestimated in its absolute value due to the larger mean atomic volume of a germanium atom in the clathrate framework and other structural contributions coming along with a non-ideal tetrahedral environment for most of the germanium atoms in the clathrate as similarly discussed before and also revealed for amorphous Ge. Hence, the expected diamagnetism should lead to an even more negative temperature-independent susceptibility contribution than estimated from the above sums of increments, so that a distinct positive deviation of the actually observed diamagnetism of the sample from the expected value is evident, and a temperature-independent Pauli-paramagnetic contribution is conclusive. Na22Ge136 is thus revealed to be a metal-like conductor. This result is well in line with the Knight shifts observed for the clathrate signals in 23Na NMR and with the behavior expected for such a metal-rich composition of an intermetallic type-II clathrate.

25527_open202100137-fig-0001.jpg

Figure 1Open in figure viewerPowerPointMelting diagram of a binary DES.

Deep eutectic solvents are prepared with mixtures of solid substances that can be described as Lewis or Bronsted acids and bases. One component acts as a hydrogen bond acceptor and other as a hydrogen bond donor. A variety of anionic and/or cationic species may be used as starting material for preparing a DES. When mixed, those compounds have the peculiarity of having a lower melting point than the species that form them. Thus, through a thermal equilibrium process, two solid substances form a liquid phase at their eutectic temperature (Figure #open202100137-fig-0001#1).

25527_open202100137-fig-0002.jpg

Figure 2Open in figure viewerPowerPointSchematic Synthesis of DES.

Figure #open202100137-fig-0002#2 shows schematically how binary DES mixing an inorganic salt and organic hydrogen donor is synthetized, in this case mixing inorganic salt and organic hydrogen bond donors.

25527_open202100137-fig-5001.jpg

Scheme 1Open in figure viewerPowerPointPET degradation.

Because of DESs physicochemical properties, biodegradability, low toxicity and lower price, DESs may be used to extract specific bioactive compounds, such as flavonoids, phenolic acids and polyphenols from various types of natural sources, favoring their synthesis and production, for use as antioxidant, anti-inflammatory, antibacterial, etc. (Figure #open202100137-fig-0003#3).

25527_open202100137-fig-0003.jpg

Figure 3Open in figure viewerPowerPointSchematic representation of extraction of bioactive molecules with DES.

25527_open202100137-fig-5002.jpg

Scheme 2Open in figure viewerPowerPointOxidation of alcohols with NBS in choline chloride urea.

25527_open202100137-fig-5003.jpg

Scheme 3Open in figure viewerPowerPointOxidation of alcohols in DES-TEMPO system.

25527_open202100137-fig-5004.jpg

Scheme 4Open in figure viewerPowerPointOxidation of toluene with H2O2 in DES.

25527_open202100137-fig-5005.jpg

Scheme 5Open in figure viewerPowerPointFumaric and maleic acids from furfural.

25527_open202100137-fig-5006.jpg

Scheme 6Open in figure viewerPowerPointGluconic acid from cellulose.

25527_open202100137-fig-5007.jpg

Scheme 7Open in figure viewerPowerPointReduction of epoxides and carbonyl compounds.

25527_open202100137-fig-5008.jpg

Scheme 8Open in figure viewerPowerPointReductive amination of aromatic carbonyl compounds.

25527_open202100137-fig-5009.jpg

Scheme 9Open in figure viewerPowerPointHydrogenation of hydroxymethyl furfural.

25527_open202100137-fig-5010.jpg

Scheme 10Open in figure viewerPowerPointCarbamate synthesis in a DES.

25527_open202100137-fig-5011.jpg

Scheme 11Open in figure viewerPowerPointAlkylation on aromatic molecules.

25527_open202100137-fig-5012.jpg

Scheme 12Open in figure viewerPowerPointBromination of 1-aminoanthra-9,10-quinone.

25527_open202100137-fig-5013.jpg

Scheme 13Open in figure viewerPowerPointEsterification of acids in choline ChCl/Cr3Cl.6H2O.

25527_open202100137-fig-5014.jpg

Scheme 14Open in figure viewerPowerPointEsterification reaction with quaternary ammonium salt DES.

25527_open202100137-fig-5015.jpg

Scheme 15Open in figure viewerPowerPointEnzymatic esterification of racemic menthol.

25527_open202100137-fig-5016.jpg

Scheme 16Open in figure viewerPowerPointTransesterification reactions can be performed with DES.

25527_open202100137-fig-5017.jpg

Scheme 17Open in figure viewerPowerPointEnzymatic aldol condensation.

25527_open202100137-fig-5018.jpg

Scheme 18Open in figure viewerPowerPointKnoevenagel condensation between aromatic aldehydes and active methylene compounds.

25527_open202100137-fig-5019.jpg

Scheme 19Open in figure viewerPowerPointNucleophilic substitution in ChCl/ ZnCl2.